"AI" is simply a field of study. There is no true bar for "AI" that GPTs fail. Because there is no true bar for AI. a symbolic AI system is as much AI as the most advanced LLM or world model or whatever.
I don't see much of an increase in those charts.
I would call that flat, not linear. Linear implies a non exponential increase. This looks flat.
Very interesting! I didn't realize there was this historical division between fortran and c. I thought c was just "better" because it came later.
From what I've read monthsl's worth of Claude queries through github copilot is estimated to have the same carbon footprint as driving 12 miles.
I do not care about IP law. My greater concern is how this stuff furthers consolidation in the tech industry.
People focus on Claude code because it's a massive improvement over the previous models. The difference between gpt 4.1 and Claude 4 is palpable. (Claude code is really just a particular interface for using Claude 4, though that interface does add some juice beyond just the model the same way that copilot does.)
You make an excellent point in the analogy between c and assembly.
Part of it is I think the current companies are extremely consolidated. Why would they want to make something new? They are only interested in strangling their remaining customers.
People aren't friendly to anti tax rhetoric when it's been used to undermine our current democracy.
Is our current system good? No. But it's getting overly theoretical to go off on taxes in this context.
Syndacalism isn't roving bands of nomads.
Ice is part of the dhs but not part of uscis. Uscis is also part of dhs but they are distinct.
Words have no inherent meaning apart from use. Quibbling over semantics, as if you can finely slice reality, is a waste of time.