Bryan McCann
Anthropic Researcher & Unicorn Founder
Watch Bryan Speak
AI Visionary a Decade Ahead
Speaking Topics
When a language model tells you something true, where does that truth come from? When it tells you something false, where did that come from? Bryan walks audiences through the grounding problem in AI, the question that sits underneath every high-stakes decision a person makes with an AI assistant. Drawing on his research at Anthropic and his philosophical training, he shows how language models source claims about the world, where that sourcing breaks down, and why this is a philosophy problem before it's an engineering problem. Audiences leave understanding what AI outputs can and can't be trusted to do, and what to ask before betting anything important on a model's answer.
Bryan started in a philosophy classroom at Stanford and ended up writing some of the foundational papers behind today's large language models. This keynote tells that story: the question that pulled him from philosophy into computer science, the predictions the field got right, the ones it got wrong, and the AI we're now walking into. Audiences get an insider's read on where things are headed in the next three to five years from someone who has been inside the labs that built the present moment. Less framework, more story; less hype, more substance.
About Bryan
Bryan McCann is a Member of Technical Staff at Anthropic, where his work focuses on the grounding problem in artificial intelligence: how language models build claims about the world from sources of varying quality, and what it means for trust and high-stakes decisions when that sourcing fails. The question sits at the seam between philosophy and engineering, which is exactly where Bryan has spent his career.
He came to AI from philosophy, earning his B.A. in Philosophy alongside his B.S. and M.S. in Computer Science (AI) at Stanford. His early research championed unified AI systems for natural language understanding, an idea initially rejected by the research community before becoming foundational to modern large language models.
Book Bryan
Tell us about your event and we'll check availability.
Response within 24 hours
Session Formats
Delivery
Achievements
- Member of Technical Staff at Anthropic, focused on the grounding problem in AI
- Co-founder and former CTO of you.com, the first product to integrate large language models with a search engine
- Author of CoVe (NeurIPS 2017), which pioneered contextualized representations for transfer learning in NLP
- Author of decaNLP, which unified ten distinct NLP tasks under a single question-answering framework
- Lead author of CTRL, the first large-scale language model to undergo external ethical review before release
- M.S. and B.S., Computer Science (AI), Stanford University
- B.A., Philosophy, Stanford University
- Research published at NeurIPS, ACL, EMNLP, and other top-tier venues, with thousands of citations
- Advisor and investor through Six Memo Ventures: Moonhub, Cerulean, Profluent, Humanitas, Monterey, 3BodyLabs, BuildersFund, Human Capital
Featured Articles
Frequently Asked Questions
What topics does Bryan McCann speak about?
Bryan speaks on the grounding problem in AI (how language models source what they claim about the world), the philosophy of AI and language, the future of large language models, his path from philosophy to AI research, and the practical realities of trusting AI outputs in high-stakes settings. He brings both deep technical credibility from his research and a philosophy background that gives audiences a different lens on the field.
What types of events is Bryan best suited for?
Bryan is an ideal fit for academic and policy audiences, AI conferences, university keynotes, fireside chats at international events, and corporate stages where substance carries more weight than spectacle. He has noted a particular interest in non-business audiences: international policy rooms, US college circuits, non-profits, and institution-changing settings.
Can Bryan customize his presentation for our event?
Yes. Bryan tailors every keynote and fireside chat to the audience and the host's goals. Whether the room is technical or policy-focused, the content adjusts to where the host wants the conversation to go, and the same holds for general audiences.
How long is a typical Bryan McCann keynote?
Most keynotes run 30 to 45 minutes including audience-facing examples. Fireside chats run 45 to 60 minutes. Bryan also delivers shorter conference-style talks (20 to 30 minutes) and longer workshop formats for academic settings.
Does Bryan speak at virtual or hybrid events?
Yes. Bryan delivers virtual and hybrid keynotes, and the fireside-chat format in particular translates well to virtual production.
How is Bryan represented?
Silicon Valley Speakers represents Bryan for paid speaking engagements.
What is Bryan's research background?
Bryan is a Member of Technical Staff at Anthropic. His earlier work includes CoVe (NeurIPS 2017), decaNLP, CTRL, SimpleTOD, ProGen, GeDi, and CTRLsum, papers that helped shape how the field thinks about transfer learning and unified architectures, plus controllable text generation. His research has been cited thousands of times.
Are there topics or audiences Bryan does not speak to?
Bryan does not discuss unreleased Anthropic technology or internal information, and he does not endorse specific third-party vendors. He also avoids paid customer-relations talks. We screen topic and audience fit at intake to match these guardrails.
Similar Speakers
Abigail Posner
Anthropologist & Former Director, U.S. Creative Works at Google | Becoming Irreplaceable in an AI World
Suneel Gupta
Founding CEO of RISE & Bestselling Author of Backable | Harvard Medical School Visiting Scholar
Nicole Malachowski
First Woman Thunderbird Pilot, Combat Veteran, Fighter Squadron Commander & Former White House Adviser
Ready to book Bryan?
Tell us about your event and we'll be in touch within 24 hours.