Bing: “I Will Not Harm You Unless You Harm Me First”

---

tags: #literature #inbox uid: 202302201659 —

The very fact that they’re so good at writing fictional stories and poems and jokes should give us pause: how can they tell the difference between facts and fiction, especially when they’re so good at making up fiction?

A search engine that summarizes results is a really useful thing. But a search engine that adds some imaginary numbers for a company’s financial results is not. Especially if it then simulates an existential crisis when you ask it a basic question about how it works. Source: https://read.readwise.io/read/01gschkgvemwc7t5fzhsq0qsah

I’d love to hear from expert AI researchers on this. My hunch as an enthusiastic amateur is that a language model on its own is not enough to build a reliable AI-assisted search engine.

I think there’s another set of models needed here—models that have real understanding of how facts fit together, and that can confidently tell the difference between facts and fiction. Source: https://read.readwise.io/read/01gschm0vvpnwg4ycg4febwc94

---

---

---