High Signal Podcasts Evidence ledger
Method
Browse
← Back to evidence

Evidence receipt / uncertainty

Published · transcript-backed

Lenny Rachitsky: uncertainty

10 Oct 2024 Lenny's Podcast Behind the product: NotebookLM | Raiza Martin (Senior Product Manager, AI @ Google Labs)

“To me it feels very similar to the ChatGPT moment where the technology already existed, they had that same GPT model out for I don't know how long, but just that new medium and new way of interacting with it changed the way people imagine.”

— Lenny Rachitsky

Source trail

Everything needed to verify it.

Speaker
Lenny Rachitsky
Attribution
Verified speaker
Claim type
uncertainty
Recorded
10 Oct 2024
Publisher
Lenny's Podcast

Transcript context

…So this is really funny. I talk about this quite a lot, especially when I'm talking to other PMs. I feel like the way that I've always built products is I start from a problem and I think about the solutions to solve it, solve it for people in a meaningful way. And in Labs in particular, we start with the technology. And it's actually a very interesting place to start where you're like, okay, what are the practical applications to this? How do I go about finding the answer? And I think we've seen different approaches to this where you just put the tool out there, you study how people are using it, and I think that's a fine approach, but we try to have a little bit more of a hypothesis about what is the shape of this thing to actually get the most learning out of it. So for the audio feature in particular, the nugget there was really, we had this ability for you to interact with text, but people were like, "The output is still always text." And I'm actually a big fan of voice modality in general, so I use a lot of voice input. And voice output, I noticed when I was experimenting with it early on, it changed a lot of things for me. It's like, oh, it changes the way that I interact with the technology, it changes the way I feel about the technology. It sort of affects the way that I'm even thinking in real-time throughout this process. And so we thought about what is a good way to introduce this to people where they could easily get the value of it and have a little bit of fun? So I think for me it's fun is such a big part of it too, which is like how do we make this cool? And so hopefully we've done that here. To me it feels very similar to the ChatGPT moment where the technology already existed, they had that same GPT model out for I don't know how long, but just that new medium and new way of interacting with it changed the way people imagine. Just they saw the power immediately. And I feel like this is such a good example of that happening again where just the tech was there and it's just this medium that you developed that really inspired people to like, "Oh wow, I had no idea AI was getting this good and LMs were getting this good." Any thoughts on that? Yeah, I think that totally strikes at the heart of what we're trying to do as well, which is a lot of technology I feel like you have to shape it and bring it closer to people. And I think it's such an interesting thing to iterate on and be like, "What is the shape?" Soon you'll get there. If you keep going at it, you'll eventually, I think, land on something that when people look at it, they're like, "Wow, I get it." And that's always what we're hunting for. But I think to your point, technology has been there. I mean it feels so crazy to say that because even with the advent of the chatbots of LLMs as we know them today, and then to introduce this sort of new technology on top of it, we've only been doing this for two years, but already we're exploring so many different ways to use it.…

Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.

Search evidence