AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
6min chapter
Scott Aaronson: Against AI Doomerism
The Gradient: Perspectives on AI
Watermarking the Outputs of GPT
The way that GPT works is inherently probabilistic, right? It's constantly taking as input like the sequence of previous words or what are called tokens. And then it outputs a probability distribution over possibilities for the next token. In some cases, almost all of the probability will be concentrated on a single choice. The idea with watermarking is that you can pick the next token pseudo-randomly. This means that to a normal user, the output will be indistinguishable from normal GPT output.
00:00
Transcript
Episode notes
Get the Snipd
podcast app
Unlock the knowledge in podcasts with the podcast player of the future.
AI-powered
Discover
highlights
Listen to the best highlights from the podcasts you love and dive into the full episode
Save any
moment
Hear something you like? Tap your headphones to save it with AI-generated key takeaways
Share
& Export
Send highlights to Twitter, WhatsApp or export them to Notion, Readwise & more
AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
Discover
highlights
Listen to the best highlights from the podcasts you love and dive into the full episode