But start-ups are responsible for a big fraction of disruptive technology, so it’s a reasonable bet.
What’s your all-things-considered view for probability that the first transformative AI (defined by your lights) will be developed by a company that, as of December 2020, either a) does not exist or b) has not gone through Series A?
(Don’t take too much time on this question, I just want to see a gut check plus a few sentences if possible).
About 40%. This is including startups that later get acquired, but the parent company would not have been the first to develop transformative AI if the acquisition had not taken place. I think this is probably my modal prediction: the big tech companies are effectively themselves huge VCs, and their infrastructure provides a comparative advantage over a startup trying to do it entirely solo.
I think I put around 40% on it being a company that does already exist, and 20% on “other” (academia, national labs, etc).
Conditioning on transformative AI being developed in the next 20 years my probability for a new company developing it is a lot lower—maybe 20%? So part of this is just me not expecting transformative AI particularly soon, and tech company half-life being plausibly quite short. Google is only 21 years old!
What’s your all-things-considered view for probability that the first transformative AI (defined by your lights) will be developed by a company that, as of December 2020, either a) does not exist or b) has not gone through Series A?
(Don’t take too much time on this question, I just want to see a gut check plus a few sentences if possible).
About 40%. This is including startups that later get acquired, but the parent company would not have been the first to develop transformative AI if the acquisition had not taken place. I think this is probably my modal prediction: the big tech companies are effectively themselves huge VCs, and their infrastructure provides a comparative advantage over a startup trying to do it entirely solo.
I think I put around 40% on it being a company that does already exist, and 20% on “other” (academia, national labs, etc).
Conditioning on transformative AI being developed in the next 20 years my probability for a new company developing it is a lot lower—maybe 20%? So part of this is just me not expecting transformative AI particularly soon, and tech company half-life being plausibly quite short. Google is only 21 years old!
Thanks a lot, really appreciate your thoughts here!