What did your AI-assisted workflow look like 1 year ago? I can only speak for myself, but I would carefully specify a class or module in great detail and then hand it off to the model to implement, then carefully review the result.
How about 2 years ago? Back then, I wouldn't even trust it to write a 5-line function without making some sort of silly mistake.
Today, I can leave an agent running by itself for 20 or 30 minutes and most of the time, it comes back with a result that's either flawless or can be refined to be good with a few back and forth messages. Maybe I still have to make some high-level decisions ahead of time, but all of the details, including exploring the codebase and figuring out what to do based on that, can be left to the agent. The amount of improvement just in the last 2 years has been staggering.
Now extrapolate how things will look if the trend continues for another 2 or 3 years.
Is this guaranteed to happen? No. But people have been predicting that we're going to hit a wall for a long time now, and we haven't yet. Maybe there's a wall just ahead of us. But maybe there's not -- and the "not" case seems likely enough that we should at least be planning for it.
I disagree with your assessment pretty strongly -- the models themselves hit a wall over a year ago once companies exhausted all existing training data. LLMs don't induce world models, and they aren't capable of real search an planning outside their training distributions. They, structurally, never will be.
I haven't noticed a change in what I trust a model to generate in response to a single prompt in a year. The failure modes are unchanged. Yes, specific failures have improved as they have been documented and passed into model training data, but the way the models fail has not changed. They still fail for me nearly every single day. I'm a pretty heavy user - 3-4 Claude code processes running at a time, all day every day.
What has gotten better is tooling around the model -- but there's no space for exponential growth there. At least, not without exponential cost increase, which would make the whole thing untenable anyway.
If you think they've been at the same level for the past year, your skill is at issue. There was a huge jump in Nov/Dec with Opus 4.5+ and GPT 5.x series, and they've been incrementally stronger over the past 6 months.
As a next step, take another look at the next practices, and apply them to your work (simon's agentic series is a good place to start). Or not, you do you.
I feel like the problem is there aren’t any great metrics. Boris Cherny probably gets paid like $2 mil per year. So what does it mean that Claude writes 100% of his code? And Claude writes 100% of code for most teams? Has Anthropic started laying people off? If Claude is writing 100% of code doesn’t that mean game over?
It’s both amazing and kind of a useless metric. How do I extrapolate out 100% 2-3 years from now? Super-duper 100%? Infinity infinity?
I use AI to generate code very rarely (basically replacing searching code from SO which was also rare) yet maybe 90-95% of my code is machine generated (macros, snippet engines, boilerplate generators, IDE/LSP features etc) if we define as how many characters in the source code out of all characters did I write by my hand.
Before collective AI psychosis nobody cared how much code is hand written and how much machine generated and the difference was absolutely massive between developers. There was also no guaranteed correlation between "productivity" and how much developer hand writes or actually uses their tools well and the same applies with AI usage.
I wonder if our difference in view could be an instance of the jagged nature of AI’s intelligence. I do computational research in a basic science so write code or build models basically all day that is (occasionally) novel. I would say that I’ve noticed exponential improvements in parts of my job but certainly not all. For example, if I’m trying to visualize a concept from a paper I now go straight to Codex, give it the paper, and describe a webapp which allows me to play with the model in a way that wasn’t possible one year ago (this is great for teaching btw). If I have a script that I want to generalize, add in better metrics, or setup for running on a cluster I use codex and it does great.
Where it fails me though is exactly when I’m doing something novel like developing a new model or trying to develop some new method to process data. I’ve tried many times to one shot these ideas with detailed descriptions of what I want, how I’d like to generate abstractions, etc and it almost always ends up changing what I want to what I can only describe as something which better matches its training data. It often quietly changes key details that means that I have to delete the whole thing and start over. Just today this happened. On this level of task I’ve found that my workflow and pace of iteration hasn’t really changed at all in the last year. I still have to go and explain in detail on a function by function level what I want in much the same way I did a year ago. While that’s obviously a harder task, it seems to me like the task this whole long term exponential argument hinges on. I obviously could be wrong and maybe LLM with eval loop will do all of this for us but it seems still quite bad at anything without a clear definition of “good”.
I’m personally much more concerned about autonomous weapons, surveillance, and people plugging these things into places they don’t belong to avoid responsibility than I am the general possibility of these models being smarter than me in every way but obviously I could be wrong on this and am just using it incorrectly, hence the question.
Why would you want to own your own giant datacenter? What would you do with it? Of course it's expensive to operate a datacenter that serves millions of people.
Well, it looks this post has already been flagged down onto page 7.
And IIRC, the same thing happened to the "oh shit" moment thread you linked to. Did the mods have to intervene to get it back on the front page?
HN might not be anti-AI, but I feel like the way flags are weighted by the ranking allows some users that are extremely anti-AI to create the impression that it is.
How is it strange? The "exceeds $5B" quote was from December 2025. Anthropic has seen tremendous growth since then, ever since Claude Code with Opus 4.5 got really good at coding.
If you've ever been at a startup, this is exactly what it looks like when you go from not having product-market fit to having it (though with a few extra zeros on the end compared to most).
All three of those are official releases from Anthropic. You can choose not to believe the if you like, but since they plan to IPO this year it's in their interest not to get caught lying to potential investors.
As a non-public company they can use whatever non-GAAP black magic accounting to claim anything while still not technically lying. It just doesn't correlate to anything we'd actually call revenue.
Not at all, because the magic can be applied differently at different times. They're undergoing a funding run now so they've got a massive incentive to come up with all kinds of revenue.
They've also signed a deal for billions in compute with xAI for april-may so they're certainly using that to fake billions in revenue using non-GAAP bullshit. It just seems a tad more likely than them legitimately increasing actual revenue by 233% in four months out of the blue.
Sorry man I hate to say this but you and many others need to stop commenting on accounting, finance and economics as its clearly way out of your realm of expertise.
Do you know what revenue recognition is? Do you know what accrual accounting is? Do you know of the phenomenon that is 'managed earnings'?
The only true objective number in finance is cash flows.
> Their revenue has quadrupled in just a few months
Maybe, maybe not. We haven't seen that S-1 yet. All we have is the 5B in lifetime so far. PLUS - revenue quadrupled or not, it only matters if their costs did not expand at the same rate or more. Revenue is not profit.
OK, so when S-1 comes out you will finally allow yourself to be wrong? Your prior is, a 1T company plans to IPO and their leader has been loudly committing an insane amount of fraud? I mean this of course is possible but that is quite the conspiracy. The scrutiny of an IPO would be a crazy thing to do if you were committing fraud at the scale you're suggesting.
Revenue is not profit yet the discussion in this particular thread is about revenue.
No, it's not. It's a stupid thing to say. Perfectly stupid assumption. There are 1000s of multi billion $ revenue companies operating and as a % the number that are fraudulent is close to zero, especially those public or looking to go public. There is always the possibility, but it's extremely naive to think it's likely.
Yes, however, my point is that the vast majority of people working for non-profits do not receive that sort of recognition. So what does "social status" mean?
The author labels COVID and the launch of ChatGPT on the graph, but fails to mention that Stack Overflow was acquired in June 2021 by Prosus, a Dutch private equity firm. That looks to me like it matches pretty well with the entire downward trend.
That would be Joel Spolsky (Fog Creek Software) and Jeff Atwood (Coding Horror), mostly. Jeff has gone on to make several large philanthropic gifts. Joel probably has too but I don't have info on them.
It’s not necessarily the sale. Some private equity companies move from “Let’s invest like we’re shooting for the moon” to “Let’s invest like we want to improve margins and flip this on 3-5 years”
It’s not inherently wrong but it is a different model, and sometimes companies suffer as a result.
Businesses (and any other kind of asset) are sold for all kinds of reasons, and trying to time the market to maximize the price is only one of them. Probably not even the most common one.
I'd disagree that the site wasn't changing. I think they were already trying to sideline job portal possibilities because it wasn't making a high enough worth calculation compared to entirely unrealized estimates. However my reaction to changes was forgiving for the old firm while feeling transactional was basically doom to my using the site as I didn't really need anything from interactions.
It had the possibility of surfacing employers or coworkers with a little more comfort about their ~support competence. The enterprise stuff seems more like the stealing an OSS contributor who can't deny competence to work on internal Jira from now on..
The 22:20 timestamp from the body of the post is wrong. The timeline section (where the 22:10 timestamp came from) is consistent with itself, and also contains:
> May 19, 22:19 UTC - Root cause identified: Google Cloud Platform has suspended Railway's production account.
They couldn't have identified the root cause before it happened.
In any practical application there'll be a known set of errors and I'm generally fine merging code that has known deficiencies. But personally, I'd not condone merging anything that causes UB. It undermines such a fundamental guarantee of the language that it should be detected and eliminated. And bun certainly rises to the level of software where I'd expect that the project runs all available tooling to detect such cases. Especially if you LLM - code it. "Do not cause UB" should be part of the test harness.
How about 2 years ago? Back then, I wouldn't even trust it to write a 5-line function without making some sort of silly mistake.
Today, I can leave an agent running by itself for 20 or 30 minutes and most of the time, it comes back with a result that's either flawless or can be refined to be good with a few back and forth messages. Maybe I still have to make some high-level decisions ahead of time, but all of the details, including exploring the codebase and figuring out what to do based on that, can be left to the agent. The amount of improvement just in the last 2 years has been staggering.
Now extrapolate how things will look if the trend continues for another 2 or 3 years.
Is this guaranteed to happen? No. But people have been predicting that we're going to hit a wall for a long time now, and we haven't yet. Maybe there's a wall just ahead of us. But maybe there's not -- and the "not" case seems likely enough that we should at least be planning for it.
reply