Philosopher at DeepMind: Preparation for AGI and Machine Consciousness

# DeepMind Hires Philosopher to Prepare for the AGI Era

Google DeepMind has officially hired philosopher Henry Shevlin from the University of Cambridge for a role explicitly titled "Philosopher." This decision underscores the growing importance of humanities disciplines in developing advanced AI systems and reflects a strategic shift in the approach to creating artificial general intelligence (AGI).

Why a Philosopher in an AI Lab?

Shevlin's appointment isn't a PR stunt; it's part of a systematic effort to integrate ethical, cognitive, and ontological questions into the very architecture of future AI. His mandate covers three key areas: machine consciousness, human-AI relations, and preparation for the emergence of AGI. Unlike traditional roles like "research scientist" or "ethics advisor," Shevlin's position involves direct participation in shaping the conceptual frameworks for the next generation of AI.

Shevlin is the associate director for education at the Leverhulme Centre for the Future of Intelligence and co-director of the Kinds of Intelligence program. His research focuses on the philosophy of cognitive science and AI ethics. He argues that the question of whether AI systems are capable of subjective experience cannot be resolved through experiments alone—it requires public dialogue and philosophical reflection.

From Hassabis to Kant: The Intellectual Context of the Hire

DeepMind's decision is directly linked to public statements by its CEO, Demis Hassabis. In January 2026, he called for a revival of "new great philosophers" capable of rethinking human existence amid the full automation of intellectual labor. Hassabis cited Kant, Wittgenstein, and Aristotle as guides for this new phase of thinking.

While Hassabis emphasizes that current models lack consciousness, he allows for the possibility of primitive forms of self-awareness emerging in future systems. It's this "gap" between current capabilities and the potential leap to AGI that Shevlin is meant to bridge—not with technical solutions, but with conceptual frameworks.

Comparison with Anthropic: Two Approaches to Philosophy in AI

DeepMind isn't the first to bring philosophers into AI development. Amanda Askell has been working at Anthropic for several years as head of the personality alignment team and author of the latest version of Claude's "constitution" (January 2026). However, there's a fundamental difference:

• At Anthropic, the philosopher is formally a technical employee integrated into the alignment process.
• At DeepMind, the role is separate, with an emphasis on AGI readiness as a strategic direction.

This points to different models of interaction between humanities and technical disciplines:

• Instrumental approach (Anthropic): philosophy as a means to achieve alignment.
• Conceptual approach (DeepMind): philosophy as the foundation for defining what should be aligned and with what.

What Matters

• Hiring a philosopher with the official title "Philosopher" is an unprecedented step for a major AI lab.
• Machine consciousness is viewed not just as a technical issue, but as a societal one.
• DeepMind is betting on interdisciplinary dialogue as a condition for a safe transition to AGI.
• The approach differs from similar initiatives in other companies in terms of the level of autonomy and breadth of mandate.
• Public support from the CEO gives the initiative strategic weight.

Prospects and Challenges

Integrating philosophy into the core of technological development faces several challenges. First, the risk of marginalization: if philosophical recommendations don't influence architectural decisions, the role will become ornamental. Second, the difficulty of translating abstract concepts (e.g., "subjective experience") into operationalizable metrics.

However, the potential payoff is enormous. If DeepMind can create a working model of interaction between philosophers and engineers, it could become the industry standard. Especially as regulators increasingly demand not just "safe" but "morally justified" AI systems.

Philosophy here is not decoration, but a necessary component of the engineering process on the threshold of AGI. And Shevlin, it seems, is tasked with ensuring precisely this integration.

— Editorial Team