Indeed this will likely happen in the future, but not today. I was experimeting with SSD streaming in DwarfStar for DeepSeek v4 PRO inference in 128GB systems (and Flash inference iwth 32/64). GPT 5.5 ran the whole night, I checked what it had accomplished regardless of all the hints I provided in the specification document. After reasoning on the problem I gave him the design fixes and the tokens/sec were 4x after 10 minutes. And this is true for every domain where the human babysitting the AI know a few things in that domain. However this is a moving target, and at the current rate, soon or later, indeed AIs will do much better than us in many domains.
I don't think this makes any sense. Companies with managers that think AI capabilities are superior will be replaced if they are wrong as the companies will perform very poorly.
Actually workflow impact in the world of software can be observed in weeks/months at max. And token spending too, is a voice that they see at the high floors. Also, there was never a strong willing in IT companies to reduce cost of work force: it is done sometimes, but it is more common to see them over-hiring.
Hello, efficient market fallacy. Markets are not actually efficient, and especially not instantaneously so. There are a million ways to observe various market inefficiencies[1], so it's childishly naive to assert that they are in fact perfectly efficient according to some ideological belief of yours without considering reality.
What I say has nothing to do with efficient market hypothesis. Here the question is simpler: in small companies where there are competitors, who does the wrong choices will be seriously hit since customers will star preferring less slop and more reliability, if AI is mis-used. And companies that instead of firing, hire the folks that are "ideas people" and can use AI efficiently, and now how to control the quality of the output, will deliver more and better. For bigger companies: AI is driving salaries at a more normal level (honestly we want a bit too high, in recent years, even for people with a very low level of knowledge, didn't we?) and to marginally reduce total spending and not deliver the timeline they have, and are used to observe for years, will be noticed. Also companies in the past had a dangerous tendency to over-hire. I don't think now they will invert the direction and over-fire. I have the feeling many managers will instead reason in terms: what is today the great programmer fit? The one with low level knowledge of each algorithm, or the one that has good ideas and understands product, quality, processes, other than programming? And they will try to mix AI and people in order to have an edge.
I think we're in partial agreement on some things. I agree that the software field overhired and overpaid people who should never have had jobs in software in the first place, and that a correction is/was overdue. I also generally agree that small companies cannot afford to produce garbage software, and if they make poor decisions about hiring or AI usage, they will die in the womb. But startups failing is not really what I think of when somebody says "companies will be replaced" or "your job security is contigent on what management thinks of AI capabilities". Those sentences both convoke images of already-successful enterprise companies, and already-successful enterprise companies are the ones that are most resistant to market forces. Indeed, we already see this in the real world, because most enterprise companies produce truly horrifyingly bad software, even before AI. The secret is that you need to produce good software to become successful, and then once successful, network effects take over and your company can become unbelievably inefficient and have little to no fear of being replaced. Tech is a ridiculously winner-take-all field, and it's very common for a single company to capture over 50% of their market, after which point they are effectively irreplaceable no matter how many bad decisions they make, at least for many years if not decades.
"The factory of the future will have only two employees, a man and a dog. The man will be there to feed the dog. The dog will be there to keep the man from touching the equipment."
How surprised people will be when they learn that their prompts and skills and etc are being saved for ai training even though they said it would not be
I had the same thought about ML engineering a while back, when Google released the AutoML suite, that was banned from Kaggle competitions. At the time, it seemed obvious to me that the closer you were to the models, the easier it was for them to replace your work, since most of the work on models was itself hill-climbing, grind searching and mutation search. So, the more your work is an explicit, measurable search loop, the more automatable it is.
Same with prompts, most attempts seem to be fidgeting with the models till they get your intend right, which is also a matter of hill-climbing, subtle mutation, and so on.
If I were to clarify anything from the article, I'd probably say that I'd rather do the factorisation of programming roles by how long they already existed. If someone is an AI engineer and his work only became relevant a month ago, very probably it will be obsolete in another month. If they do the same thing for the past 10 years, changes are that their skills would be useful for another 10 years to come.
Of course not, all these sloppers are doing is training the models so at the eyes of management they are good enough for a replacement. The ones who stay will have 10x more work.
They select for people who are beholden to AI, probably to eventually have the model do the job of prompting if the model is expected to be doing the doing anyhow. Anthropic job posts I've seen have explicitly said you should use claude to claude-ify your resume before submitting. I'm guessing it's an auto reject if you don't. If they are asking for you to use their ai tool for step 0 before you even work there, they are going to want you to use it for all your job functions and communications. And all of that will be logged, used as training data, and will justify not hiring to fill your seat when you leave or get canned.
It's fun to think about why that is. Most compilers explore the code into sea of nodes, with explicit relation kinds (depends-on, computes, effect relations etc) that humans don't have access to in surface languages. Then they just reduce and shorten a lot of the unnecessary edge chains, sometimes duplicate code that improves scheduling and a lot of not-so-semantically relevant stuff to get a dataflow that is guaranteed to have the same external behaviour with the code they're compiling. So basically compilers work on representations that humans almost never see, with different objectives that humans have. If humans were to code directly in dataflow networks and could find a way to keep them tidy and neat, I think humans would have a chance to beat solutions generated from surface code and then compiled automatically.
I am pretty familiar with a 500k LOC codebase. If for every feature request/bug the agent has to go through a lot of it, spend a gazillion thinking tokens for understanding what it needs to do, plan, and then execute (assuming it gets it right) given the current cost of tokens I argue I am often more cost effective.
In fact, I believe that the most cost effective way is a collab of human+agent. Ie giving the agent direction as it goes along with the plan I can cut the thinking while keeping the speed. Basically helping the agent going from a breadth first search into a guided depth first one which is much more token efficient.
Additionally, humans have long term memory and knowledge of the context around your codebase. Agents do not, and while you can fit a lot in 1M context window, once you fill that the quality goes down considerably.
Not to mention the fact that even the most well documented codebase will have documentation blindspots about real-world concerns or limitations that LLMs cant know about. Cursor yesterday tried to remove a document format from the codebase because it was convinced that it was non-existant, turns out that not only does it exist and is vitally important for our shipping process, but also the API it comes from does not document its existence at all.
This is why you can't be replaced today. I'm not sure you can rely on that remaining true for very long. And this goes for the vast majority of us.
To be clear, I'm also not saying LLMs will definitely displace a lot of us very soon. I'm just saying I wouldn't be surprised by either outcome and I don't know how anyone claims to know one way or another given the past year or so of progress.
Im curious if that point comes before it automates away the entire mid-upper management caste.
In a hypothetical world where LLMs have enough context window and "understanding" to have no need for an experienced user to give inputs I would assume its also going to have enough information to make most business decisions and provide well formatted info to the C-Suite.
I think the missing piece here is nuance. Of course there are certain tasks that software engineers do that will be replaced. But will AI replace _everything_ a software engineer does?
The most difficult bit about software engineering is to keep a mental model of _everything_ a product does with varying levels of granularity. The way I see LLMs fail at my company the most is that they are very good at the big picture, and very good at the very small picture, but have difficulty moving between those two levels. And especially when changes have occurred or accumulated over time. Most of all production systems have an extremely long tail of gotchas which are only managed by people who have been around for long enough to have some kind of deep storage access in their heads to those little tidbits of information.
And I think current LLMs might be fundamentally incapable of replacing that.
Maybe tomorrow, but they're just still not there quite yet. For non-software developers, there's a "vibe coding wall" that gets hit and the project needs a software developer to unwind. Maybe Mythos or codex-5.6 will be able to do it, but Opus 4.8 and codex-5.5 still get stuck without proper guidance.
I've yet to hear an argument that argues that software engineers can be replaced by AI that doesnt boil down to slop apologism, inability to detect slop or simple gaslighting.
These are things I've come to expect from bots, clueless journalists, clueless juniors, clueless expert beginners and clueless members of the professional managerial class but almost never from experienced software engineers.
To be fair, seasoned software engineers always seem to get shouted down online by the former group which is louder and more numerous so you could argue that we "lost" the argument.
Meanwhile big tech's vibe coded monstrosities are increasingly exploding all around us in ever more humiliating ways while the humans who had this tech rammed down their throats get thrown under the bus.
This undeserved halo effect over AI is maintained in order to keep the needle from pricking the ginormous stock market bubble that hinges upon the religious belief in the lie AI Will Replace Us All Soon.
The argument is "Software Engineer" sounds like "Programmer" to me and "Programming" is just typing lines of code, AI can do all that typing quicker than a human so there we go.
Currently leading an Integration that for the most part needs no new code written and the CEO is breathing down my neck telling me to cut my 4 week estimate down to 1 because "can't i just use AI like the other firms do?".
There's a morbid part of me that wants to give him what he wants and let claude make critical process decisions on internal processes that are very domain specific and have no online documentation, but alas I would rather not have the project go down in flames so I smile and nod.
Much less emasculating than accepting that all that weird tech mumbo jumbo that those overpaid senior engineers babble about actually has a deeper meaning and is not just there to artificially slow down the all important company growth!
> I've yet to hear an argument that argues that software engineers can be replaced by AI that doesnt boil down to slop apologism, inability to detect slop or simple gaslighting.
That, or extreme extrapolation from events that form a vanishingly tiny part of the job of a software engineer. "Last week my AI solved this amazing software problem that I had struggled with" very quickly becomes "the AI is better at software than I am". Any pushback suggesting that the fact that something (or someone) did one tiny part of your job better than you one time does not mean you should be replaced, is quickly met with "yeah, but that's today, imagine how amazing the models will be in n years".
You can't win a debate with this much moving of goalposts.
The article is quite general. Here's some notes on how AI is being used to do AI research at frontier labs specifically. It's not the singularity (yet?) but it's heading in that direction.
Most training is now actually inference, not directly gradient descent. Reinforcement learning requires the generation of lots of 'rollouts' that are then compared with each other via an algorithm like GRPO. Or they might be compared using a critic model - AI judging AI and causing it to self improve. Generating a rollout means inference. And there's lots of data cleaning by older models. This has been called in the past 'textbook' or 'curriculum' learning, not sure what it's called now. But AI is also used for things like data/document labelling, transcription of videos, detection of images/videos with watermarks or subtitles, elimination of content that shouldn't be in the dataset, creation of new content that should and so on.
AI has proven capable of some routine work, like brute-force optimizing GPU kernels or doing hyperparameter sweeps.
Obviously, researchers are all using coding agents too.
So that's a few ways AI is self-improving. But there are lots of other ways in which even frontier models are still beaten by human researchers. Experiments in closing the loop have failed. For instance, people have tried giving the latest models access to some GPUs and an old version of an AI codebase that was recently optimized by human researchers (a NanoChat speed run goal, I believe). Could the models match the performance of the AI researchers? Nope. They only got 10% as far as the humans did, mostly because their approach was uninspired. They wasted a lot of time and budget doing low-IQ stuff like hyperparameter tuning. The humans had many other tactics like studying the research literature and inventing new algorithms that the models didn't even attempt.
The bottleneck is therefore currently the level of insight and inspiration the models are capable of. I've also seen this in my own work. I come up with an idea I think is novel and see if I can get a frontier model to reach the same idea. It never works without questions so leading it's more or less pointless.
It's very unclear why AI struggles so much with innovation yet can invent new songs, poems etc without apparent difficulty. Obvious answers like "it's not in the training set" don't feel right to me, the issue is deeper.
It's the Einstein question. Given an LLM trained on all written word up until right before Einstein's work, could the current SOTA in LLM architecture rediscover relativity? The jury is out on that, but until someone runs that experiment and proves otherwise, we have to assume LLMs simply aren't capable of that kind of brilliance. They're still impressive and useful and also stupid at times, but at the end of the day, no LLM has a gut to make a gut decision with. Unlike us.
I'm not convinced by this. I've had good managers and bad managers. Generally the manager isn't "doing the work", so much as setting the direction and smoothing the path. The current state of AI tools still need "good managers" to set the direction, otherwise they end up nowhere. Especially in large complex projects.
Maybe at some point the AI tooling will be good enough for me to say "do my work for me today" and sit back. At that point, yes, I am irrelevant and could be replaced by anyone else. But is it anywhere close to that right now? My experience says no.
Perhaps the current models are capable of that with the right tooling - some system to define clear goals and stick to them. I haven't seen evidence of that yet though. Have other people?
I'm not sure what the take-home of this message is since you can replace "AI" with just about anything... "If you use a keyboard to do your work, you can be replaced by someone else using a keyboard to do your work." Sure? You can always be replaced by someone/something that can do your job better.
A keyboard doesn't "do" the work. In other words, the more work you outsource to AI, the lower your value-add becomes, the easier your are to replace by someone doing the same thing for cheaper.
I’d reframe it: you won’t be replaced by someone using AI, you’ll be replaced by someone who is better at using AI and understands the code it generates
Over the last couple of years, I’ve seen plenty of developers who remain barely competent despite having access to powerful AI tools. Generating code is easy. Evaluating whether it’s actually correct and maintainable is the hard part.
If Ai can generate code, review code, validate correctness, understand reqs, make architectural tradeoffs, operate systems, and take responsibility for outcomes, then we're no longer talking about replacing programmers. We’re talking about replacing most knowledge workers
At that point the debate isn’t really about software engineering anymore
> > Evaluating whether it’s actually correct and maintainable is the hard part.
> But AI can also do that.
Citation needed.
> So, what’s the point?
The point is that there haven't been broad demonstrations of your claim.
> And if you think it can’t, wait one more year
You surely must understand that this isn't an argument? How many hundreds of billions have been burned through now? Yet we still have to suffer "soon" as an argument? I can't take any of this seriously anymore.
PS: Just to be absolutely sure you don't misunderstand me: I am NOT claiming that AI will never be able to do this stuff. Nor am I even claiming that it's too far off or too expensive. Just, for the love of god, you cannot build an industry on promises of how amazing it'll be in the future. Technology is evaluated based on how it performs. Not how you think it might perform in the future.
PPS: The last paragraph does also not mean that I think it's bad to invest in things that haven't yet paid off. On the contrary! What I am saying is you cannot claim success until there's success!
Indeed this will likely happen in the future, but not today. I was experimeting with SSD streaming in DwarfStar for DeepSeek v4 PRO inference in 128GB systems (and Flash inference iwth 32/64). GPT 5.5 ran the whole night, I checked what it had accomplished regardless of all the hints I provided in the specification document. After reasoning on the problem I gave him the design fixes and the tokens/sec were 4x after 10 minutes. And this is true for every domain where the human babysitting the AI know a few things in that domain. However this is a moving target, and at the current rate, soon or later, indeed AIs will do much better than us in many domains.
Your job safety doesn’t depend on the capabilities of AI but on what management thinks are the capabilities of AI
I don't think this makes any sense. Companies with managers that think AI capabilities are superior will be replaced if they are wrong as the companies will perform very poorly.
The lag time between firing your core team and finding out that was a bad idea can be measured in years of slow attrition.
Actually workflow impact in the world of software can be observed in weeks/months at max. And token spending too, is a voice that they see at the high floors. Also, there was never a strong willing in IT companies to reduce cost of work force: it is done sometimes, but it is more common to see them over-hiring.
Hello, efficient market fallacy. Markets are not actually efficient, and especially not instantaneously so. There are a million ways to observe various market inefficiencies[1], so it's childishly naive to assert that they are in fact perfectly efficient according to some ideological belief of yours without considering reality.
[1] Some examples: https://danluu.com/nothing-works/
What I say has nothing to do with efficient market hypothesis. Here the question is simpler: in small companies where there are competitors, who does the wrong choices will be seriously hit since customers will star preferring less slop and more reliability, if AI is mis-used. And companies that instead of firing, hire the folks that are "ideas people" and can use AI efficiently, and now how to control the quality of the output, will deliver more and better. For bigger companies: AI is driving salaries at a more normal level (honestly we want a bit too high, in recent years, even for people with a very low level of knowledge, didn't we?) and to marginally reduce total spending and not deliver the timeline they have, and are used to observe for years, will be noticed. Also companies in the past had a dangerous tendency to over-hire. I don't think now they will invert the direction and over-fire. I have the feeling many managers will instead reason in terms: what is today the great programmer fit? The one with low level knowledge of each algorithm, or the one that has good ideas and understands product, quality, processes, other than programming? And they will try to mix AI and people in order to have an edge.
I think we're in partial agreement on some things. I agree that the software field overhired and overpaid people who should never have had jobs in software in the first place, and that a correction is/was overdue. I also generally agree that small companies cannot afford to produce garbage software, and if they make poor decisions about hiring or AI usage, they will die in the womb. But startups failing is not really what I think of when somebody says "companies will be replaced" or "your job security is contigent on what management thinks of AI capabilities". Those sentences both convoke images of already-successful enterprise companies, and already-successful enterprise companies are the ones that are most resistant to market forces. Indeed, we already see this in the real world, because most enterprise companies produce truly horrifyingly bad software, even before AI. The secret is that you need to produce good software to become successful, and then once successful, network effects take over and your company can become unbelievably inefficient and have little to no fear of being replaced. Tech is a ridiculously winner-take-all field, and it's very common for a single company to capture over 50% of their market, after which point they are effectively irreplaceable no matter how many bad decisions they make, at least for many years if not decades.
And such management faults never ever happened before.
When all the prompts are by AI, and all the commits by AI, and all the use by AI,only then will corporations realize that..something ..
"The factory of the future will have only two employees, a man and a dog. The man will be there to feed the dog. The dog will be there to keep the man from touching the equipment."
How surprised people will be when they learn that their prompts and skills and etc are being saved for ai training even though they said it would not be
I'm not sure if the authors means "ai engineer" like that.
I had the same thought about ML engineering a while back, when Google released the AutoML suite, that was banned from Kaggle competitions. At the time, it seemed obvious to me that the closer you were to the models, the easier it was for them to replace your work, since most of the work on models was itself hill-climbing, grind searching and mutation search. So, the more your work is an explicit, measurable search loop, the more automatable it is.
Same with prompts, most attempts seem to be fidgeting with the models till they get your intend right, which is also a matter of hill-climbing, subtle mutation, and so on.
If I were to clarify anything from the article, I'd probably say that I'd rather do the factorisation of programming roles by how long they already existed. If someone is an AI engineer and his work only became relevant a month ago, very probably it will be obsolete in another month. If they do the same thing for the past 10 years, changes are that their skills would be useful for another 10 years to come.
"Engineers"
Of course not, all these sloppers are doing is training the models so at the eyes of management they are good enough for a replacement. The ones who stay will have 10x more work.
If they can actually deliver 10x more work: well, that's how economic progress looks like.
[dead]
The people who stick around will be the ones who deeply understand the problem, not the ones who are good at wrangling the tools.
They select for people who are beholden to AI, probably to eventually have the model do the job of prompting if the model is expected to be doing the doing anyhow. Anthropic job posts I've seen have explicitly said you should use claude to claude-ify your resume before submitting. I'm guessing it's an auto reject if you don't. If they are asking for you to use their ai tool for step 0 before you even work there, they are going to want you to use it for all your job functions and communications. And all of that will be logged, used as training data, and will justify not hiring to fill your seat when you leave or get canned.
Good news - Anthropic will pay you £600k to build the AI which replaces you.
Drag it out for a couple of years and you'll be set.
Compiler developers couldn't beat compilers at generating code either.
It's fun to think about why that is. Most compilers explore the code into sea of nodes, with explicit relation kinds (depends-on, computes, effect relations etc) that humans don't have access to in surface languages. Then they just reduce and shorten a lot of the unnecessary edge chains, sometimes duplicate code that improves scheduling and a lot of not-so-semantically relevant stuff to get a dataflow that is guaranteed to have the same external behaviour with the code they're compiling. So basically compilers work on representations that humans almost never see, with different objectives that humans have. If humans were to code directly in dataflow networks and could find a way to keep them tidy and neat, I think humans would have a chance to beat solutions generated from surface code and then compiled automatically.
Loosely related: Long read in today's FT on "The race to build AI that can improve itself" (https://www.ft.com/content/7cc7800f-18ed-47d8-9539-221ae3e16...).
Although this may be more relevant to replacing AI researchers, not AI engineers...
(Submitted as https://news.ycombinator.com/item?id=48380643 )
... and so the skynet saga begins
For some of us that day can't come soon enough @}-;-'---
I have yet to hear a single convincing argument by a person that works in software why they can't be replaced.
I am pretty familiar with a 500k LOC codebase. If for every feature request/bug the agent has to go through a lot of it, spend a gazillion thinking tokens for understanding what it needs to do, plan, and then execute (assuming it gets it right) given the current cost of tokens I argue I am often more cost effective.
In fact, I believe that the most cost effective way is a collab of human+agent. Ie giving the agent direction as it goes along with the plan I can cut the thinking while keeping the speed. Basically helping the agent going from a breadth first search into a guided depth first one which is much more token efficient.
Additionally, humans have long term memory and knowledge of the context around your codebase. Agents do not, and while you can fit a lot in 1M context window, once you fill that the quality goes down considerably.
Not to mention the fact that even the most well documented codebase will have documentation blindspots about real-world concerns or limitations that LLMs cant know about. Cursor yesterday tried to remove a document format from the codebase because it was convinced that it was non-existant, turns out that not only does it exist and is vitally important for our shipping process, but also the API it comes from does not document its existence at all.
This is why you can't be replaced today. I'm not sure you can rely on that remaining true for very long. And this goes for the vast majority of us.
To be clear, I'm also not saying LLMs will definitely displace a lot of us very soon. I'm just saying I wouldn't be surprised by either outcome and I don't know how anyone claims to know one way or another given the past year or so of progress.
Im curious if that point comes before it automates away the entire mid-upper management caste.
In a hypothetical world where LLMs have enough context window and "understanding" to have no need for an experienced user to give inputs I would assume its also going to have enough information to make most business decisions and provide well formatted info to the C-Suite.
Well the fact that 0 developers are replaced by AI should tell you something
I think the missing piece here is nuance. Of course there are certain tasks that software engineers do that will be replaced. But will AI replace _everything_ a software engineer does? The most difficult bit about software engineering is to keep a mental model of _everything_ a product does with varying levels of granularity. The way I see LLMs fail at my company the most is that they are very good at the big picture, and very good at the very small picture, but have difficulty moving between those two levels. And especially when changes have occurred or accumulated over time. Most of all production systems have an extremely long tail of gotchas which are only managed by people who have been around for long enough to have some kind of deep storage access in their heads to those little tidbits of information. And I think current LLMs might be fundamentally incapable of replacing that.
to be fair, I'm not really seeing many people insisting otherwise, not anymore at least.
Don't fall for it. Its a trick question. Capitalism made everybody replaceable a long time ago. :P
The more interesting question is: Has the ease of replacibility increased because of AI?
Maybe tomorrow, but they're just still not there quite yet. For non-software developers, there's a "vibe coding wall" that gets hit and the project needs a software developer to unwind. Maybe Mythos or codex-5.6 will be able to do it, but Opus 4.8 and codex-5.5 still get stuck without proper guidance.
I've yet to hear an argument that argues that software engineers can be replaced by AI that doesnt boil down to slop apologism, inability to detect slop or simple gaslighting.
These are things I've come to expect from bots, clueless journalists, clueless juniors, clueless expert beginners and clueless members of the professional managerial class but almost never from experienced software engineers.
To be fair, seasoned software engineers always seem to get shouted down online by the former group which is louder and more numerous so you could argue that we "lost" the argument.
Meanwhile big tech's vibe coded monstrosities are increasingly exploding all around us in ever more humiliating ways while the humans who had this tech rammed down their throats get thrown under the bus.
This undeserved halo effect over AI is maintained in order to keep the needle from pricking the ginormous stock market bubble that hinges upon the religious belief in the lie AI Will Replace Us All Soon.
The argument is "Software Engineer" sounds like "Programmer" to me and "Programming" is just typing lines of code, AI can do all that typing quicker than a human so there we go.
Currently leading an Integration that for the most part needs no new code written and the CEO is breathing down my neck telling me to cut my 4 week estimate down to 1 because "can't i just use AI like the other firms do?".
There's a morbid part of me that wants to give him what he wants and let claude make critical process decisions on internal processes that are very domain specific and have no online documentation, but alas I would rather not have the project go down in flames so I smile and nod.
Doctor sounds like Nurse, which sounds like applying bandages and taking temperatures.
Physicist sounds like Lab Technician, which sounds like managing samples.
Electrical Engineer sounds like Electrician, which sounds like installing a bunch of wire.
Stunt Driver sounds like Uber Driver which sounds like pushing pedals and turning a wheel.
It’s fun to pretend the world is much simpler than it is.
Much less emasculating than accepting that all that weird tech mumbo jumbo that those overpaid senior engineers babble about actually has a deeper meaning and is not just there to artificially slow down the all important company growth!
> I've yet to hear an argument that argues that software engineers can be replaced by AI that doesnt boil down to slop apologism, inability to detect slop or simple gaslighting.
That, or extreme extrapolation from events that form a vanishingly tiny part of the job of a software engineer. "Last week my AI solved this amazing software problem that I had struggled with" very quickly becomes "the AI is better at software than I am". Any pushback suggesting that the fact that something (or someone) did one tiny part of your job better than you one time does not mean you should be replaced, is quickly met with "yeah, but that's today, imagine how amazing the models will be in n years".
You can't win a debate with this much moving of goalposts.
The article is quite general. Here's some notes on how AI is being used to do AI research at frontier labs specifically. It's not the singularity (yet?) but it's heading in that direction.
Most training is now actually inference, not directly gradient descent. Reinforcement learning requires the generation of lots of 'rollouts' that are then compared with each other via an algorithm like GRPO. Or they might be compared using a critic model - AI judging AI and causing it to self improve. Generating a rollout means inference. And there's lots of data cleaning by older models. This has been called in the past 'textbook' or 'curriculum' learning, not sure what it's called now. But AI is also used for things like data/document labelling, transcription of videos, detection of images/videos with watermarks or subtitles, elimination of content that shouldn't be in the dataset, creation of new content that should and so on.
AI has proven capable of some routine work, like brute-force optimizing GPU kernels or doing hyperparameter sweeps.
Obviously, researchers are all using coding agents too.
So that's a few ways AI is self-improving. But there are lots of other ways in which even frontier models are still beaten by human researchers. Experiments in closing the loop have failed. For instance, people have tried giving the latest models access to some GPUs and an old version of an AI codebase that was recently optimized by human researchers (a NanoChat speed run goal, I believe). Could the models match the performance of the AI researchers? Nope. They only got 10% as far as the humans did, mostly because their approach was uninspired. They wasted a lot of time and budget doing low-IQ stuff like hyperparameter tuning. The humans had many other tactics like studying the research literature and inventing new algorithms that the models didn't even attempt.
The bottleneck is therefore currently the level of insight and inspiration the models are capable of. I've also seen this in my own work. I come up with an idea I think is novel and see if I can get a frontier model to reach the same idea. It never works without questions so leading it's more or less pointless.
It's very unclear why AI struggles so much with innovation yet can invent new songs, poems etc without apparent difficulty. Obvious answers like "it's not in the training set" don't feel right to me, the issue is deeper.
It's the Einstein question. Given an LLM trained on all written word up until right before Einstein's work, could the current SOTA in LLM architecture rediscover relativity? The jury is out on that, but until someone runs that experiment and proves otherwise, we have to assume LLMs simply aren't capable of that kind of brilliance. They're still impressive and useful and also stupid at times, but at the end of the day, no LLM has a gut to make a gut decision with. Unlike us.
If you use AI to do your work, you can be replaced by someone else using AI to do your work.
I'm not convinced by this. I've had good managers and bad managers. Generally the manager isn't "doing the work", so much as setting the direction and smoothing the path. The current state of AI tools still need "good managers" to set the direction, otherwise they end up nowhere. Especially in large complex projects.
Maybe at some point the AI tooling will be good enough for me to say "do my work for me today" and sit back. At that point, yes, I am irrelevant and could be replaced by anyone else. But is it anywhere close to that right now? My experience says no.
Perhaps the current models are capable of that with the right tooling - some system to define clear goals and stick to them. I haven't seen evidence of that yet though. Have other people?
I'm not sure what the take-home of this message is since you can replace "AI" with just about anything... "If you use a keyboard to do your work, you can be replaced by someone else using a keyboard to do your work." Sure? You can always be replaced by someone/something that can do your job better.
A keyboard doesn't "do" the work. In other words, the more work you outsource to AI, the lower your value-add becomes, the easier your are to replace by someone doing the same thing for cheaper.
I’d reframe it: you won’t be replaced by someone using AI, you’ll be replaced by someone who is better at using AI and understands the code it generates
Over the last couple of years, I’ve seen plenty of developers who remain barely competent despite having access to powerful AI tools. Generating code is easy. Evaluating whether it’s actually correct and maintainable is the hard part.
> Evaluating whether it’s actually correct and maintainable is the hard part.
But AI can also do that. So, what’s the point? And if you think it can’t, wait one more year
If Ai can generate code, review code, validate correctness, understand reqs, make architectural tradeoffs, operate systems, and take responsibility for outcomes, then we're no longer talking about replacing programmers. We’re talking about replacing most knowledge workers
At that point the debate isn’t really about software engineering anymore
What time to be alive, eh?
Sam?
> > Evaluating whether it’s actually correct and maintainable is the hard part.
> But AI can also do that.
Citation needed.
> So, what’s the point?
The point is that there haven't been broad demonstrations of your claim.
> And if you think it can’t, wait one more year
You surely must understand that this isn't an argument? How many hundreds of billions have been burned through now? Yet we still have to suffer "soon" as an argument? I can't take any of this seriously anymore.
PS: Just to be absolutely sure you don't misunderstand me: I am NOT claiming that AI will never be able to do this stuff. Nor am I even claiming that it's too far off or too expensive. Just, for the love of god, you cannot build an industry on promises of how amazing it'll be in the future. Technology is evaluated based on how it performs. Not how you think it might perform in the future.
PPS: The last paragraph does also not mean that I think it's bad to invest in things that haven't yet paid off. On the contrary! What I am saying is you cannot claim success until there's success!
Let's not offend actual real engineers.
I suspect everyone has a bit of „my job is special“ delusion tbh just with varying degrees of self awareness
There are already AI certificafions for engineer on very new/ever changing technology.
This comment certifies you as ai expert.
[dead]
[dead]