🧠 MindStream

AI capabilities have advanced from high school math to solving unsolved research problems within one year

“A year ago AI was just able to do high school math... By last summer, uh it was competing at the hardest mathematics competitions we had in the in the world... last week uh there was a new thing called first proof... I believe our latest AI got seven of those problems right”

AGI is very close at hand, and given faster takeoff expectations, superintelligence will follow shortly after.

“If you had asked most people six years ago what would you think if we had systems that could do new research on their own... I think super intelligence is not that far off.”

Significant professional categories, including manual software coding, face complete obsolescence requiring total workforce adaptation.

“A lot of professions will almost go away... writing, you know, C++ code by hand, that's over... big categories of jobs AI is just going to completely obsolete.”

Delivering enough compute capacity for everyone to have thousands of GPUs is physically impossible in the near future

“if you go multiply that out and say there's 8 billion people in the world, we have no way to deliver 8 trillion GPUs anytime soon... We're not going to do that to be clear.”

Human-centric roles involving care and fine art retain value because people prefer human involvement despite AI efficiency.

“For fine art, the price of AI generated art is a zero and the price of human generated graphic art has continued to go up... I really cared about the nurse that was taking care of me.”