Depends on what you mean by login wall. They are, technically, behind a login/auth wall as you can only make 60 requests to their API per hour. Scrapping these days is extremely challenging as defenses do in-depth checks on your browser to make sure you are a "real" person.
People live in very stochastic and volatile environments and they manage that in ways no LLMs currently ever can. (ie: imagine sending an LLM all the data - sensory/auditory/etc… - that a human receive)
People’s job is to partially reign in this volatile environment by creating processes with stable output.
The problem is that with text/code, judgement is hard. Here is what it looks like for physical activity: https://www.youtube.com/shorts/lK7TjujKQLw It's hard to see how that it's not useful at best and could be a disaster for any unsupervised use.
ISIS-style soldiers usually have light-weaponry because they need to be mobile. Having heavy artillery or bombers will make them an easy target for an organized army which they are very not equipped to fight. Their advantage is in there ability to hit in random unprotected areas with little damage but to do it constantly and unpredictably.
It'll find a non-existent bug - fix it - figure out it broke a previously working thing - try to fix again - etc..
The "keep improving" the code base prompt have been tried and it never works. The LLM has no consciousness of where to stop and where to draw the lines of reasonableness.
I talked with a friend on a different field (academic) and he had to re-review all things written by AI. Basically, he used AI to read/summarize/find stuff in large academic papers but realized later that many times AI makes glaring mistakes that on a first read pass the smell test.
I am now in the process of fixing code I wrote using AI. I have come to the realization that AI can't really write software and I am annoyed that it took me that long (months) to realize that.
Photopea predates LLM coding by a lot. The fact that it was made by single developer is not particular, so was Figma.
The reality is, Photopea is unique because no one with LLMs can easily produce something similar. Also Photopea does not compare to Adobe offering similar to how Figma does. Not even close.
It's simple: AI can't really write software. AI is good in generating garbage code averaged from millions of repos.
Here is an example of software that can scale with a single person assuming AI can write software: a web-based alternative to Photoshop built on wasm. Figma is analogous to that and it's worth $10Bn.
Is everyone project so simple that it can fit in these "vp check" / "vp dev" commands? Like even for my amateurish web app, I have a custom web server with a self-signed certificate with an "/etc/hosts" domain; and for checks I need to do custom checks for GraphQL and a couple of cloned NPM packages.
Is that really that interesting? Imagine giving a bunch of tech nerds some funds and freedom to do whatever. The end result will be software/tech (whether the product is useful is actually another story).
The OP is trying to summarize a 1.4Bn country failure to achieve jet engine in: if you do these couple tricks; or because of these two things.
If you're a small country, sure, that kind of strategy might make sense. Pick your battles, run the country like a startup (i.e. bet on one or two industries). China's strategy is the opposite of that: just make production costs low across the board (transport, energy, housing, etc.) and let everything else follow. With 1.4 billion people, something somewhere is bound to pop off.
People are reading way too much into the 5-year plans. It basically boils down to "do science across the board, but lean a bit more into these areas."
> All the money and talent in the world can't replicate real users generating real data that you can use to improve
Basically, China excellence in EVs and Solar was driven by the market being new. It's hard (almost impossible?) to outrank an incumbent very entrenched in a big market. You need a paradigm shift (ie: iphone vs nokia) to make the change.
The $20/month model is exclusionary as you are excluding people who are not interested in regularly reading a single newspaper. Pay-per-page is more reasonable. I have rarely paid for news subscription (partly because I find news here which comes from many sources). Obviously, I am not going to pay hundreds in subscriptions, so I am essentially paying $0.
With this, I can see myself paying $20-30/month (less than my coffee spend). That's money that was previously unlocked. Also, being able to pay some random writer/journalist outside the mega-news-corps, has a special feel to it and this gives me the option to do that.
You are missing that the product is the hype cycle around AI and that's worth Trillions of $ (Trillions with a T). Why build a PDF parser that generate text when you can BS in a podcast and get paid.
This discussion was about measures, goals and incentives. Follow the incentives.
I mean, if the execution happen on the VM then the problem is trust on the programs and then you can't trust any program by that logic? That or you think AI-companies software is serious slop.
blog: https://omarabid.com
building: https://codeinput.com - tools for developers
=====
The world is now a closed club of aging incumbents, circled by middle powers, developing countries, and failing states.