> Of course both are useful, but I’d rather have familiarity with the codebase than a deep general understanding of software systems.
In my experience, getting that familiarity with a particular codebase in a way that isn't surface-level has always been a hands-on process. E.g. just because I know many general things about software, I need to know the particulars of the current codebase I'm in to know what is reasonable to actually apply to it.
This is a chicken and egg problem I find hard to resolve with LLMs. If we're pushed to delegate most work to them, how do you build that expertise? Sure you can ask questions about the codebase, but IMHO that falls under surface-level information, and the devil is often in the deeper details. Hmm.
I think this is only a partial answer but I've been surprised by how familiar dev leads become with the app even if they are not in the code.
They tend to stick around and they engage in the problem solving on a higher level and develop a detailed picture of how the app does and should behave. So at least that part of the expertise may come from working with an LLM to solve problems.
the devil is always in the details. those details are on every level you look at: human minds, nature around us, space. so if your inputs are vague, you should only expect outputs that are vague and generalized
The amplifying mirror analogy works best here. LLMs are ultimately a reflection of your own interactions with its weights, the tone you use, the structure with which you construct your prompt, aspects of an issue you tend to focus on, your breadth of vocabulary and world knowledge and whatnot.
People who (carefully) use it as an extension of their own mind and senses will very likely thrive, and those who use it as a replacement for their minds and their senses will struggle.
One of the Claude skills I made Claude itself generate was the 'learning a concept across tiers' skill -- from ELI5 level to a PhD level, and it triggers whenever I ask it a very general question on a complex topic that isn't my bread-and-butter. The fact that I'm able to choose explanation level from a super smart LLM (that's available 24x7) that can explain any topic under the sun would've been mind-bogglingly sci-fi-ish just 4 years ago in 2022.
When I get out of my domain, I always ask it to describe things to me "like I'm a senior CS student who doesn't know any industry terms". I find it really easy to skim past the definitions I know and very useful to get the explicit clarification on unfamiliar terminology. Telling it to talk down to you a little also seems to calm down it's tendency to talk like it's trying really hard to convince you it's smart like a 8th grader trying to use every single vocab word they learned last week.
I've been using a prompt that comes from the opposite direction for non-programming stuff: "Assume the user is an expert in all fields; while this is clearly logically untrue on a literal level, the user prefers to see a field's technical terminology and then ask the AI about terms the user does not understand rather than get an inaccurate statement about some issue."
Whether you have to reassure the LLM that this is obviously untrue, I don't know, but they do have a knowledge baseline to know it's not true and I have a sneaking suspicion it would be less effective without that.
This has ended up in some of the most interesting incidental knowledge exploration I've ever done. A recent example is that I was asking about some stretches and it started talking about how useful they are for the sarcomeres, which I had not heard of. Now I have.
I'm not saying this is better, just that it is different. I think there's a time and a place for both approaches.
Not the OP but you can whack this into your prompt and get most of the way there: "no jargon goes unearned, nothing gets dumbed down, every abstraction touches ground"
This is something that really needs to be formally studied.
I'm inclined to say that this matches my own experience, but I can't rule out confirmation bias on my part.
As a meticulous person generally looking for a very specific code outcome, I prompt in a way intended to get exactly the thing I have in mind, and my results reflect that. But on the other hand, I have coworkers who type ten-word prompts with very limited specificity, and they seem to get results that way as well, and that makes me wonder.
It would certainly be beneficial for my career and financial well-being for the assertion to be true, because it means I don't have to worry about being pushed out of my job by an army of $15/hr vibe coders. But the convenience of that assumption is exactly why I think it's important to be skeptical.
I do find that "signalling expertise" is important. "I have a significant background in biblical scholarship. You can assume I've read the most important works in NT studies in particular. Do not translate Greek, Latin, Hebrew, or Syriac. Now, I would like to know . . ." That changes things significantly. So does telling it you have 20+ years of experience with C programming, that you have a robust understanding of machine organization, memory layouts, embedded systems, etc.
For sure. On a personal coding project I said "I'm a professional software engineer, and while this is a hobby project I'm not just vibe-coding and want to build reliable software" and the agent suddenly started suggesting all kinds of things to make its code more robust.
LLMs skew toward over-focusing on things that you mention.
The reason "the agent suddenly started suggesting all kinds of things to make its code more robust" is because you said you "want to build reliable software".
It's not a signal of good judgment or understanding. It's just how LLM attention works.
I thought exactly the same at first. But then i wondered if that still holds true with today's advanced thinking, RLHF involved, frontier models. I guess to a certain extend it did indeed behave better, as a reaction to his self description into account.
EDIT: I mean, those systems accumulated so much complexity around the attention based next token predictor.
this feels like "make no mistakes" level of prompting. reliable software isn't as simple as making it reliable, it's about choosing the trade-offs in the areas that don't matter as much as the areas that do. if you keep prompting the LLM to make your software more robust it will keep giving you things to do. they aren't all good things. eventually you'll end up needing kubernetes to run a calculator app.
> In the 2010s, if you had technical gaps (say, you couldn’t write CSS), you had to either rely on a skilled colleague or just hope that the answer to your exact problem was out there on the internet.
You could read some general reference/guide/tutorial documentation on CSS, and then probably solve your problem (without searching for "how to center a div", or whatever your exact problem was, and copy&pasting the answer and moving on), also becoming more knowledgeable in the process.
The rest of the short blog post has some good points, but the first sentence sounds like it's targeted at the percentage of developers who did StackOverflow copy&paste to close Jira tickets, never becoming experts.
Delegating to LLM-ish AI is just a natural evolution of that. The question is whether they can still add value if kept in the loop.
The article author suggests that the answer is to be expert, and is addressing people who... "either rely on a skilled colleague or just hope that the answer to your exact problem was out there on the internet."
I think you're talking about a different type of expertise from TFA. Consider this: What if I never enjoyed frontend programming and so I never wanted to be an expert on that?
In fact, I never enjoyed frontend programming because it was such a pain to deal with matters I considered trivial yet so frustratingly hard to do right... like centering a div. And yet the slightest misalignment is visually jarring and forces me to get a bit OCD about fixing it, which made it even more frustrating.
I questioned the whole premise of the situation: is working around a bad developer experience something worth spending my time on? Unless I actively wanted to get in there and fix the situation, not really. So yes, in those cases I would outsource my problem to a colleague or StackOverflow and move on. And as a career choice, I preferred to do more backend dev.
I would posit that that was the type of expertise that did not matter. The type of expertise that really matters here is good UI design. That is entirely orthogonal to the drudgery that is implementing and debugging webpage rendering, and I am eternally grateful to LLMs for freeing us from it.
You can extend that line of thought to the entire article. What really matters (and what LLMs reward) is domain expertise rather than technical expertise.
> You could read some general reference/guide/tutorial documentation on CSS, and then probably solve your problem
Hours + Hours of reading and a lot of trial-and-error. The loop was so long and sooo slow. Now it's instant. As if your very first Google search just solved the problem for you immediately.
The problem with CSS is it's a) all broad and no depth and b) constantly adding major new paradigms. If you don't get up to speed and stay there, you lose it.
Pre-AI I'd say I've gotten over the hump 3 times with CSS, only to lose it again by the next time I had to use it in volume.
There was some learning, but also a lot of waste. As a self-learner, I've never been able to learn so fast as I can now with an LLM to instantly answer my specific questions, and incrementally correct and grow my mental model. And while with some subjects you need to worry about accuracy, LLMs are generally very good with programming questions, and (for most types of questions) you can verify their claims yourself.
I lost my desire to learn anything about web development twenty years ago and specialized my career in a completely different direction. I had not touched HTML at all since then until earlier this year, when I discovered that AI robots could do it for me. Now I am happy to whip up HTML/CSS/JavaScript visualizers, explorers, and even one-off interactive report presentations whenever they might be useful, precisely because I don't have to learn any of that crap to do it. My time & attention are far more productively spent focused on the work I am actually good at and interested in doing.
Why do you think thats worth learning? Doing a few times manually sure… but to force yourself to solely rely on it is foolish. Just use the calculator.
The people who make the tools that generate your assembly instructions need to learn it. Just like the people who make the browser rendering engine and push CSS forward still need to learn it.
The people who don't need to learn asm never needed to learn it. If you wouldn't code in asm now, you wouldn't have ever.
I just want to point out the exception - the person who the AI learns from.
Someone has to know it. And to advance it to the next level requires domain level expertise. Or are we allowing LLM's to determine the next CSS snapshot/version/level/whatever?
I am a frontend engineer and I see so many bugs from vibe-coded CSS from people who think just like you. CSS is very powerful and easy to get wrong if you don't understand it. Since most CSS code on the internet is not good, LLMs do not generate good CSS code and create a lot of bugs ranging from bad performance to completely broken reflow.
Honestly, we don't really need to learn any more. These days, I mostly copy and paste from tickets to the LLM, and then rubber stamp the output. I have people in my team that try to care about what they're making, but they're generating less code, and filling tickets more slowly, than I am. Even when it comes to system design and architecture, the LLMs aren't great at it, but honestly, they're better than my coworkers (I'd say system architecture is probably at the same readiness level that coding was last year). Expertise wouldn't really help you get things done faster.
You could probably replace me with a minimum wage worker to do some manual testing and copy-paste errors from the console into the LLM, and still be fine.
Let's see how long it is before the next round of layoffs, I guess. For now, the money's fine and the work's boring but ok.
But, no, the LLMs rewarding expertise line is pure cope. Software is not really skilled labor any more.
(obviously, throwaway -- I'm not interested in potentially advertising this to my employer)
> Honestly, we don't really need to learn any more.
Perhaps for your task of closing Jira tickets that’s true - but the you have a decision to make: wait until you’re eliminated or try to anticipate where future competitiveness comes from.
Yeah, that's the thing; future competitiveness isn't going to come from learning, at least not from learning about software. The LLMs have that covered, and software is becoming low status work anyways.
It's largely going to come from interpersonal skills and being likeable enough to be the face of the projects and companies. Technical skill is rapidly dropping in value.
Yes, but sometimes I don't need and want to learn. One example from my recent experience in research -- building custom dashboard pages for results of scientific analyses. Each analysis is bespoke, and building interactive webpages is simply not the skill many researchers have (and it's boring IMO). But here with LLM you could easily explore the results visually/share them with collaborators etc. There are plenty examples like that.
But certainly there are cases where learning is required.
You seem to be hung up on the part about "learning". Do you know how x86 registers work? Or atomic instructions available on ARM, SPARC, or POWER architectures?
No, of course not. Because all of that got abstracted to higher-level instructions decades ago.
Well that's great for you. That's some knowledge that you will possess for the rest of your life.
I want a green lawn and big bushy shrubs in front of my house. Do I need to know the intricate biology of my soil and habitat? No, of course not. I just do the surface-level things that make the lawn and shrubs thrive.
>No, of course not. I just do the surface-level things that make the lawn and shrubs thrive.
And others want the green lawn and big bushy shrubs, but don't want to learn all of the surface level things to make their lawn thrive so they hire a service to do it for them. And there's nothing wrong with that - not everyone enjoys the yard work, but they still want the thriving lawn and shrubs... and they are happy to write a check to OpenAI... err...Lawn Doctor every month to get that result since the time they save by not dealing with their lawn, they can pursue things they do want to do.
Not everyone wants to (or needs to) learn every detail along the way of getting the results they want.
I understand your point, but also imagine applying that across the whole spectrum of computing.
It's very useful that there's an intermediary that knows how that stuff works so I can build things without thinking about it in excruciating detail the whole time, I can dip down lower and learn stuff when it's relevant (like cache access and nested arrays) but I can also not do that in many situations.
There's also the argument that you can do engineering without understanding the underlying science as seen in th pyramids, the beautiful old european churches, etc
But maybe there's a limit to the usefulness of this abstraction. Perhaps forgetting too many basic skills lower down isn't always as useful as going from Assembly to C. I'd argue that there's a law of diminishing returns there, and it's not an endless series of benefits.
So, since you care so much and caring requires knowing everything about everything that even remotely intersects with your project, as (presumably) a software developer presumably you understand in detail how every bit of the CPUs you're targeting, plus the lithography, chemistry, and how solid state physics, and the underlying quantum mechanics work?
You seem to be massively exaggerating here for effect.
I think the main point is, if you're building frontend web apps, you should probably know how HTML, CSS, and JavaScript interact to some decently high level. Can you look stuff up on the margins via AI, or StackOverflow, or whatever? Sure.
If you're building <something else>, you should probably know the core tools and concepts at the appropriate level to build them.
No we do not need to know every aspect of CPU branch prediction and whatnot to make a webpage.
I think it's pretty obvious from the context of "if you had technical gaps (say, you couldn’t write CSS)" that this isn't about someone that regularly works on frontend web apps, it's about someone that is having to do something that is not normal for them. E.g. I mainly work on HPC code, maybe once in a year or two I need to throw together an animated poster to accompany a printed one. It's obviously ridiculous to claim that this means I have to learn frontend web dev.
All the boring stuff you skipped to get right to the answer you want was just as important whether you want to admit it or not.
Having to read through a structured resource describing something to figure something out has intrinsic value that an LLM is not going to provide you with.
> All the boring stuff you skipped to get right to the answer you want was just as important
This is obviously not the case. There are mountains and mountains of boring, unimportant stuff that LLMs can do for us now.
For example, at work I can now make a nestJS dashboard without having any idea what nestJS is. I can just tell the LLM what I want, and it makes it so. This allows me to make my tool useful to people without having to become an expert on something unimportant.
I don't think AI use is supposed to replace foundational learning such as reading a C++ book or Python book or CSS tutorial when you're a beginner. You still have to do those things if you want to be a professional or a strong amateur. But many people just want to get the thing done. They don't want to become a mechanic, they just want to drive from A to B.
> They don't want to become a mechanic, they just want to drive from A to B.
I'm fairly certain the article is directed at professionals, or at least the AI companies are basing their valuations off of directly taking a slice of that professional "productivity".
I've seen junior engineers be productive on their first day in the industry because of AI, so I don't think the article is the whole truth.
The example math is boundary-pushing and definitely not a solved problem. But most of us work on CRUD backends with a React frontend. Those are more or less solved problems that have well-documented solutions. For those kinds of tasks, LLMs just reward usage.
I can count on one hand the number of times in my career I've needed to solve a problem that's not described on Stack Overflow.
Except LLMs will only tend to share the most common or average of what it knows as the standard and deviating from it (including new ways) it can be resistant to.
An expert can lay a different kind of frame to prevent the llm to fell out of its way of being generally too verbose, and that can transfer as well to code generation and complication.
Not sure I agree with this. The math guy at anthropic's prompts are essentially:
"suppose you’ve gotta resolve the $CONJECTURE, like absolutely have to, everything depends on it. think really hard, and try to come up with a bunch of ideas to try. but remember to trust yourself and not necessarily in conventional wisdom!!"
https://claude.ai/share/25740bd5-aa97-4bd7-bf58-c4df3793fda7
https://xcancel.com/__alpoge__/status/2083855298239078748
Tao's chat was for him to gain intuition, not to solve the problem from the outset.
What's funny is that every other person gets a different conclusion about who these models reward/empower. I've seen people say that the generalist stands to gain the most and others say that it's the experts. Like all of life, maybe the "winner" is the person who just does stuff.
It depends on the levels. People with differing fitness levels and ages run at very different paces. Now, do cars make them more equal or less? On the bottom end, the tide lifts all boats. Most healthy people can learn to drive and will drive "fine", they get from A to B. Out there in the city streets the car flattens the differences, everyone roughly takes the same time to get from A to B in a car.
But at the top of top, the gap probably widens. A professional F1 driver will drive laps around some random guy. It amplifies reflexes etc, because at that speed little differences in timing make a big difference.
Now, AI coding isn't exactly analogous, but I think it also has these two regimes. It flattens things for simple tasks. If your task is to shovel data, do some trivial compiler wrangling staring at badly designed error messages, looking through GitHub issues hunting for the comment with many tadaa emojis to fix an issue etc, those things can now be done by anyone. Just as grandpa can also drive to the grocery store. But if you're pushing at things on a higher level, now only your above-AI ability matters. If all the things that AI can do well are subtracted out, how much other expertise do you have left? This will be proportionally a bigger and bigger difference between different people.
So you're saying that it depends whether you're measuring "success at task X" (where in average everyone improves) vs comparative success (where people with knowledge can do far and away better at specific tasks).
This works better for math because math is self-verifiable. Once you have a proof it needs no outside evidence.
Expertise is needed to evaluate model outputs where it can't verify itself, or at the very least one's expertise can help steer the model in the right direction.
However this is irrelevant if models themselves are better at evaluating/leveraging expertise/information.
Corollary to this is an important part of LLM usage is what I call pinning it to reality. That is, designing verification steps that interact with the real world in some way not easy to hallucinate or work around. This means things like having code that interacts with the physical world, round trip tests, arriving at the same result using different paths, interoperability / replication with external libraries / competing products, performance improvement projects that start with robust performance test suites, and similar sorts of things that reduce to "how do I provide evidence that's difficult to fool myself about".
This includes things like "before you start fixing this bug, write two tests that fail proving it exists".
Expertise is good, but a wise expert will set up methods for the machine to prove to itself that a desired result is achieved removing the expert from the tight development loop.
But the guy who writes the “just do it” prompt can neither formulate the conjecture in the first place, nor come up with any follow-up questions to build on the result.
Yeah, the people who say no expertise is needed for these things confuse me somewhat. This is indeed the case if you want to be a meat wrapper around an LLM, understanding neither your inputs nor your outputs. But at that point, what is the point of you versus going to the LLM myself? Expertise is necessary because it adds understanding and structure to the blob of text produced by an LLM. Progress can only be built on such understanding.
I am tempted to say (uncharitably) that the 'No knowledge needed! Just add LLMs!' byline is wishful thinking by non-experts who do not want to confront the reality that they will ultimately need to learn things.
There was one math proof that was AI generated going around Twitter and the chat transcript to generate it was basically the human prompting "keep going" until it solved it.
Tao's chat was fascinating because the questions he was asking belied expert knowledge of the subject that only a handful of people could have asked.
Yup I linked that in my other comment but copy-pasted here for everyone's convenience:
The counterexample of the Dinitz-Garg-Goemans conjecture was basically just "keep going" and finally "enough of partial results. now finish with a complete unconditional counterexample"
Yes, we're in the low hanging fruit stage where folks can just say "here's the problem" and "keep going" until a result is found and it will work sometimes.
The low hanging fruit will run short. Ultimately mathematics is a field of subjective selections of problems and proofs as beautiful and interesting. Machines absolutely will struggle with what to study, what theorems are desirable, and when do be done with a proof.
> Machines absolutely will struggle with what to study, what theorems are desirable, and when do be done with a proof
And why do you think this would be the case? I'm not talking about today but in 1-2 years. For reference o1 was released less than 2 years ago, and we've had reasonable coding agents for 9 months or so.
Mathematics is ultimately an aesthetic pursuit. Outside of a well defined goal ML models don't have any sense of taste and regardless of the scaling that's been enabled in the last year or so of capability if they haven't memorized the process of doing something they have the same limitations of inability to make choices about unknowns not trained into them.
Real synthetic intelligence seems to me to be still very far away and not a matter of making models bigger or more efficient.
LLMs are a collection of biases. Humans are also a collection of biases. So we project our biases as input through the biases of an LLM and get an output. Hence why I think getting optimal output requires being an optimal person. And in that sentence there are many points of expression.
Finally, we train our LLMs on who we are. Another reinforcement of biases.
The short version I give to non-technical people who ask me about whether "AI will replace coding" is this: it accelerates you. You can get much further much more quickly.
If you don't know where you're going or how to get there, or even if you're just not paying enough attention, it will get you very far in the wrong direction before you've realised.
I totally see this. I just did 3 hours of bot sitting to put together some thrash loops that thrash our provisioning working flow for a BLE gadget we make. It was pretty straightforward and productive. But then, I have a lot of experience with BLE, and a quite a bit of experience with python and shell scripting. So I was able to guide the process through stages, do some intermediate testing, make some adjustments, and proceed. Domain experience made this really easy and straightforward. Me two junior engineers who have only superficial/high level knowledge of BLE and some of the other pieces, couldn't have done this as effectively.
Where my angst comes, is worrying that no one will ever get that experience anymore. They might have had some eventual success, who knows what monstrosity a much less guided LLM would have done, but experential learning may be mostly a thing of the past. And it creates a real tension between the person with experience and the person without.
A good moment to remind everyone that if we took the promise for granted, that AI will in fact prevail and prompting is the one skill that will rule them all... we'll lose all domain experts in one generation.
It's less of "signaling expertise" and more about actually having said "expertise".
In my experience with LLMs it's not uncommon to be having a deep conversation about making pasta, only to be told, after asking for a sample recipe, to get a bucket of paint and a bag of concrete. Of course these hallucinations are way more subtle and easy to miss for someone who doesn't have deep domain knowledge.
Agree so much with this! In domains I know well I get much better results then someone who doesn't know the domain because I know where to challenge the LLM. LLMs need to be pushed because otherwise their answers are typically average.
That's why when people like Pieter Levels tweet "I cancelled and then vibecoded 100% of my SaaS subscriptions", you need to take it with a huge grain of salt because you're not Pieter Levels, you cannot vibe code your SaaS subscriptions.
The LLM is like the death star. If you don't know exactly where to point it, you will likely miss your target and have no/negative effect. The further away the target, the more accurate your firing solution needs to be. If all you need to do is add something like a dark mode theme to an existing product, this is probably a point blank shot in this metaphor. Building an entire codebase from zero, or even refactoring a legacy codebase into a new codebase, are lightyears away by comparison. You can still land the shot, but you need to deeply understand the metrology and astrodynamics.
The information system required to encode the aesthetic preferences needed to make a technology experience not suck is likely in excess of what any near-term solution will offer. Knowing when to say "no" is perhaps the most important skill here. You can't just say it arbitrarily either. You really have to mean it and be willing to fight other humans for it.
I am feeling this a lot lately. Getting the most out of agents seems to require being able to ask the right question. And how can you ask the right questions without deep domain expertise?
Yes sometimes it’s a matter of just using the right word. You can talk to an agent about a general concept for hours and hours and it may never mention $Concept_X, but you mention $Keyword_Y and all of a sudden the AI is going on about how $Concept_X is foundational to understanding the whole thing.
This matches my experience. Just Talk To It is the best method for working with LLMs if you're an expert.
I've seen this at work (as eng manager/lead/principal/whoevenknowsanymore) – all the big APIs give you stats. We see how much people burn in tokens and we know how much output they produce. There is a pretty strong inverse correlation between token burn and output.
The more tokens people burn, the less likely they are to produce a good outcome.
There's a growing and fascinating divide between people who see LLMs as more of a "bicycle for the mind" in the vein of Jobs vs those who see them as whollly supplanting the role of human intelligence. I can't help but wonder if these aren't primarily two human archetypes more than anything - the LLMs can be both and they erect a mirror of the human using them. Some humans really don't want deep individual expertise and intelligence to be the deciding factor because they don't identify with that. Others are completely the opposite. We really can't tell which will be more effective yet, because LLMs are very good in both modes. But most of the predictions currently are people executing on wishful thinking about what they hope will be the outcome.
I've just localized and translated all our products to 30 odd new languages (We had 3). This is something both LLMs are great at and boy was there a lot of actually knowing what I was doing. The most important first step was to create scripts that could traverse the apps on simulators/emulators taking screenshots of each screen. Being able to dump UIs, add accessibility identifiers, OCR screenshots etc. Another big part was not translating strings in isolation but building massive contexts around each string so the quality of the translation would be high. Then verifying those translations by using scripts to navigate to them and LLMs "viewing" them in context. Then at the back end of a massive multi week constant LLM usage there is the uploading the 13 screenshots per app per language (ios phone x5 ipad x3 android x5). etc etc. Even localizing images in marketing and in-app to be culturally relevant. Having familiarity with fastlane etc was very important. A naive approach would have been done in a day and been terrible.
So we could run a lighter LLM in front of humans, which translates from 'no domain knowledge' to 'domain expert' and in turn prompts over to the larger LLM.
Then the larger LLM gets all the right lights on, yields better outputs and we translate back into user domain.
I kinda thought the chain-of-thought reasoning already did this, no?
The counterexample of the Dinitz-Garg-Goemans conjecture was basically just "keep going" and finally "enough of partial results. now finish with a complete unconditional counterexample" lol
Well, nice post. Actually, there may be some truth behind it, but basically, it captures what I—as a programmer—want to read: expertise will remain valuable.
But how I am observing is different, though. Since LLMs the gap between experts and non-experts has been shrinking. And yes, there is still a gap, but vanishing.
In my experience (scientific programming) AI is a giant multiplier for people with specialized knowledge.
But it’s also a giant devaluer for that same knowledge as people with no idea what they’re doing can clog the field with plausible bullshit.
It’s now the case that if someone tells me they’ve done something, and I look into it and find out it’s completely AI slop, then I will have spent more time on the project than the person who “made” it. The situation is completely untenable and only serves to drain time and resources from people with better things to do.
this will have educational consequences (that I'm trying to solve). I don't think that we can adjust without rapid education and making extreme specialists of us all.
This requires coordination, certification, licensing, and other tiers of authenticity. False experts can ruin sample gathering, can ruin training. False expertise is exemplified by the current American Administration. Look at Robert F. Kennedy Jr.; he's a false expert. He is responsible for the measles outbreak. He is responsible for ivermectin abuse by humans. False expertise is overtaking real expertise. And the results are continuously disastrous and large-scale.
i find this post re-assuring (as who doesn't like to feel like they are an expert at something and llm definitely strips that away)... but it still feels like you are rewarded just as much for being a 6/10 expert as you are for being a 9/10 expert. It definitely is an equalizer it is just a question of to what degree.
I agree. When I talk to LLMs about fields I am familiar with, I can push back on bad suggestions and ignore faulty/incorrect advice and assumptions, which is much harder for unfamiliar subjects.
Of course, simple common sense and extremely basic Googling on unfamiliar subjects can produce similar results, but it's much faster if you are truly understanding what the AI is suggesting.
Yes. I agree with most, if not all of this. For instance, I am seeing folks either relying in the LLM as an _assumed_ expert or, assuming someone - who knows the structure of skill definitions - also has some expertise (in the area of the skill).
It's a difficult situation; there is not much point in explaining _why_ the LLM output or skill in use (on a domain problem) isn't what the person actually _needs_ to address the domain problem, because the person isn't a domain expert or indeed, adjacent to domain expertise.
But, it is an interesting experiemnt to arm folk with little domain expertise with the _skill_ necessary to be able to extract the right solution from the model.
This is also why people's experience with LLMs/AI varies so much, because some people can see a use for AI for their needs and go about using the tool, while others do not as it relates to whatever they're working on and so they may say "LLMs/AI are useless" (it doesn't mean they're not experts though, although some people who have totally no expertise might also see no use for AI for themselves).
Might be a very noob question but in this era of LLMs, let me ask the reverse, how do you gain expertise? It seems this rewards people who had expertise pre LLMs, but what about people who don’t have that in a specific domain? What approaches are viable now in this current system?
this 100%. Skills are gained via effort. Not too much effort that it discourages you, but enough so it's a challenge and engages you. Ofc it helps to have wonderful teachers, coaches, mentors, books, even ai to help guide you, but YOU must put in the effort. You don't get something for nothing.
The same as it's been! Make things without using LLMs. Don't debug with them, don't use them to research things, just do it yourself. It'll be painful and that pain is learning.
But getting an answer is not the same thing as understanding why that is the correct answer, or going deeper and learning more about the subject.
IMHO, the people who genuinely desire to learn will trudge through whatever they need to in order to grow their understanding - be it through reading books, original research papers or what have you.
If, OTOH, all you seek is the answers and that alone is satisfying to you, then of course you simply will not be motivated to do it the old school way anyway. But that's hardly different now in the age of AI.
I often do my best to represent a genuine interest in the subject at hand and learning in general to models. Imagine the model's response prose and mannerisms being on the other polar end of answering questions simply to get the correct answers as they're often scoped for on quantitative benchmarks. Not sure I explained this well, sorry. An LLM could help
I have lengthy conversations with my LLM, almost like an interview. I agree on the expertise part, because I wouldn't be able to go in depth on a subject with it if I lacked the expertise.
Some work is a result of design and negotiations in those designs.
I don't think Tao's style works with everyone/thing, especially if we don't know what style he's tuned his LLM on.
Overall, I agree, when I ask things I'm an expert in and do professionally every day. I get very good useful answers. When, for example, our marketing people, ask about the science, they often get confusing and wrong answers.
I'm surprised nobody mentioned (including the author) the Gell-Mann Amnesia Effect [1]... Just substitute "LLM" for "journalist" and there you have it.
And to be honest, I have seen it, as I'm sure (almost) everyone has, who has demonstrated experience/expertise in their own fields, and correct the LLM's responses one time or another...
Many, including myself, report having a lot of success with braindumping and not structuring anything. Just talking into speech recognition for 2-10 minutes as a stream of consciousness about what my context is, what I want, what I know already, what I have a vague hunch about, how it fits into a bigger picture, what aspects are most important to me, any footguns I already know about, really like having a chat with a person on the phone, with someone you have to guide remotely because they have to implement the thing right now but you have to be out of office and so your only interface is speech. Except you can be more structureless because the AI won't be offended. Just keep on rambling, and press enter, don't even correct mistranscriptions. It will understand it anyway.
Now, the key is, that while rambling without structure, you do have to drop the key facts into your speech, and you have to know what you're talking about in at least a good portion of it.
I think people are afraid of doing it, because it seems "not the right way" or "not scientific" or whatnot. They want to believe there is some magic to writing the right prompt. So let me tell you, it works.
I don't completely disagree with the concept of giving a free association thought process ramble into context. But I also bet that when you start getting it to actually generate code and link modules of things together, subroutines, functions, code structure and filenames, you still pay attention to what it does and you guide it into the architecture that makes logical sense to you.
For real work yes. For personal projects, less and less since Fable came out (probably the same if true of the other frontier models). You can get a lot done if it's just some one off, or a personal tool, even without looking at the code, just trying the application. Frontier models now automatically test it before handing the thing to you, they take screenshots, they fix the superficial issues themselves. To get something up and running, it's enough to send chat messages.
Unfortunately the software industry is saying things like "don't look at the code", "LLMs have made developers 10-100x faster", etc. The only way they can make such claims is by saying what you said above: "Claude, make me Microsoft Flight Simulator, make no mistakes".
Additionally engineers are facing pressures via deadlines to work in the paradigm of "Claude, make me Microsoft Flight Simulator, make no mistakes"...
The question i wonder about is, when will an event come along that persuades everyone that human understanding is still required? Or will it never come?
I wonder the same thing. I think we've already seen some of this happening, however the consequences haven't been large enough to the organization, for example:
Such an event would have to be pretty catastrophic at this point to slow down the inertia. Perhaps the tech debt will just pile up until someone's product implodes, or there's a massive safety issue that causes loss of life, or some big hedge fund goes bust.
The event could be when fair pricing comes from the model providers. We're still at the cash burning stage. When the economy crashes a little and departments start monitoring their spending, and the prices for inference are 10x what they are, there will be less tolerance for employees to substitute constant AI usage for understanding.
> Of course both are useful, but I’d rather have familiarity with the codebase than a deep general understanding of software systems.
In my experience, getting that familiarity with a particular codebase in a way that isn't surface-level has always been a hands-on process. E.g. just because I know many general things about software, I need to know the particulars of the current codebase I'm in to know what is reasonable to actually apply to it.
This is a chicken and egg problem I find hard to resolve with LLMs. If we're pushed to delegate most work to them, how do you build that expertise? Sure you can ask questions about the codebase, but IMHO that falls under surface-level information, and the devil is often in the deeper details. Hmm.
I think this is only a partial answer but I've been surprised by how familiar dev leads become with the app even if they are not in the code.
They tend to stick around and they engage in the problem solving on a higher level and develop a detailed picture of how the app does and should behave. So at least that part of the expertise may come from working with an LLM to solve problems.
the devil is always in the details. those details are on every level you look at: human minds, nature around us, space. so if your inputs are vague, you should only expect outputs that are vague and generalized
Read the code.
Reading it is good, but working with it more directly seems to help retention.
What if the code sucks, because it was vibe coded by an LLM over a dozen disparate sessions?
Then that's the point. You know the code sucks, the guy who vibecoded it either didn't know or didn't care. That's the added value.
Then fix it using the llm
Claude, make this codebase less ass
That will actually work pretty well.
The amplifying mirror analogy works best here. LLMs are ultimately a reflection of your own interactions with its weights, the tone you use, the structure with which you construct your prompt, aspects of an issue you tend to focus on, your breadth of vocabulary and world knowledge and whatnot.
People who (carefully) use it as an extension of their own mind and senses will very likely thrive, and those who use it as a replacement for their minds and their senses will struggle.
One of the Claude skills I made Claude itself generate was the 'learning a concept across tiers' skill -- from ELI5 level to a PhD level, and it triggers whenever I ask it a very general question on a complex topic that isn't my bread-and-butter. The fact that I'm able to choose explanation level from a super smart LLM (that's available 24x7) that can explain any topic under the sun would've been mind-bogglingly sci-fi-ish just 4 years ago in 2022.
When I get out of my domain, I always ask it to describe things to me "like I'm a senior CS student who doesn't know any industry terms". I find it really easy to skim past the definitions I know and very useful to get the explicit clarification on unfamiliar terminology. Telling it to talk down to you a little also seems to calm down it's tendency to talk like it's trying really hard to convince you it's smart like a 8th grader trying to use every single vocab word they learned last week.
I've been using a prompt that comes from the opposite direction for non-programming stuff: "Assume the user is an expert in all fields; while this is clearly logically untrue on a literal level, the user prefers to see a field's technical terminology and then ask the AI about terms the user does not understand rather than get an inaccurate statement about some issue."
Whether you have to reassure the LLM that this is obviously untrue, I don't know, but they do have a knowledge baseline to know it's not true and I have a sneaking suspicion it would be less effective without that.
This has ended up in some of the most interesting incidental knowledge exploration I've ever done. A recent example is that I was asking about some stretches and it started talking about how useful they are for the sarcomeres, which I had not heard of. Now I have.
I'm not saying this is better, just that it is different. I think there's a time and a place for both approaches.
That sounds useful. Can you share that skill?
Not the OP but you can whack this into your prompt and get most of the way there: "no jargon goes unearned, nothing gets dumbed down, every abstraction touches ground"
This is something that really needs to be formally studied.
I'm inclined to say that this matches my own experience, but I can't rule out confirmation bias on my part.
As a meticulous person generally looking for a very specific code outcome, I prompt in a way intended to get exactly the thing I have in mind, and my results reflect that. But on the other hand, I have coworkers who type ten-word prompts with very limited specificity, and they seem to get results that way as well, and that makes me wonder.
It would certainly be beneficial for my career and financial well-being for the assertion to be true, because it means I don't have to worry about being pushed out of my job by an army of $15/hr vibe coders. But the convenience of that assumption is exactly why I think it's important to be skeptical.
I do find that "signalling expertise" is important. "I have a significant background in biblical scholarship. You can assume I've read the most important works in NT studies in particular. Do not translate Greek, Latin, Hebrew, or Syriac. Now, I would like to know . . ." That changes things significantly. So does telling it you have 20+ years of experience with C programming, that you have a robust understanding of machine organization, memory layouts, embedded systems, etc.
For sure. On a personal coding project I said "I'm a professional software engineer, and while this is a hobby project I'm not just vibe-coding and want to build reliable software" and the agent suddenly started suggesting all kinds of things to make its code more robust.
LLMs skew toward over-focusing on things that you mention.
The reason "the agent suddenly started suggesting all kinds of things to make its code more robust" is because you said you "want to build reliable software".
It's not a signal of good judgment or understanding. It's just how LLM attention works.
I thought exactly the same at first. But then i wondered if that still holds true with today's advanced thinking, RLHF involved, frontier models. I guess to a certain extend it did indeed behave better, as a reaction to his self description into account.
EDIT: I mean, those systems accumulated so much complexity around the attention based next token predictor.
>build reliable software
this feels like "make no mistakes" level of prompting. reliable software isn't as simple as making it reliable, it's about choosing the trade-offs in the areas that don't matter as much as the areas that do. if you keep prompting the LLM to make your software more robust it will keep giving you things to do. they aren't all good things. eventually you'll end up needing kubernetes to run a calculator app.
> In the 2010s, if you had technical gaps (say, you couldn’t write CSS), you had to either rely on a skilled colleague or just hope that the answer to your exact problem was out there on the internet.
You could read some general reference/guide/tutorial documentation on CSS, and then probably solve your problem (without searching for "how to center a div", or whatever your exact problem was, and copy&pasting the answer and moving on), also becoming more knowledgeable in the process.
The rest of the short blog post has some good points, but the first sentence sounds like it's targeted at the percentage of developers who did StackOverflow copy&paste to close Jira tickets, never becoming experts.
Delegating to LLM-ish AI is just a natural evolution of that. The question is whether they can still add value if kept in the loop.
The article author suggests that the answer is to be expert, and is addressing people who... "either rely on a skilled colleague or just hope that the answer to your exact problem was out there on the internet."
I think you're talking about a different type of expertise from TFA. Consider this: What if I never enjoyed frontend programming and so I never wanted to be an expert on that?
In fact, I never enjoyed frontend programming because it was such a pain to deal with matters I considered trivial yet so frustratingly hard to do right... like centering a div. And yet the slightest misalignment is visually jarring and forces me to get a bit OCD about fixing it, which made it even more frustrating.
I questioned the whole premise of the situation: is working around a bad developer experience something worth spending my time on? Unless I actively wanted to get in there and fix the situation, not really. So yes, in those cases I would outsource my problem to a colleague or StackOverflow and move on. And as a career choice, I preferred to do more backend dev.
I would posit that that was the type of expertise that did not matter. The type of expertise that really matters here is good UI design. That is entirely orthogonal to the drudgery that is implementing and debugging webpage rendering, and I am eternally grateful to LLMs for freeing us from it.
You can extend that line of thought to the entire article. What really matters (and what LLMs reward) is domain expertise rather than technical expertise.
> like centering a div
We'll of course you're gonna be frustrated if you start with the hardest problem in HTML :/
> You could read some general reference/guide/tutorial documentation on CSS, and then probably solve your problem
Hours + Hours of reading and a lot of trial-and-error. The loop was so long and sooo slow. Now it's instant. As if your very first Google search just solved the problem for you immediately.
>Hours + Hours of reading and a lot of trial-and-error. The loop was so long and sooo slow.
But that's how you learn...
The problem with CSS is it's a) all broad and no depth and b) constantly adding major new paradigms. If you don't get up to speed and stay there, you lose it.
Pre-AI I'd say I've gotten over the hump 3 times with CSS, only to lose it again by the next time I had to use it in volume.
There was some learning, but also a lot of waste. As a self-learner, I've never been able to learn so fast as I can now with an LLM to instantly answer my specific questions, and incrementally correct and grow my mental model. And while with some subjects you need to worry about accuracy, LLMs are generally very good with programming questions, and (for most types of questions) you can verify their claims yourself.
I lost my desire to learn anything about web development twenty years ago and specialized my career in a completely different direction. I had not touched HTML at all since then until earlier this year, when I discovered that AI robots could do it for me. Now I am happy to whip up HTML/CSS/JavaScript visualizers, explorers, and even one-off interactive report presentations whenever they might be useful, precisely because I don't have to learn any of that crap to do it. My time & attention are far more productively spent focused on the work I am actually good at and interested in doing.
It is not the only way to learn.
Why do you think thats worth learning? Doing a few times manually sure… but to force yourself to solely rely on it is foolish. Just use the calculator.
We don't need to learn CSS anymore. Just like we don't need to learn ASM since GCC does a great job generating it from higher-level code.
Learning ASM will actually teach you a lot about how the computer works, and I doubt learning CSS deeply teaches you much about design fundamentals.
Define "we".
The people who make the tools that generate your assembly instructions need to learn it. Just like the people who make the browser rendering engine and push CSS forward still need to learn it.
The people who don't need to learn asm never needed to learn it. If you wouldn't code in asm now, you wouldn't have ever.
This " logic" is so irrational.
I just want to point out the exception - the person who the AI learns from.
Someone has to know it. And to advance it to the next level requires domain level expertise. Or are we allowing LLM's to determine the next CSS snapshot/version/level/whatever?
Experts work on gcc so everyone else doesn’t have to.
Yes
I am a frontend engineer and I see so many bugs from vibe-coded CSS from people who think just like you. CSS is very powerful and easy to get wrong if you don't understand it. Since most CSS code on the internet is not good, LLMs do not generate good CSS code and create a lot of bugs ranging from bad performance to completely broken reflow.
Funnily enough, I seem to remember some people saying that compiler generated assembly wasn’t as good as handmade assembly.
For early compilers in certain contexts, this was once true.
the question is, what's the business cost of some amount of css bugs vs paying specialists or people to learn these details.
the generalists win overall, except of course for specific cases where specialists are great
I want to believe this is sarcasm, but it's so hard to tell these days.
For many cases like the one described, its true.
It's not sarcasm.
We've all accepted that code-generation has been required and accepted for decades.
Code generation (by compiler) has one major difference: it's deterministic.
Here are the reported miscompilation bugs in GCC so far in 2026. The ones labeled "wrong-code".
https://gcc.gnu.org/bugzilla/buglist.cgi?chfield=%5BBug%20cr
I count 500+ of them.
Honestly, we don't really need to learn any more. These days, I mostly copy and paste from tickets to the LLM, and then rubber stamp the output. I have people in my team that try to care about what they're making, but they're generating less code, and filling tickets more slowly, than I am. Even when it comes to system design and architecture, the LLMs aren't great at it, but honestly, they're better than my coworkers (I'd say system architecture is probably at the same readiness level that coding was last year). Expertise wouldn't really help you get things done faster.
You could probably replace me with a minimum wage worker to do some manual testing and copy-paste errors from the console into the LLM, and still be fine.
Let's see how long it is before the next round of layoffs, I guess. For now, the money's fine and the work's boring but ok.
But, no, the LLMs rewarding expertise line is pure cope. Software is not really skilled labor any more.
(obviously, throwaway -- I'm not interested in potentially advertising this to my employer)
> Honestly, we don't really need to learn any more.
Perhaps for your task of closing Jira tickets that’s true - but the you have a decision to make: wait until you’re eliminated or try to anticipate where future competitiveness comes from.
Yeah, that's the thing; future competitiveness isn't going to come from learning, at least not from learning about software. The LLMs have that covered, and software is becoming low status work anyways.
It's largely going to come from interpersonal skills and being likeable enough to be the face of the projects and companies. Technical skill is rapidly dropping in value.
This claim can only be true if LLMs have completely mastered all intellectual aspects of creating and shipping software.
That's so far from being true it's like saying that humans are currently an interplanetary, post-scarcity species.
Yes, but sometimes I don't need and want to learn. One example from my recent experience in research -- building custom dashboard pages for results of scientific analyses. Each analysis is bespoke, and building interactive webpages is simply not the skill many researchers have (and it's boring IMO). But here with LLM you could easily explore the results visually/share them with collaborators etc. There are plenty examples like that. But certainly there are cases where learning is required.
This "instant" loop is so fast because it doesn't involve the part where you learn
You seem to be hung up on the part about "learning". Do you know how x86 registers work? Or atomic instructions available on ARM, SPARC, or POWER architectures?
No, of course not. Because all of that got abstracted to higher-level instructions decades ago.
Yes, I do. And it wouldn't take you that long to learn, either.
Well that's great for you. That's some knowledge that you will possess for the rest of your life.
I want a green lawn and big bushy shrubs in front of my house. Do I need to know the intricate biology of my soil and habitat? No, of course not. I just do the surface-level things that make the lawn and shrubs thrive.
>No, of course not. I just do the surface-level things that make the lawn and shrubs thrive.
And others want the green lawn and big bushy shrubs, but don't want to learn all of the surface level things to make their lawn thrive so they hire a service to do it for them. And there's nothing wrong with that - not everyone enjoys the yard work, but they still want the thriving lawn and shrubs... and they are happy to write a check to OpenAI... err...Lawn Doctor every month to get that result since the time they save by not dealing with their lawn, they can pursue things they do want to do.
Not everyone wants to (or needs to) learn every detail along the way of getting the results they want.
I understand your point, but also imagine applying that across the whole spectrum of computing.
It's very useful that there's an intermediary that knows how that stuff works so I can build things without thinking about it in excruciating detail the whole time, I can dip down lower and learn stuff when it's relevant (like cache access and nested arrays) but I can also not do that in many situations.
There's also the argument that you can do engineering without understanding the underlying science as seen in th pyramids, the beautiful old european churches, etc
Who is more productive? A python dev who learns about x86 registers in his free time or one that uses AI at work?
The one who uses AI at work will eventually be replaced by someone cheaper who uses AI at work.
The one who learns about x86 registers is much more likely to continue to evolve
But maybe there's a limit to the usefulness of this abstraction. Perhaps forgetting too many basic skills lower down isn't always as useful as going from Assembly to C. I'd argue that there's a law of diminishing returns there, and it's not an endless series of benefits.
Learn what? The thing the tool can do instantly? Take the win and spend your energy on bigger problems.
In a world where there are endless things to do and only limited time to get them done, not every issue needs to be a learning matter.
If you are making something that you care about you should probably understand how it works. I didn't know this was such a hot take.
So, since you care so much and caring requires knowing everything about everything that even remotely intersects with your project, as (presumably) a software developer presumably you understand in detail how every bit of the CPUs you're targeting, plus the lithography, chemistry, and how solid state physics, and the underlying quantum mechanics work?
You seem to be massively exaggerating here for effect.
I think the main point is, if you're building frontend web apps, you should probably know how HTML, CSS, and JavaScript interact to some decently high level. Can you look stuff up on the margins via AI, or StackOverflow, or whatever? Sure.
If you're building <something else>, you should probably know the core tools and concepts at the appropriate level to build them.
No we do not need to know every aspect of CPU branch prediction and whatnot to make a webpage.
I think it's pretty obvious from the context of "if you had technical gaps (say, you couldn’t write CSS)" that this isn't about someone that regularly works on frontend web apps, it's about someone that is having to do something that is not normal for them. E.g. I mainly work on HPC code, maybe once in a year or two I need to throw together an animated poster to accompany a printed one. It's obviously ridiculous to claim that this means I have to learn frontend web dev.
All the boring stuff you skipped to get right to the answer you want was just as important whether you want to admit it or not.
Having to read through a structured resource describing something to figure something out has intrinsic value that an LLM is not going to provide you with.
> All the boring stuff you skipped to get right to the answer you want was just as important
This is obviously not the case. There are mountains and mountains of boring, unimportant stuff that LLMs can do for us now.
For example, at work I can now make a nestJS dashboard without having any idea what nestJS is. I can just tell the LLM what I want, and it makes it so. This allows me to make my tool useful to people without having to become an expert on something unimportant.
You’re not the only one who doesn’t know what nestJS is
Well, no, it might not be important. Maybe you won't use that knowledge about CSS grids and flexbox for another year or two, or maybe ever.
I don't think AI use is supposed to replace foundational learning such as reading a C++ book or Python book or CSS tutorial when you're a beginner. You still have to do those things if you want to be a professional or a strong amateur. But many people just want to get the thing done. They don't want to become a mechanic, they just want to drive from A to B.
> They don't want to become a mechanic, they just want to drive from A to B.
I'm fairly certain the article is directed at professionals, or at least the AI companies are basing their valuations off of directly taking a slice of that professional "productivity".
I've seen junior engineers be productive on their first day in the industry because of AI, so I don't think the article is the whole truth.
The example math is boundary-pushing and definitely not a solved problem. But most of us work on CRUD backends with a React frontend. Those are more or less solved problems that have well-documented solutions. For those kinds of tasks, LLMs just reward usage.
I can count on one hand the number of times in my career I've needed to solve a problem that's not described on Stack Overflow.
As they said in the 80s or maybe earlier RTFM. I think if you got a good enough duster TFM was still readable in 2010.
the author forgot you could also do a secret third thing: learn!
dont really see the point when LLM compiles english
Except LLMs will only tend to share the most common or average of what it knows as the standard and deviating from it (including new ways) it can be resistant to.
An expert can lay a different kind of frame to prevent the llm to fell out of its way of being generally too verbose, and that can transfer as well to code generation and complication.
Not sure I agree with this. The math guy at anthropic's prompts are essentially:
Tao's chat was for him to gain intuition, not to solve the problem from the outset.What's funny is that every other person gets a different conclusion about who these models reward/empower. I've seen people say that the generalist stands to gain the most and others say that it's the experts. Like all of life, maybe the "winner" is the person who just does stuff.
It depends on the levels. People with differing fitness levels and ages run at very different paces. Now, do cars make them more equal or less? On the bottom end, the tide lifts all boats. Most healthy people can learn to drive and will drive "fine", they get from A to B. Out there in the city streets the car flattens the differences, everyone roughly takes the same time to get from A to B in a car.
But at the top of top, the gap probably widens. A professional F1 driver will drive laps around some random guy. It amplifies reflexes etc, because at that speed little differences in timing make a big difference.
Now, AI coding isn't exactly analogous, but I think it also has these two regimes. It flattens things for simple tasks. If your task is to shovel data, do some trivial compiler wrangling staring at badly designed error messages, looking through GitHub issues hunting for the comment with many tadaa emojis to fix an issue etc, those things can now be done by anyone. Just as grandpa can also drive to the grocery store. But if you're pushing at things on a higher level, now only your above-AI ability matters. If all the things that AI can do well are subtracted out, how much other expertise do you have left? This will be proportionally a bigger and bigger difference between different people.
So you're saying that it depends whether you're measuring "success at task X" (where in average everyone improves) vs comparative success (where people with knowledge can do far and away better at specific tasks).
This works better for math because math is self-verifiable. Once you have a proof it needs no outside evidence.
Expertise is needed to evaluate model outputs where it can't verify itself, or at the very least one's expertise can help steer the model in the right direction.
However this is irrelevant if models themselves are better at evaluating/leveraging expertise/information.
Corollary to this is an important part of LLM usage is what I call pinning it to reality. That is, designing verification steps that interact with the real world in some way not easy to hallucinate or work around. This means things like having code that interacts with the physical world, round trip tests, arriving at the same result using different paths, interoperability / replication with external libraries / competing products, performance improvement projects that start with robust performance test suites, and similar sorts of things that reduce to "how do I provide evidence that's difficult to fool myself about".
This includes things like "before you start fixing this bug, write two tests that fail proving it exists".
Expertise is good, but a wise expert will set up methods for the machine to prove to itself that a desired result is achieved removing the expert from the tight development loop.
But the guy who writes the “just do it” prompt can neither formulate the conjecture in the first place, nor come up with any follow-up questions to build on the result.
Yeah, the people who say no expertise is needed for these things confuse me somewhat. This is indeed the case if you want to be a meat wrapper around an LLM, understanding neither your inputs nor your outputs. But at that point, what is the point of you versus going to the LLM myself? Expertise is necessary because it adds understanding and structure to the blob of text produced by an LLM. Progress can only be built on such understanding.
I am tempted to say (uncharitably) that the 'No knowledge needed! Just add LLMs!' byline is wishful thinking by non-experts who do not want to confront the reality that they will ultimately need to learn things.
It's not contradictory to say that expertise is a multiplier, and that models are systematically underconfident in themselves.
Who's end state took / is going to take more tokens / money, however?
"LLMs reward expertise" is the title, not that "LLMs only make things possible for those with expertise"
And what about problems that cannot be one-shotted but helped along?
There was one math proof that was AI generated going around Twitter and the chat transcript to generate it was basically the human prompting "keep going" until it solved it.
Tao's chat was fascinating because the questions he was asking belied expert knowledge of the subject that only a handful of people could have asked.
Yup I linked that in my other comment but copy-pasted here for everyone's convenience:
The counterexample of the Dinitz-Garg-Goemans conjecture was basically just "keep going" and finally "enough of partial results. now finish with a complete unconditional counterexample"
https://x.com/DmitryRybin1/status/2079904005652893709
https://chatgpt.com/share/6a60b2eb-0b64-83ee-9c76-7931ca1de0...
Yes, we're in the low hanging fruit stage where folks can just say "here's the problem" and "keep going" until a result is found and it will work sometimes.
The low hanging fruit will run short. Ultimately mathematics is a field of subjective selections of problems and proofs as beautiful and interesting. Machines absolutely will struggle with what to study, what theorems are desirable, and when do be done with a proof.
> Machines absolutely will struggle with what to study, what theorems are desirable, and when do be done with a proof
And why do you think this would be the case? I'm not talking about today but in 1-2 years. For reference o1 was released less than 2 years ago, and we've had reasonable coding agents for 9 months or so.
> why do you think this would be the case?
Mathematics is ultimately an aesthetic pursuit. Outside of a well defined goal ML models don't have any sense of taste and regardless of the scaling that's been enabled in the last year or so of capability if they haven't memorized the process of doing something they have the same limitations of inability to make choices about unknowns not trained into them.
Real synthetic intelligence seems to me to be still very far away and not a matter of making models bigger or more efficient.
While they may struggle on what to study, we'll be able to spin up millions of them.
> who these models reward/empower
The easy, straightforward answer is "the people who own the models". Who else benefits feels like a more complex question and we'll have to see...
> Like all of life, maybe the "winner" is the person who just does stuff.
Someone who just does stuff still has to be able to deal with errors and failures. That’s where an expert or a generalist may have an advantage.
LLMs are a collection of biases. Humans are also a collection of biases. So we project our biases as input through the biases of an LLM and get an output. Hence why I think getting optimal output requires being an optimal person. And in that sentence there are many points of expression.
Finally, we train our LLMs on who we are. Another reinforcement of biases.
The short version I give to non-technical people who ask me about whether "AI will replace coding" is this: it accelerates you. You can get much further much more quickly.
If you don't know where you're going or how to get there, or even if you're just not paying enough attention, it will get you very far in the wrong direction before you've realised.
I totally see this. I just did 3 hours of bot sitting to put together some thrash loops that thrash our provisioning working flow for a BLE gadget we make. It was pretty straightforward and productive. But then, I have a lot of experience with BLE, and a quite a bit of experience with python and shell scripting. So I was able to guide the process through stages, do some intermediate testing, make some adjustments, and proceed. Domain experience made this really easy and straightforward. Me two junior engineers who have only superficial/high level knowledge of BLE and some of the other pieces, couldn't have done this as effectively.
Where my angst comes, is worrying that no one will ever get that experience anymore. They might have had some eventual success, who knows what monstrosity a much less guided LLM would have done, but experential learning may be mostly a thing of the past. And it creates a real tension between the person with experience and the person without.
A good moment to remind everyone that if we took the promise for granted, that AI will in fact prevail and prompting is the one skill that will rule them all... we'll lose all domain experts in one generation.
It's less of "signaling expertise" and more about actually having said "expertise".
In my experience with LLMs it's not uncommon to be having a deep conversation about making pasta, only to be told, after asking for a sample recipe, to get a bucket of paint and a bag of concrete. Of course these hallucinations are way more subtle and easy to miss for someone who doesn't have deep domain knowledge.
Agree so much with this! In domains I know well I get much better results then someone who doesn't know the domain because I know where to challenge the LLM. LLMs need to be pushed because otherwise their answers are typically average.
That's why when people like Pieter Levels tweet "I cancelled and then vibecoded 100% of my SaaS subscriptions", you need to take it with a huge grain of salt because you're not Pieter Levels, you cannot vibe code your SaaS subscriptions.
The LLM is like the death star. If you don't know exactly where to point it, you will likely miss your target and have no/negative effect. The further away the target, the more accurate your firing solution needs to be. If all you need to do is add something like a dark mode theme to an existing product, this is probably a point blank shot in this metaphor. Building an entire codebase from zero, or even refactoring a legacy codebase into a new codebase, are lightyears away by comparison. You can still land the shot, but you need to deeply understand the metrology and astrodynamics.
The information system required to encode the aesthetic preferences needed to make a technology experience not suck is likely in excess of what any near-term solution will offer. Knowing when to say "no" is perhaps the most important skill here. You can't just say it arbitrarily either. You really have to mean it and be willing to fight other humans for it.
I am feeling this a lot lately. Getting the most out of agents seems to require being able to ask the right question. And how can you ask the right questions without deep domain expertise?
Yes sometimes it’s a matter of just using the right word. You can talk to an agent about a general concept for hours and hours and it may never mention $Concept_X, but you mention $Keyword_Y and all of a sudden the AI is going on about how $Concept_X is foundational to understanding the whole thing.
This matches my experience. Just Talk To It is the best method for working with LLMs if you're an expert.
I've seen this at work (as eng manager/lead/principal/whoevenknowsanymore) – all the big APIs give you stats. We see how much people burn in tokens and we know how much output they produce. There is a pretty strong inverse correlation between token burn and output.
The more tokens people burn, the less likely they are to produce a good outcome.
There's a growing and fascinating divide between people who see LLMs as more of a "bicycle for the mind" in the vein of Jobs vs those who see them as whollly supplanting the role of human intelligence. I can't help but wonder if these aren't primarily two human archetypes more than anything - the LLMs can be both and they erect a mirror of the human using them. Some humans really don't want deep individual expertise and intelligence to be the deciding factor because they don't identify with that. Others are completely the opposite. We really can't tell which will be more effective yet, because LLMs are very good in both modes. But most of the predictions currently are people executing on wishful thinking about what they hope will be the outcome.
Agree with this. LLMs multiply the human user's ability. More ability, more impact!
And unfortunately, more ineptitude, more chaos.
I've just localized and translated all our products to 30 odd new languages (We had 3). This is something both LLMs are great at and boy was there a lot of actually knowing what I was doing. The most important first step was to create scripts that could traverse the apps on simulators/emulators taking screenshots of each screen. Being able to dump UIs, add accessibility identifiers, OCR screenshots etc. Another big part was not translating strings in isolation but building massive contexts around each string so the quality of the translation would be high. Then verifying those translations by using scripts to navigate to them and LLMs "viewing" them in context. Then at the back end of a massive multi week constant LLM usage there is the uploading the 13 screenshots per app per language (ios phone x5 ipad x3 android x5). etc etc. Even localizing images in marketing and in-app to be culturally relevant. Having familiarity with fastlane etc was very important. A naive approach would have been done in a day and been terrible.
So we could run a lighter LLM in front of humans, which translates from 'no domain knowledge' to 'domain expert' and in turn prompts over to the larger LLM.
Then the larger LLM gets all the right lights on, yields better outputs and we translate back into user domain.
I kinda thought the chain-of-thought reasoning already did this, no?
This has been my experience as well. I’ve also been thinking a lot about Terrence Tao and his chats and presentation.
The counterexample of the Dinitz-Garg-Goemans conjecture was basically just "keep going" and finally "enough of partial results. now finish with a complete unconditional counterexample" lol
https://x.com/DmitryRybin1/status/2079904005652893709
https://chatgpt.com/share/6a60b2eb-0b64-83ee-9c76-7931ca1de0...
Prompt an image or video generator without knowledge in photography or art skills and your results will look sloppy.
Well, nice post. Actually, there may be some truth behind it, but basically, it captures what I—as a programmer—want to read: expertise will remain valuable.
But how I am observing is different, though. Since LLMs the gap between experts and non-experts has been shrinking. And yes, there is still a gap, but vanishing.
You're basically playing the role of team lead to the LLM's junior dev.
I love larping as a vacant scrum master
This is true but also false.
In my experience (scientific programming) AI is a giant multiplier for people with specialized knowledge.
But it’s also a giant devaluer for that same knowledge as people with no idea what they’re doing can clog the field with plausible bullshit.
It’s now the case that if someone tells me they’ve done something, and I look into it and find out it’s completely AI slop, then I will have spent more time on the project than the person who “made” it. The situation is completely untenable and only serves to drain time and resources from people with better things to do.
we are slowly punishing reading comprehension
this will have educational consequences (that I'm trying to solve). I don't think that we can adjust without rapid education and making extreme specialists of us all.
This requires coordination, certification, licensing, and other tiers of authenticity. False experts can ruin sample gathering, can ruin training. False expertise is exemplified by the current American Administration. Look at Robert F. Kennedy Jr.; he's a false expert. He is responsible for the measles outbreak. He is responsible for ivermectin abuse by humans. False expertise is overtaking real expertise. And the results are continuously disastrous and large-scale.
i find this post re-assuring (as who doesn't like to feel like they are an expert at something and llm definitely strips that away)... but it still feels like you are rewarded just as much for being a 6/10 expert as you are for being a 9/10 expert. It definitely is an equalizer it is just a question of to what degree.
I agree. When I talk to LLMs about fields I am familiar with, I can push back on bad suggestions and ignore faulty/incorrect advice and assumptions, which is much harder for unfamiliar subjects.
Of course, simple common sense and extremely basic Googling on unfamiliar subjects can produce similar results, but it's much faster if you are truly understanding what the AI is suggesting.
Yes. I agree with most, if not all of this. For instance, I am seeing folks either relying in the LLM as an _assumed_ expert or, assuming someone - who knows the structure of skill definitions - also has some expertise (in the area of the skill). It's a difficult situation; there is not much point in explaining _why_ the LLM output or skill in use (on a domain problem) isn't what the person actually _needs_ to address the domain problem, because the person isn't a domain expert or indeed, adjacent to domain expertise. But, it is an interesting experiemnt to arm folk with little domain expertise with the _skill_ necessary to be able to extract the right solution from the model.
This is also why people's experience with LLMs/AI varies so much, because some people can see a use for AI for their needs and go about using the tool, while others do not as it relates to whatever they're working on and so they may say "LLMs/AI are useless" (it doesn't mean they're not experts though, although some people who have totally no expertise might also see no use for AI for themselves).
Might be a very noob question but in this era of LLMs, let me ask the reverse, how do you gain expertise? It seems this rewards people who had expertise pre LLMs, but what about people who don’t have that in a specific domain? What approaches are viable now in this current system?
Same way you get strong in an age of heavy machinery: Lift heavy weights yourself.
Skills will have to be built through artificial constraints. Pen & paper, reading books, not using AI, etc.
this 100%. Skills are gained via effort. Not too much effort that it discourages you, but enough so it's a challenge and engages you. Ofc it helps to have wonderful teachers, coaches, mentors, books, even ai to help guide you, but YOU must put in the effort. You don't get something for nothing.
The same as it's been! Make things without using LLMs. Don't debug with them, don't use them to research things, just do it yourself. It'll be painful and that pain is learning.
I'm not sure I understand the question. What would prevent you from doing what these people did now that LLMs are here?
Those people had no choice. In my opinion, it’s harder to grind through problems knowing very well an answer is a prompt away.
But getting an answer is not the same thing as understanding why that is the correct answer, or going deeper and learning more about the subject.
IMHO, the people who genuinely desire to learn will trudge through whatever they need to in order to grow their understanding - be it through reading books, original research papers or what have you.
If, OTOH, all you seek is the answers and that alone is satisfying to you, then of course you simply will not be motivated to do it the old school way anyway. But that's hardly different now in the age of AI.
This feels like a moment in time, not the end state of AI.
Like I read there was a time when teams of people + AI could beat pure AI at chess. But that these days, pure AI wins.
For all the things people say about "how AI works" you have to add the missing piece "how current AI works".
I often do my best to represent a genuine interest in the subject at hand and learning in general to models. Imagine the model's response prose and mannerisms being on the other polar end of answering questions simply to get the correct answers as they're often scoped for on quantitative benchmarks. Not sure I explained this well, sorry. An LLM could help
I have lengthy conversations with my LLM, almost like an interview. I agree on the expertise part, because I wouldn't be able to go in depth on a subject with it if I lacked the expertise. Some work is a result of design and negotiations in those designs.
I don't think Tao's style works with everyone/thing, especially if we don't know what style he's tuned his LLM on.
Overall, I agree, when I ask things I'm an expert in and do professionally every day. I get very good useful answers. When, for example, our marketing people, ask about the science, they often get confusing and wrong answers.
I'm surprised nobody mentioned (including the author) the Gell-Mann Amnesia Effect [1]... Just substitute "LLM" for "journalist" and there you have it.
And to be honest, I have seen it, as I'm sure (almost) everyone has, who has demonstrated experience/expertise in their own fields, and correct the LLM's responses one time or another...
[1] https://en.wikipedia.org/wiki/Michael_Crichton#%22Gell-Mann_...
I think if we have a large population of generalists, then none of them are generalists after all
I believe they would still be called generalists.
Now everyone who feels rewarded by LLMs will conclude that it demonstrates their expertise. ;)
I cant have an overall opinion but in my personal experience i have analysed that LLMs do reward concreteness
Skilled use may or may not matter, depending on the task. Do you need to do what Terence Tao is doing?
Yes. This is called the Matthew Principle:
> For to every one who has will more be given, and he will have abundance; but from him who has not, even what he has will be taken away.
LLMs reward architecture knowledge of how to structure things and how to not just say "Claude, make me Microsoft Flight Simulator, make no mistakes".
Many, including myself, report having a lot of success with braindumping and not structuring anything. Just talking into speech recognition for 2-10 minutes as a stream of consciousness about what my context is, what I want, what I know already, what I have a vague hunch about, how it fits into a bigger picture, what aspects are most important to me, any footguns I already know about, really like having a chat with a person on the phone, with someone you have to guide remotely because they have to implement the thing right now but you have to be out of office and so your only interface is speech. Except you can be more structureless because the AI won't be offended. Just keep on rambling, and press enter, don't even correct mistranscriptions. It will understand it anyway.
Now, the key is, that while rambling without structure, you do have to drop the key facts into your speech, and you have to know what you're talking about in at least a good portion of it.
I think people are afraid of doing it, because it seems "not the right way" or "not scientific" or whatnot. They want to believe there is some magic to writing the right prompt. So let me tell you, it works.
I don't completely disagree with the concept of giving a free association thought process ramble into context. But I also bet that when you start getting it to actually generate code and link modules of things together, subroutines, functions, code structure and filenames, you still pay attention to what it does and you guide it into the architecture that makes logical sense to you.
For real work yes. For personal projects, less and less since Fable came out (probably the same if true of the other frontier models). You can get a lot done if it's just some one off, or a personal tool, even without looking at the code, just trying the application. Frontier models now automatically test it before handing the thing to you, they take screenshots, they fix the superficial issues themselves. To get something up and running, it's enough to send chat messages.
Unfortunately the software industry is saying things like "don't look at the code", "LLMs have made developers 10-100x faster", etc. The only way they can make such claims is by saying what you said above: "Claude, make me Microsoft Flight Simulator, make no mistakes". Additionally engineers are facing pressures via deadlines to work in the paradigm of "Claude, make me Microsoft Flight Simulator, make no mistakes"...
The question i wonder about is, when will an event come along that persuades everyone that human understanding is still required? Or will it never come?
I wonder the same thing. I think we've already seen some of this happening, however the consequences haven't been large enough to the organization, for example:
- https://www.theguardian.com/technology/2026/mar/20/meta-ai-a...
- https://tech.yahoo.com/articles/ai-code-wreaked-havoc-amazon...
- https://alexeyondata.substack.com/p/how-i-dropped-our-produc...
We can only hope that engineers working in safety critical systems haven't fallen to these working conditions.
Such an event would have to be pretty catastrophic at this point to slow down the inertia. Perhaps the tech debt will just pile up until someone's product implodes, or there's a massive safety issue that causes loss of life, or some big hedge fund goes bust.
The event could be when fair pricing comes from the model providers. We're still at the cash burning stage. When the economy crashes a little and departments start monitoring their spending, and the prices for inference are 10x what they are, there will be less tolerance for employees to substitute constant AI usage for understanding.
That question makes me think about Boeing. Or NASA. Or Enron. Reality always wins, no matter what management and Investor Relations says.
It makes me think about The Terminator.
Yes. If you use the right technical terms together it’s lights up more specific feature spaces to your task.
Specificity matters to LLMs a lot.