Poisoning the generative AI well

Bewitching the bots by accident or by design.

This week, Tristan and Tasia discuss how Google plans to monetize its AI-powered search experience and how it plans to challenge ChatGPT's dominance with its next-generation LLMs. Meanwhile, OpenAI's developing a tool to detect images created by DALL-E 3. We'll also explore how AI "hallucinations" are affecting web search, and how a new data poisoning tool lets artists fight back against AI web scrapers. Join us as generative AI begins to break in chaotic and unpredictable ways.

LISTEN

• Acast, Amazon Music, Apple Podcasts, Google Podcasts, iHeart or Spotify

FOLLOW

• AI Named This Show on Facebook, Instagram, YouTube and X (Twitter)

• Tristan & Tasia

• AI Named This Show podcast on Acast, Amazon Music, Apple Podcasts, Google Podcasts, iHeart or Spotify

FOLLOW-UP

• Google is ready to fill its AI searches with ads

AI NEWS

• Google’s Gemini AI is a Serious Threat to ChatGPT – Here’s Why

• The misfires of AI — what about hallucinations of the chatbots?

• Chatbot Hallucinations Are Poisoning Web Search

• OpenAI is working on a tool to detect DALL-E 3 AI-generated images

• This new data poisoning tool lets artists fight back against generative AI

LLM PRIMERS

• A jargon-free explanation of how AI large language models work

• Generative AI exists because of the transformer

• What Is ChatGPT Doing … and Why Does It Work?