Comment by bccdee
1 year ago
> The valuations seem to be primarily based on the R&D progress
There hasn't been much R&D progress, though. Sure, as another commenter pointed out, context lengths have gotten longer and chat models can interpret images now, but the industry figureheads have been pushing agents, and we're not much closer to those than we were two years ago when GPT-4 came out. Current models simply are not consistent enough to do the kind of agentic stuff that AI valuations are predicated upon, nor is there any sign that a significantly smarter GPT-5 is just around the corner. Multi-modal chat is cute, but OpenAI is burning money. They're all burning money, and they don't have a product. They imply and imply that there's something big on the horizon, but it's been years, and there just isn't a killer app yet. Their platform isn't good enough, and it's not improving in the ways it would need to in order for Godot to arrive and for agents to be feasible.
Recent results are showing exponential improvement in reasoning and dramatic decreases in the time and cost to train models. O3 now ranks 50th on code forces according to openai staff. Are you aware of all of this and still say R&D hasn’t progressed?
You can invest in building bigger and more complicated pipe structures, but until you show the field that is supposed to be irrigated, you can't say you're disrupting farming business.
In the context of Moore's law exponential growth was measured in the number of transistors per integrated circuit. This seems vigorous and straightforward.
With AI the improvements have certainly been impressive but it isn't straightforward how you can define "reasoning" to measure whether or not the reasoning is exponentially "improving".
Again: Those are incremental improvements. The valuations are based on the promise that agents are just around the corner, and we just haven't seen the kind of categorical shift in intelligence that agents would require.
I think you can simultaneously think that there is some real value being made with LLMs and also look at OpenAI losing $5B a year or thereabouts and really wonder how they're not going to run out of money.
That said, I'm learning a new sdk and I've moved 500-1k searches a month from kagi and google to llms.
By this I mean it’s a bet on what R&D might yield, current progress being some kind of signal. No one has certainty here. It’s an emergent technology and no one knows for certain how far it can be pushed.