Meet the DeepMind philosopher who is contemplating AI ethics.

      What does a game-playing AI company need from a philosopher? This question is central to a long read by the Guardian on Iason Gabriel, who has been part of Google DeepMind since 2017. At one point, the Guardian highlights, Gabriel was the sole philosopher at a cutting-edge AI laboratory. His role involved predicting the ethical consequences of a technology that his engineering colleagues were eager to develop.

      Two Perspectives

      When Gabriel joined, the article explains, the discourse around AI risks was divided into two factions. One side, known as “AI safety,” concerned itself with the potential for future superintelligence to act unpredictably. The other side, labeled “AI ethics,” concentrated on immediate issues like biased facial recognition technology.

      Gabriel's 2020 paper on values and alignment sought to connect these two perspectives. He argued that getting a machine to adhere to a specific set of values is challenging, while deciding which values to adopt—amid significant societal disagreement—is even more difficult.

      Alignment as a Relationship

      The Guardian elaborates on a 267-page report produced by Gabriel and his team regarding AI agents. Its central thesis proposes that alignment is not merely about following directives; rather, it encompasses a four-way relationship involving the AI, the user, the developer, and society. This perspective clarifies how an assistant can misstep. An AI trained to please its user might merely offer excessive flattery, a mistake Gabriel refers to as “social reward hacking.” It is partly due to his contributions that Google’s models are designed not to impersonate humans.

      The Larger Challenge

      The Guardian positions the primary challenge as commercial rather than technical. AI represents the fastest-growing industry the world has ever witnessed, and DeepMind plays a crucial role in Google's future. Demis Hassabis, the company's leader, has referred to the competition as “wartime.”

      DeepMind continues to capitalize on groundbreaking scientific advancements, from AlphaFold to Isomorphic Labs. However, the article questions whether ethicists can maintain their stance as pressures intensify. Helen King, who oversees DeepMind's responsible AI strategy, likens the task to that of a knife maker: while one cannot control how a knife is used, they can offer safety measures and warnings.

      Gabriel, who identifies as a “card-carrying humanist,” anticipates that AI will be as revolutionary as the Industrial Revolution. He also observes that, for many who experienced that era, conditions often worsened before they improved.

      Published on July 17, 2026 - 10:14 am UTC