Why Tejal Patwardhan stopped underestimating the models - Episode 21

单集文稿 ...

• 00:00

  Hello, I'm Andrew Main and welcome to the OpenAI podcast.

• 00:03

  On today's episode, we're talking to the research lead, Tejal Pat Warden,

• 00:07

  about the need to build frontier e-valves as old benchmarks get saturated.

• 00:11

  Generally bad, benchmarking is bad.

• 00:13

  How can we make these models useful for people in their real work?

• 00:16

  We were really nervous because we were like, this human baseline is kind of hard.

• 00:19

  We don't know if the model is going to beat it, but we should never underestimate the model.

• 00:24

  Tejal, I have a question.

• 00:26

  How did you end up where you were?

• 00:27

  What brought you into OpenAI?

• 00:29

  Oh, I thought we weren't going to start with this.

• 00:31

  Tagel, I have a question for you.

• 00:32

  What would you like to start with?

• 00:33

  Can we start with, like, tell us, like, what you did when you started OpenAI?

• 00:37

  And you can, like, work backwards.

• 00:38

  Don't you want to talk about your early days?

• 00:40

  No.

• 00:40

  I grew up at OpenAI.

• 00:45

  Tell me a bit about your journey here working inside artificial intelligence, inside OpenAI.

• 00:50

  So I joined OpenAI in fall 23, and it was right after Chatchapiti had come out, GPG4 was out,

• 全部文稿