A company with no moat. I can train / run a model on my NVIDIA / AMD gpu. Advances in mathematics and in chip speed will eat away at their market share. There's only so much cognition that is needed at some point.
This is why they're so obsessed with regulation. You can go ahead and regulate it in America, but the rest of the world is not so stupid, and matrices are not a sufficient differentiating factor.
implication: Beyond a certain level, there is little productivity gains to be made by higher level cognitive capabilities, and will not provide a competitive barrier to others.
Machine cognition will reach a certain level after which it becomes a commodity and priced by energy use, most likely.
A company with no moat. I can train / run a model on my NVIDIA / AMD gpu. Advances in mathematics and in chip speed will eat away at their market share. There's only so much cognition that is needed at some point.
This is why they're so obsessed with regulation. You can go ahead and regulate it in America, but the rest of the world is not so stupid, and matrices are not a sufficient differentiating factor.
> There's only so much cognition that is needed at some point.
What do you mean by this?
implication: Beyond a certain level, there is little productivity gains to be made by higher level cognitive capabilities, and will not provide a competitive barrier to others.
Machine cognition will reach a certain level after which it becomes a commodity and priced by energy use, most likely.
One that was good enough to replace human workers could be worth a lot. Time will tell I guess.
Lots of people have been paying them quite a lot of money suggesting there is something there beyond what you can do open source.