Non-determinism means that each AI generation can produce slightly different results, such as different wording of a sentence or different pixels in a generated image. In highly regulated industries this is particularly challenging as AI models must be explainable, and organizations must be able to prove their outputs are correct. Nobody wants “hallucination” in a banking or payments transaction. Join Googlers Aja Hammerly and Jason Davenport as they dive into non-determinism in AI, learn how it affects AI projects, and what developers need to know when working with generative AI models.
Chapters:
0:00 – Intro
0:15 – What does non-determinism mean for developers?
0:43 – How to handle non-determinism in AI?
1:25 – How developers can determine if they need deterministic responses
1:54 – How can devs ensure agent responses are reasonable with non-determinism?
2:25 – How to add an evaluations to every step of the flow
4:25 – What other options do we have if the evaluator for an AI agents behavior isn’t up to standard?
5:03 – Debugging LLMs
6:16 – Recap
Watch more Real Terms for AI → https://goo.gle/AIwordsExplained
Subscribe to Google Cloud Tech → https://goo.gle/GoogleCloudTech
#GoogleCloud #GenerativeAI
Speakers: Aja Hammerly, Jason Davenport
Products Mentioned: Gemini, Gemma,
