Menu
Sign In Search Podcasts Libraries Charts People & Topics Add Podcast API Blog Pricing

Dwarkesh Patel

πŸ‘€ Speaker
15656 total appearances
Voice ID

Voice Profile Active

This person's voice can be automatically recognized across podcast episodes using AI voice matching.

Voice samples: 1
Confidence: Medium

Appearances Over Time

Podcast Appearances

Dwarkesh Podcast
Andrej Karpathy β€” AGI is still a decade away

So you've talked about how you were at Tesla leading self-driving from 2017 to 2022.

Dwarkesh Podcast
Andrej Karpathy β€” AGI is still a decade away

And then you firsthand saw this progress from, we went from cool demos to now thousands of cars out there actually autonomously doing drives.

Dwarkesh Podcast
Andrej Karpathy β€” AGI is still a decade away

Why did that take a decade?

Dwarkesh Podcast
Andrej Karpathy β€” AGI is still a decade away

Like what was happening through that time?

Dwarkesh Podcast
Andrej Karpathy β€” AGI is still a decade away

That's very interesting to hear you say that the sort of safety guarantees you need from software are actually not dissimilar to self-driving because what people will often say is that self-driving took so long because the cost of failure is so high.

Dwarkesh Podcast
Andrej Karpathy β€” AGI is still a decade away

Like a human makes a mistake on average every 400,000 miles or every seven years.

Dwarkesh Podcast
Andrej Karpathy β€” AGI is still a decade away

And if you had to release a coding agent that couldn't make a mistake for at least seven years, it would be much harder to deploy.

Dwarkesh Podcast
Andrej Karpathy β€” AGI is still a decade away

But I guess your point is that if you made a catastrophic coding mistake, like breaking some important system every seven years.

Dwarkesh Podcast
Andrej Karpathy β€” AGI is still a decade away

And in fact, in terms of sort of wall clock time, it would be much less than seven years because you're like constantly outputting code like that, right?

Dwarkesh Podcast
Andrej Karpathy β€” AGI is still a decade away

So it's like per tokens, or in terms of tokens, it would be seven years.

Dwarkesh Podcast
Andrej Karpathy β€” AGI is still a decade away

But in terms of wall clock time, it would be pretty close.

Dwarkesh Podcast
Andrej Karpathy β€” AGI is still a decade away

There's another objection people make to that analogy, which is that with self-driving, what took a big fraction of that time was solving the problem of having basic perception that's robust and building representations and having a model that has some common sense so it can generalize to when it sees something that's slightly out of distribution.

Dwarkesh Podcast
Andrej Karpathy β€” AGI is still a decade away

If somebody's waving down the road this way, you don't need to train for it.

Dwarkesh Podcast
Andrej Karpathy β€” AGI is still a decade away

The thing will...

Dwarkesh Podcast
Andrej Karpathy β€” AGI is still a decade away

have some understanding of how to respond to something like that.

Dwarkesh Podcast
Andrej Karpathy β€” AGI is still a decade away

And these are things we're getting for free with LLMs or VLMs today.

Dwarkesh Podcast
Andrej Karpathy β€” AGI is still a decade away

So we don't have to solve these very basic representation problems.

Dwarkesh Podcast
Andrej Karpathy β€” AGI is still a decade away

And so now deploying AIs across different domains will sort of be like deploying a self-driving car with current models to a different city, which is hard, but not like a 10-year-long task.

Dwarkesh Podcast
Andrej Karpathy β€” AGI is still a decade away

You let self-driving for five years at Tesla.

Dwarkesh Podcast
Andrej Karpathy β€” AGI is still a decade away

Because one, the start is at 1980, not 10 years ago.