the opinions in question are that bash should be enabled by default with no restrictions, that the agent should have access to every file on your machine from the start, and that npm is the only package manager worth supporting. Bold choices.
To save others a click, though the article is worth reading.
He also mentions no subagents by default in pi as well.
Part of how AI works is that it's just really complicated compression, you can get AI to write out Harry Potter novels word for word with the right prompting.
When it picks out a rare bit of code, it will be simply copying that code, illegally, and presenting it without attribution or any licenses which is in fact breaking the law but AI companies are too important for the law to apply to them.
There's been instances where models have spat out comments in code that mention original authors, etc., effectively outing itself as a copyright thief.
There's nothing anyone can do about it, but the suspicion is that the big companies have taken everyone's code on GitHub, without consent, and trained on it.
And now are spitting out big chunks of copyrighted code and presented it as somehow transformed even though all they've actually done is change a few variable names.
It is copyright theft, but because programmers are little people, not Disney, we don't have any recourse.
And now are spitting out big chunks of copyrighted code and presented it as somehow transformed even though all they've actually done is change a few variable names.
It's pretty likely that I've done the same thing. I mean, I've written enough CRUD functions in my life, for example, that in all likelihood I'm regurgitating stuff that's a copy, for all practical purposes, of stuff I've done before as work-for-hire for my employer. I'm not stealing intentionally or consciously, but it seems quite likely that it's happening. And that's probably true for many of you, at least that have been in the industry for a while.
> There's nothing anyone can do about it, but the suspicion is that the big companies have taken everyone's code on GitHub, without consent, and trained on it.
I asked agent X what is the source of training data it generated code from, it couldn’t say. Then I asked why the code implementation is exactly the same as the output of agent Y. It said they were trained on the same ‘high-quality library’, and still couldn’t say which one.
So I guess that’s fine because everyone is doing it.
When I write fizzbuzz do I owe royalties to the inventor of fizzbuzz? Is my brain copyright thieving because I can write out the song lyrics from memory?
It's a shame you feel that way as articles like this were extremely common 30 years ago in the Saturday and Sunday papers. And I do miss them.
I wonder if it's a generational thing, where now every essay must focus on one idea instead of taking a meandering path of curiosity to the author's final point.
I agree with you that this is a Parade-level essay, and I think those were always basically fluff: enough to give a sense that something had been learned about two or three seeming-disparate topics from a great distance, but not enough to actually satisfy curiosity about any of them. In the 80s as a pre-teen, those were the first part of the Sunday paper I claimed, but they never did more than whet.
I don't think it's even about staying on one topic. It's just that the topics were too broad for such a short essay, even had it been relatively dense, and then to that problem the author added long phrases where a short phrase would do, as if trying to pad the length to some predetermined mark.
But OpenAI/Anthropic are not selling the compute as they're buying that from Google/Amazon/etc.
So they're selling the transformation, or the model. Or the ability to make a model. And their brand and their harness.
And it seems like the model is definitely not worth 380 billion. Models depreciate incredibly fast. There are lots of models and the other models aren't that far behind.
And it seems like the harness is not worth much as there's already open source alternatives that people claim are better.
And all these companies are paying lots of money for these AI training experts.
But I suspect that any regular Hacker News reader of 10 years dev experience could become a training expert in months if allowed to play with a load of compute and a lot of data for a bit.
Just like any of us could have become a data scientist, this stuff is not particularly hard. Random horny dudes on the internet are putting out loras and quantized models in days against the open source image models.
So what's worth 380 billion exactly? The brand?
These valuations just look really off. Not by one order of magnitude, but more like by 4 orders of magnitude. Like 380 million might be a reasonable valuation, but not billion.
What I also don't get is that it's pretty obvious to me that the Europeans should all be spinning up their own, not necessarily massive, data centers and throwing a few billion at some guys in Cambridge or Stockholm or London or Berlin to make their own AI models.
Only the French have done it.
But instead the rest seem to be trying to court Anthropic or OpenAI to build data centers. Which is just stupid politics given what's happening in the world right now.
The technical task is not the business task... unless the task really is a commodity.
Coding facebook isn't rocket surgery either. Neither is Visa, Salesforce or many other tech-centric companies. Replicating their business model is.
Those are locked in by network effects. Path dependencies and suchlike can play a role. But... the upshot is that anthropic, open Ai and whatnot have the model people are using for work.
A government sponsored model isn't a bad thing to have, but I thing it's unlikely (but possible) that it will also be the product people want to use or the business that succeeds.
Whatever it is that leads to a $30bn run rate, growing >200%. Right now it's having the better model and being able to show how to use it in specific verticals.
But I suspect in the long run only platforms have high margins (and they will need margins not just revenues to justify their valuation). Are they becoming platforms? Google seems to think (or fear) that they might.
Not directly related to the valuation question you asked, but for Google there's a lot of value in getting as much Anthropic workload to run on their hardware as possible. The value comes from getting the insights and learnings of running these workloads, especially when they run on custom Google hardware. That hardware will get better as a result and increase the likelihood that Google has world class AI hardware in the future.
I can't say with any confidence that the $40B is a reasonable amount to pay for that value, but it doesn't seem unreasonable over a multi year time horizon given the stakes.
>These valuations just look really off. Not by one order of magnitude, but more like by 4 orders of magnitude. Like 380 million might be a reasonable valuation, but not billion.
The motivation behind what you've classified 'good' is bad, you've just twisted the symptoms to appear good, while the underlying motivation and root cause is bad.
Unfortunately the parent is suffering from a complete lack of self confidence, and even telling them to go to a therapist won't help as they never will.
Seen it IRL, even if they book an appointment, they'll convince themselves there's some good reason not to go. The two people I've met with it both somehow convinced themselves that therapy didn't work without ever trying it. To the point of lecturing me, who has been to therapy and found it helped immensely, at how useless it is.
It really seems to be a nefarious affliction.
Reading that list above is like watching a car crash in slow motion. You desperately want to help them, they could have so much of a better life if they just believed in themselves even a little bit, but they won't listen to you.
One of my friends I once asked 'Do you want me to push you any morez or is it better if we just talk about other things?'. They dejectedly admitted that they found being pushed depressing and preferred if we didn't talk about it any more.
One of the funniest, insightful people I know, with a great talent, is working a warehouse job and we meet and talk and have a great time but we now talk about anything but his failure to launch.
Ditto for a CERN physicist that now is a part-time tutor for high schoolers living at home with his parents.
Simple examples are interceptors and error handling.
Fetch is one of those things I keep trying to use, but then sorely regret doing so because it's a bit rubbish.
You're probably reinventing axios functionality, badly, in your code.
It's especially useful when you want consistent behaviour across a large codebase, say you want to detect 401s from your API and redirect to a login page. But you don't want to write that on every page.
Now you can do monkey patching shenanigans, or make your own version of fetch like myCompanyFetch and enforce everyone uses it in your linter, or some other rubbish solution.
Or you can just use axios and an interceptor. Clean, elegant.
And every project gets to a size where you need that functionality, or it was a toy and who cares what you use.
ourFetch is more likely to be buggy, unmaintained, undocumented and nobody knows it well because the guy who wrote it left the org 2 years ago and so you have to waste time reading and maintaining it yourself.
Axios is something where you get most of that work done for you by the community for free, and a lot of people know it. As long as you don’t get pwned due to it. Oh and you will actually find community packages that integrate with it, vs ourFetch, which again, nobody knows or even cares that it exists.
Applies to web frameworks, databases and other types of software and dependencies - if you work with brilliant people, you might succeed rolling your own, but for most people taking something battle tested, something off the shelf is a pretty sane way to go about it.
In this case it’s a relatively small dependency so it’s not the end of the world, but it’s the exact same principle.
I think the ideal model would be being able to depend on upstream code, but being able to review ALL of the actual code changes when pulling in new dependency versions (with a nice UI) and being able to vendor things and branch off with a single command whenever you need it, so you don't have to maintain it yourself by default but it's trivial when you want to.
It's actually surprising that in regards to front end development the whole shadcn approach hasn't gotten more popular. Or anywhere else for that matter, focusing on making code way more easy to maintain, to compile/deploy, with less complexity along the way.
It's the difference between using a SQL library and some person on your team writing their own SQL library and everyone having to use it. There's a vast gulf between the two, professionally speaking.
People dissing axios probably suffer from other NIH problems too.
You can read the article and see it's not a tech debt trade off but someone not doing a back of the envelope guesstimate about how much DyanmoDB would cost to run their payments system on it.
We can all see the vast gulf between paid + open AI in image and video, it's really visible. Compare Grok to wan or LTX or whatever and the difference is vast. There is no debate that those sort of models are 3 or 4 generations behind, because you can't argue with your eyes.
But DIYers like you claim that text LLMs are up to scratch with the frontier models?
Again, I simply don't believe you. I can't be bothered to download like however many GB it is to find out, because the result is going to be completely underwhelming and going back to 2023.
And worse, when these 'open' models do start getting good, what makes you think these companies will carry on open sourcing their models?
At the moment they're trying to stay relevant, get investment. When these models do start getting good, they won't give away the weights, they'll sell them.
They're not actually open.
And then in a year or two your 'open' model will be horrifically out-of-date with completely out of date knowledge, because you can't add to the knowledge of the model, it's stuck at whatever date the data it was trained on finished.
So in a year or two, those models will be worthless. That's why Ali, Meta, etc. are giving them away.
To save others a click, though the article is worth reading.
He also mentions no subagents by default in pi as well.
reply