Jonesing For The Next Disruptor
9 points by brtkdotse
9 points by brtkdotse
I can't recall that large groups of programmers didn't immediately understand how to leverage the Net to increase productivity.
Memory is an artist. It wasn’t until smart phones that we went from the internet as a convenient way to run what used to be called mail order to the pervasive computing phenomenon we experience today.
The article rolls together an extraordinary collection of ideas that don’t even sit together comfortably. There’s modern capitalism bad (fair enough) and also the idea that for technology to represent real value it has to result in profitable companies. It’s a good thing when investment results in a technology that creates social value without any particular company or person being able to capture a huge slice and create a new billionaire.
If we believe that new software creates social value (a proposition that I think is itself hitting diminishing returns), then LLMs clearly create value by allowing engineers to create better software more quickly. Or not, if you focus on the worst reports. But that’s still completely separate from whether or not there are private fortunes being made in AI vendors.
Unlike the dot-com boom there's no initial proof of success, no profitable ISPs or EBays to spark the flames of hype.
I was not around for the dot-com boom, but I don't think this statement necessarily holds as these AI companies already have tens (hundreds?) of billions in current revenue. ChatGPT also has 800M MAUs. Now all of this still might not be enough for the current level of spending to not cause a bubble with bad outcomes, but I wouldn't say there's no initial proof of success.
Seconding this – the proof of success is already there, it's just not evenly distributed.
What I've seen around me, here in Estonia:
Nobody I've spoken to is under any illusions about the quality of the output – they all know that in the long run they need to hire an actual artist for their board game art, or an engineer to put their software on a solid foundation.
What's new with AI is that they get to try first, build some momentum in the short term, and invest long-term into their projeect once they know that there is a long-term perspective in the first place.
At the time of writing, even Anthropic themselves are hiring developers, rather than cutting down and cashing in on that self-reported 90% AI generated code.
This only makes sense assuming that the scope of the work remains fixed, in which case productivity gains would indeed result in a smaller amount of manpower required.