The Problem With Large Language Models

The problem with like the very nature of the LLM's is A, they're creative, B, they're sort of changing slightly all the time. So you can't write a test suite that gives the same result every time. It's just simply not going to unless you have, you know, temperature set to zero and the model for some reason doesn't change. And if you then are forced to deploy maybe a degraded version or a more aligned version into your app and degrade the experience, like you have no control. That level of control is going to push and steer people further and further away from these large language models.

Transcript

Play full episode

Transcript

Episode notes

The AI-powered Podcast Player

Save insights by tapping your headphones, chat with episodes, discover the best highlights - and more!

Get the app