Reflections on the path toward AGI
that benefits all of humanity.

"We set out to do something crazy,
unlikely, and unprecedented."

Reinforcement Learning
from Human Preferences
showed a path to aligning
Al with human values.

The unsupervised
sentiment neuron learned
semantics, not just syntax.

Pushed reinforcement
learning to new levels with
Dota 1v1 results.

We knew we needed to scale up each of these
results with massive computational power.

"All of a sudden, AGl was no longer a crazy thing to consider."

This technology has gotten
integrated into the world at a
scale and speed that no
technology ever has before.

"These last three years
have been extremely
intense and full of stress
and heavy responsibility."

"This was quite controversial at the time, but I think it has been one of
our best decisions ever and become the industry standard."

We successfully put early versions of the technology into the world, so that
people can form intuitions and society and the technology can co-evolve.

"I have never felt more optimistic about
our research and product roadmaps, and
overall line of sight towards our mission."

In ten more years, I believe
we are almost
certain to build
superintelligence.

In some sense, daily life and the things we care most about will change very little.

"The people of 2035 will be capable of doing things that
I just don't think we can easily imagine right now."

Healthcare Briefing