The Philosopher Inside Google DeepMind Who Asks: What Is AI, Really?
Iason Gabriel has spent seven years at Google thinking through what artificial intelligence actually is and what it might do to society. As business pressure mounts, his job is getting harder.
Key points
- Iason Gabriel joined Google in 2017, making him one of the longest-serving AI ethicists at any major technology company.
- Gabriel works at Google DeepMind, the research division that builds some of the world's most advanced AI systems.
- Commercial pressure and the US-China race over AI supremacy are making ethical review harder, not easier.
- Nobody yet fully understands what a large language model, the technology behind chatbots like ChatGPT and Gemini, actually is.
Iason Gabriel is a philosopher working inside one of the most powerful AI labs on earth. His job, broadly, is to ask uncomfortable questions before a product ships rather than after.
Gabriel joined Google in 2017, when the phrase "AI ethics" still sounded academic. The stakes look very different now.
He sits within Google DeepMind, the London-based research division Google formed in 2023 by merging its two main AI research arms. DeepMind built AlphaFold, the system that cracked the structure of nearly every known protein, and it also works on Gemini, Google's family of large language models, which power conversational chatbots.
What does an AI ethicist actually change?
Honestly, that's contested. Gabriel's work involves anticipating harm before it happens: thinking through how a system could be misused, who it might hurt, what values it should reflect. Anticipating harm and preventing it are two different things.
The pressure is real. Governments in the US and China are treating AI as a strategic asset, pumping billions into development and urging speed. Companies face investor expectations and competitive threats from rivals releasing new models every few months. A philosopher saying "wait, let's think about this" is swimming against a strong current.
Gabriel told The Guardian that one question keeps surfacing no matter how far the technology advances. "There's this deep mystery of what, actually, is this thing?" Nobody, including the engineers who build these systems, fully understands why large language models behave the way they do. They process enormous amounts of text and generate responses that can seem thoughtful or even creative, yet the internal mechanism remains, in important ways, opaque.
That opacity matters for ethics. If you cannot explain why a system made a decision, holding anyone accountable when something goes wrong becomes very hard.
AI2Day first covered Gabriel's work on 17 July 2026, and that piece sits within our 44 AI ethics stories published since 13 July covering exactly this tension between institutional caution and commercial speed. Two weeks earlier, we reported that DeepMind's own chief wanted a global AI watchdog in place before development got away from everyone. Gabriel's daily reality is what that proposal is trying to solve.
Should you worry?
For most people the practical implication is direct: the AI tools you use at work, in school or in a clinic are built by companies where people like Gabriel are fighting for a seat at the table. Whether they win that argument shapes what those tools do.
Gabriel's presence doesn't guarantee good outcomes. His absence would likely make them worse. Seven years in, the real question is whether ethics functions as a genuine brake or as reassuring decoration on a machine moving too fast to stop.
