Transparency works at several layers. At the interaction layer, it means telling someone they are talking to an AI. At the system layer, it means explaining what a tool is for and where it fails. At the model layer, it means documenting how a model was trained and on what data. Explaining an individual decision is a related but narrower idea; see explainability.
Hiring is a clear example. In episode 58 at 21:49, Greenhouse CEO Daniel Chait says companies “haven’t always been transparent” about whether a candidate will be interviewed by a person or an AI, and argues that candidates warm to AI interviews when the process is explained up front. Galileo was the season’s presenting sponsor (sponsor read). At the model layer, IBM’s Maryam Ashoori says in episode 17 at 8:43 that IBM trained its Granite models from scratch and has “been very transparent in terms of how the models are trained.” That is her account of IBM’s practice; episode 17 was sponsored by Galileo, whose CTO co-hosted.
Regulation now requires some of this. Article 50 of the EU AI Act requires that people be informed when they interact with an AI system and that synthetic content be machine-detectable, which is where AI watermarking and content provenance come in. Editorially, transparency is a precondition for algorithmic accountability, not a substitute: disclosing that AI is used does not by itself make the use fair or correct.
Hear it from the guest
“Companies haven't always been transparent, surprisingly, about whether you're gonna speak to an interview with a person or whether you're gonna speak to an AI”
“We've been very transparent in terms of how the models are trained”
Quotes lightly edited to remove filler words.
Sources
- OECD.AI: Transparency and explainability (OECD AI Principle 1.3) — Asks AI actors to provide meaningful information about AI systems, including making people aware of their interactions with AI; checked October 24, 2026.
- EU AI Act Explorer: Article 50, Transparency Obligations for Providers and Deployers of Certain AI Systems — Requires people to be informed when they interact with an AI system and AI-generated content to be marked in a machine-readable format; applies from 2 August 2026; checked October 24, 2026.
