Video

Transcript

Imagine the wildest idea that you’ve ever had and you’re curious about how

it might scale to something that’s hundred a thousand times bigger.

This is a place where you can get to do that.

OpenAI is building the general technologies that can extend the reach

of the human brain and our creativity.

Dall-E really brings high quality creations to anyone.

Our coding models serve as a coding assistant in tools like Co-Pilot.

And GPT3 is a system that has a robust understanding of language.

Our mission is to create highly capable AI technologies and deploy them to the

world, for the benefit of humanity in a way that is safe and beneficial for all.

I think that AI will be a technological revolution on the

scale of the agricultural, the industrial, the computer revolution.

And we have no goal other than the creation and deployment

of safe, beneficial AGI.

We have a non-profit that governs everything, and the ability to block

deployments that would make commercial sense but we think create safety issues.

Our whole structure is around that.

I really like the way OpenAI are trying to build AGI.

Namely by harnessing large amounts of real world data to train powerful

unsupervised models, and then steering those with human alignment to create

agents that can actually do useful things.

OpenAI is truly unique in that sense of being able to

work in a very nimble fashion.

At the same time having all the resources in the world to solve the problems.

That mentality is what made it really possible to ship Dall-E.

At OpenAI we felt very strongly that engineering is an integral part to

the success of any progress in AI.

And that research needs to be infused with engineering and engineering

needs to be infused with research.

In my first month I did a kind of an outrageously large scaling project.

They were like yeah we have a piece of software that runs on

one computer can you make it run on 3000 and you have three weeks?

So that was wild.

I don’t think you need a specific machine learning background to

come here to OpenAI and succeed.

I had just taken one class in college.

People were very willing to help and teach.

And I worked on Co-Pilot, one of the more successful products that are based

on ML that people are using day-to-day.

People are really both friendly and approachable and ambitious.

There’s this shared sense of we’re going to do something very very

big together, we’ve set our sights high, and also we’re collaborating

together to make that happen.

The number of times I’ve been sitting at lunch and you overhear the

other team talking about something and you learn a trick from that.

Just kind of going on the signal of what is everyone else excited about is already

a really strong sense of what do I need to pay attention, to what do I need to learn.

For the systems that we’re developing to have a big impact we have

to figure out how to make them accessible and how to do so safely.

So it’s very important to work with a set of people that have diverse

experiences, because we’re really entering an era where we have to innovate

in almost every aspect of society.

I truly believe that the technology we are creating is going to

have a really profound impact.

Can we create an environment where we have AI and technology

broadly amplifying human wisdom?

Giving humans another resource for connecting with what’s actually important

to them, what actually gives them meaning.

And we need a lot of different kinds of people to make that happen.