How is generative AI evolving, and what can we do about it? While at NDC in Oslo, Richard chatted with Alison Cossette about her work as a data scientist before the ChatGPT explosion in November 2022 and what life has been like since the LLM came to town. Alison talks about the rigor of building AI models using generative AI before ChatGPT and how many of those efforts have diminished when confronted with a friendly, confident language model. Eventually, this rigor will be needed - as the dangers of not managing language models cause problems, and the need for rigor will re-appear. Alison describes steps you can take today to understand how the LLMs you are using are trained and how they are tested. Generative AI is evolving, and you can be part of making it better!

Links

• GitHub Copilot
• Fairly Trained

Recorded June 12, 2024