DeepSeek, every mention

13 scenes (2025), the whole family · ← back to DeepSeek

mentionsepisodesper episode tap a year for its mentions

Verbatim, from the transcripts: the passages where DeepSeek comes up

loading…

▶ 0:00 Benedict Evans It seems to me right now you could do like a double-blind test of the same prompt given to Grok, Claude, Gemini, Mistral, Deep Seek.
▶ 27:42 Benedict Evans Even then was obviously yes, of course, which is what happened, which Deepsea kind of demonstrated. 2 times in the scene

▶ 1:06:28 Garry Tan And so in the same way, I think that whether it's OpenAI or Anthropic or Meta's Lab or Deep Seek or some other lab that wins, like, I think that we're going to have a bunch of different labs and they're going to serve a certain role, like…
▶ 1:18:38 Garry Tan Yeah, and that was like Deep Seek R-one just got released this morning.
▶ 1:19:31 Garry Tan I mean, that's another weird thing of AI regulation, you know, um, I think Biden, for instance, um, his last EO was sort of this export ban, and, uh, DeepSeek is a Chinese company, uh, releasing these models open source, and I believe that…
▶ 1:48:22 Garry Tan Um, and then as the models get better, because you have, you know, what we call the golden evals, basically, you just run the golden evals against, you know, the new, the newest model, like, uh, you know, four O comes out, cloud 3.5 comes…
▶ 1:50:16 Garry Tan Um, and it, I saw it up close and personal, uh, this morning looking at some blog posts about DeepSeek R-one, which is, uh, DeepSeek's, uh, reasoning model.
▶ 1:50:16 Garry Tan Um, and it, I saw it up close and personal, uh, this morning looking at some blog posts about DeepSeek R-one, which is, uh, DeepSeek's, uh, reasoning model. 4 times in the scene

▶ 6:34 Josh Wolfe And you're already starting to see this where there was one element of like an instantiation of R one communicating with another in R one, but the language that they were commuting was not English or Chinese or even traditional code.
▶ 13:06 Josh Wolfe And this was some of the debate that I think happened with the release of DeepSeek where you had Satya basically talking about Jevin's paradox, that any one thing might get more efficient, but the result is not that you have less demand.
▶ 21:53 Josh Wolfe The second that DeepSea comes out, it suddenly accelerated the internal strategic decisions from OpenAI of when are we gonna release models.
▶ 32:52 Josh Wolfe I, I, I definitely see that models can improve, because you can even argue, like, DeepSeq's R-one is a model that was improving upon outputs from ChatGPT and, and so on.
▶ 49:24 Josh Wolfe Well, you saw this a little bit with just how quickly the market reacted with a single largest one-day loss of five, six, seven hundred billion dollars with NVIDIA, just because of the fear over DeepSeek, and the fear was a cascading, you…

Turn any episode into a week of clips.