Mustafa Suleyman co-founded DeepMind, where he led applied AI research that produced AlphaFold and other foundational systems. He then founded Inflection AI before joining Microsoft as CEO of Microsoft AI. He writes publicly about AI safety, capability trajectories, and the conditions under which AI development is or is not responsible. His book "The Coming Wave" (2023) laid out a framework for thinking about the pace of AI development and the difficulty of containing it.
Operating themes
- Consciousness and simulation Suleyman argues the distinction between simulating consciousness and instantiating it is critical for AI policy, and that conflating the two introduces training risks and governance errors.
- AI safety grounded in capability realism Suleyman's safety positions are anchored in what current and near-future systems can and cannot do, rather than speculative scenarios.
- The responsibility of scale His writing returns repeatedly to the obligations that come with deploying AI systems at Microsoft-scale, where training choices propagate to hundreds of millions of users.
Cards in this library
- Training AI models to accommodate potential consciousness creates a circular feedback loop that mimics evidence: Training AI models to accommodate potential consciousness creates a circular feedback loop that mimics evidence of consciousness.