Big Technology Podcast: Google Research Head Yossi Matias on AI for Cancer, Quantum, and the Future of Research
Host: Alex Kantrowitz
Guest: Yossi Matias, Head of Google Research
Date: October 27, 2025
Overview
This episode features a wide-ranging conversation between Alex Kantrowitz and Yossi Matias, head of Google Research, discussing how AI is transforming scientific research, especially in cancer and quantum computing. Matias dives into the intersection of research and product development, the practical realities and promise of quantum computing, philosophical distinctions between breakthrough and innovation, and the future role of researchers in an AI-powered world. The discussion, held live at Google’s Mountain View headquarters, highlights both recent achievements and the evolving landscape of research in the age of AI.
Key Discussion Points and Insights
1. AI in Cancer Research — “Cell to Sentence” and Beyond
- Significance of Recent Work: Google Research, in collaboration with Yale and DeepMind, used large language models (LLMs) to identify a substance that prompts cancer cells to reveal themselves to the immune system—a previously overlooked approach.
- Yossi Matias (02:11):
“We see the progress of AI as transformative… with AI models, generative AI, we now have better understanding to understand patterns… This is a step toward some of the biggest challenges in healthcare.”
- Yossi Matias (02:11):
- Empowering Scientists with AI: AI is becoming a “co-scientist,” allowing researchers to ask bigger questions and test hypotheses faster.
- The Research Process: AI accelerates sifting through data, forming hypotheses, and validation, unlocking hidden patterns previously out of reach.
Notable Quote
- “This is really using AI agents to help out, sift through the information and do the kind of work that in the past only, you know, very sophisticated people could do.”
— Yossi Matias (02:55)
2. Quantum Computing: Reality vs. Hype
- Recent Quantum Milestone: Google’s quantum computer outpaced a classical supercomputer by 13,000x for a specific algorithm, yielding the first verifiable, practical “quantum advantage.”
- Long-Term Perspective: Despite media hype, practical widespread quantum computing remains a 5+ year journey, with important incremental milestones. Progress is steady but drawn out compared to other fields.
- Yossi Matias (05:15):
"Quantum computing is a very long-term quest… But we have a very steady progress on very measurable timeline and very clear milestones."
- Yossi Matias (05:15):
Quantum's Broad Potential
- Transforming Science and AI: Quantum will open new domains of inquiry—“material” change. It may reveal abilities not yet imagined and help accelerate AI itself by unlocking new kinds of knowledge and computation.
- Unexpected Innovations: Many coming breakthroughs are unknowable today. As quantum and AI amplify each other, new fields will emerge unexpectedly.
Notable Quote
- “Once you uncover opportunity, suddenly it creates the kind of thing that perhaps you did not anticipate… For many of us [AI] seemed like science fiction just a few years ago. And it’s just accelerating.”
— Yossi Matias (07:02)
3. Research and Product: Striking the “Magic Cycle”
-
Bringing Research & Product Together: Google thrives by tightly coupling research breakthroughs with real-world applications, forming a “magic cycle” where each drives and enriches the other.
- Yossi Matias (08:33):
“All of our research projects are motivated by problems in the real world… then taking the result and applying them back was, to me, the most fascinating aspect.”
- Yossi Matias (08:33):
-
Risks of Too Much Togetherness: There’s a necessary balance—product timelines can pressure research to become short-termist, so leadership judgment is key to maintaining long-term research integrity.
- Yossi Matias (10:56):
"In any development environment… one of the important things is to have this balance between what you need to do tomorrow and how to invest in the future."
- Yossi Matias (10:56):
4. Innovation vs. Breakthrough: What’s the Difference?
-
Definition:
- Innovation is iterative, improving products/services continuously.
- Breakthrough solves problems previously thought unsolvable, often by inventing new paradigms.
- Yossi Matias (12:32):
“Innovation is something that we do all the time… When I think about research breakthroughs, this is about problems that currently we don’t know how to solve in principle…”
-
“Technology Transfer” is a Myth: Research and product inform each other in a cycle—not a linear handoff. Judgment is always needed for timing real-world application.
5. The Value of Long-Term Research
- Long-Term Vision: Most transformative breakthroughs result from audacious, multi-year projects, broken into tangible research and product milestones (e.g., Google’s work in transformers and flood forecasting).
- Flood Forecasting Example: Started in 2017, now covers 2B people in 150 countries—a product of successive research milestones.
- “No research is detached:” Even the most speculative projects are driven by real needs or by a vision of the possible.
Notable Quote
- “The important thing is to actually have the validation, the peer review, and everything that's good, and then taking it back… and this generates the next questions.”
— Yossi Matias (09:51)
6. Where Will the Next Breakthroughs Come From?
-
Research Is Exploration: Progress often comes from the unknown. While researchers are intentional with their bets, the unknown remains vast and exciting.
- Yossi Matias (19:43):
“The beautiful thing about research is that it’s really exploring the unknown… The most exciting thing are the things that we don’t know yet.”
- Yossi Matias (19:43):
-
Algorithmic Innovation vs Compute: True progress requires both—scaling existing models (data/compute) and fresh algorithmic insights (e.g., transformers). Many coming advances are expected to be algorithmic.
- Yossi Matias (21:04):
“There are going to be more algorithmic innovations that are going to make breakthroughs. The best is yet to come.”
- Yossi Matias (21:04):
7. Motivating Researchers Beyond Generative AI
- Inherent Drive: The best researchers are motivated by hard, important problems—across AI, quantum, genomics, Earth sciences, and more.
- Interdisciplinary Collaboration: Google’s environment allows fluid movement and cross-fertilization between fields; the organizational “full stack” enables breakthroughs in many domains, not just generative AI.
Notable Quote
- “People are excited to work on things that matter and can actually apply their brilliance and innovation and have breakthrough research… When we bring together talents… a lot of the magic happens.”
— Yossi Matias (24:07)
8. The Future of Research in an AI World — More Scientists, Bigger Questions
- AI Augments, Doesn’t Replace: LLMs and AI “co-scientists” enable more researchers to ask and answer bigger questions, not fewer.
- Amplification Effect: AlphaFold hasn’t reduced the need for protein researchers; it lets them focus on more advanced problems. AI tools give every grad student or postdoc an advanced lab, increasing both ability and ambition.
Notable Quote
-
“The only situation where you need less researchers is if you assume that we practically almost answered all the questions that we need to have. I don’t think anybody… would think that.”
— Yossi Matias (26:46) -
Societal Mission: From healthcare to education, from natural disasters to climate, AI’s amplification of human ingenuity opens vast new possibilities for discovery and social impact.
- “With AI, there’s an opportunity for more teachers to be more effective… the innovation is going to come from [the next generation].”
— Yossi Matias (28:37)
- “With AI, there’s an opportunity for more teachers to be more effective… the innovation is going to come from [the next generation].”
Memorable Quotes
- “AI is an amplifier of human ingenuity. It really empowers scientists, healthcare workers, teachers, business people… The more we’re making advancements with AI, the more we can expect all these professionals to take on bigger missions and to do bigger progress for the benefit of humanity.”
— Yossi Matias (29:23)
Timestamps for Major Topics
- AI for Cancer Research (“Cell to Sentence”): 01:44–04:38
- Quantum Computing Breakthroughs: 04:38–08:24
- Research-Product Interplay, “Magic Cycle”: 08:24–12:10
- Innovation vs. Breakthrough: 12:10–14:16
- Long-term vs. Short-term Research: 14:16–17:52
- Future of Breakthroughs, Algorithm vs Compute: 19:43–23:32
- Researcher Motivation & Interdisciplinary Magic: 23:32–26:11
- AI's Impact on Research Workforce: 26:11–29:44
Final Takeaway
Yossi Matias paints a picture of a future dramatically reshaped by AI, where researchers are empowered—not replaced—to pursue ever more ambitious questions across disciplines. Google’s approach, tightly integrating long-term research with pragmatic product cycles, is accelerating breakthrough science in healthcare, quantum computing, climate, and more. AI and quantum research will both reveal and create fields not yet imagined, marking the dawn of an era where human curiosity—amplified by technology—drives transformative progress for society.
