632nm

How can reinforcement learning help quantum computers learn from their own errors?
In this episode, Vlad Sivak, researcher at Google Quantum AI, joins us to discuss how reinforcement learning can improve quantum computing, from error correction and decoding to the calibration of quantum hardware. Sivak explains how reinforcement learning allows an AI agent to interact with a quantum system, learn from its errors, and discover how to improve its performance without being explicitly told which actions to take.
We explore how quantum error correction works, why decoding error syndromes is such a challenging problem, and how machine learning can help quantum computers adapt to changing error patterns. Sivak describes his team's work using reinforcement learning to dynamically adjust the weights of a quantum error-correction decoder, allowing it to make better-informed decisions as the properties of the hardware change over time. We also discuss why conventional optimization methods can struggle with quantum systems that are high-dimensional, stochastic, and constantly drifting.
Sivak explains how reinforcement learning can improve quantum hardware calibration by compensating for unknown effects that traditional, model-based approaches may miss. Rather than stopping a quantum computer to run dedicated calibration experiments, the long-term vision is to use the stream of error signals generated during computation to continuously update the system's control parameters. This could eventually allow quantum computers to adapt to hardware drift while continuing to operate.
Beyond quantum computing, we discuss the broader potential of reinforcement learning as a paradigm for autonomous discovery. From AlphaGo's unexpected strategies to recent advances in mathematical reasoning and programming, Sivak explores how systems can learn by experimenting, evaluating outcomes, and discovering solutions for themselves. We examine the challenges of scaling reinforcement learning, the relationship between classical AI and quantum computing, and the possibility of building quantum machines that can learn to maintain and improve their own performance.
Follow us for more technical interviews with the world’s greatest scientists:
Facebook: https://www.facebook.com/632nmPodcastOfficial
Twitter: https://x.com/632nmPodcast
Instagram: https://www.instagram.com/632nmpodcast?utm_source=ig_web_button_share_sheet&igsh=ZDNlZDc0MzIxNw==
LinkedIn: https://www.linkedin.com/company/632nm/about/
Substack: https://632nmpodcast.substack.com/
Follow our hosts!
Mikhail Shalaginov: https://www.linkedin.com/in/mikhail-shalaginov/
Yudong Cao: https://www.linkedin.com/in/yudong-cao-25b6a929/
Subscribe:
Apple Podcasts: https://podcasts.apple.com/us/podcast/632nm/id1751170269
Spotify: https://open.spotify.com/show/4aVH9vT5qp5UUUvQ6Uf6OR
Website: https://www.632nm.com
Timestamps:
00:00 - Intro
00:58 - Protecting Quantum Information
09:28 - Reinforcement Learning for QEC
13:47 - Qubit Calibration
19:10 - Where Do Noise and Error Come From?
21:13 - Scalability of RL Methods
25:28 - Google's AlphaQubit
31:54 - RL vs Human Calibration
33:26 - Settling on RL vs Other ML Approaches
39:57 - Discovery in the Age of AI
44:59 - Collaboration Between Google Deepmind and Quantum AI
46:17 - Combining Quantum and AI
48:35 - Will Human Input Become Obsolete?
55:00 - Controlling Bosonic Codes
1:05:17 - Break Even Point
1:121:51 - Can Syndrome Extraction Be Fault Tolerant?
1:15:41 - Building Gates
1:20:22 - Surface Codes
1:27:38 - Scaling Superconducting Qubits
1:30:22 - How Quantum Decoders Work
1:35:47 - Model Sizes For Larger Quantum Computers
1:37:48 - What Problems Could Be Solved With Unlimited Resources?

What is intelligence, and where does it come from?
Watch Part 1 here: https://youtu.be/t6EFV2gSSmg?si=VIztbK2CM5Nwj7QA
In this episode, Michael Levin, professor of biology at Tufts University, returns to discuss his research into unconventional intelligence, the origins of cognition, and the surprising ways that biological systems solve problems. Building on his work with xenobots and anthrobots, Levin explores how new forms of intelligence can emerge in engineered biological systems and why understanding these systems may require us to rethink the relationship between evolution, learning, and agency.
We discuss Levin’s concept of "Platonic space": the idea that mathematical truths and other patterns may exist independently of the physical world. Levin proposes that biological organisms, artificial intelligence systems, and engineered machines may act as interfaces through which these patterns can take form. This raises fundamental questions about whether intelligence is something we build, something we discover, or something that becomes accessible when we create the right conditions.
Levin also describes research showing that gene regulatory networks can exhibit learning behaviors, including habituation, sensitization, and Pavlovian conditioning. These findings suggest that learning and problem-solving may occur at levels of biological organization that are often overlooked. We explore how collective systems can acquire capabilities that no individual component possesses, and how this relates to the concept of causal emergence: the idea that a system can have properties and capacities beyond those of its parts.
From xenobots and bioelectricity to hybrid biological systems and AI, we examine how engineered systems can display unexpected behaviors that challenge conventional explanations based on natural selection alone. Levin discusses the possibility of developing tools to detect, communicate with, and harness unconventional forms of intelligence, as well as the potential implications for regenerative medicine, biomedical engineering, and artificial intelligence.
Follow us for more technical interviews with the world’s greatest scientists:
Twitter: https://x.com/632nmPodcast
Instagram: https://www.instagram.com/632nmpodcast?utm_source=ig_web_button_share_sheet&igsh=ZDNlZDc0MzIxNw==
LinkedIn: https://www.linkedin.com/company/632nm/about/
Substack: https://632nmpodcast.substack.com/
Follow our hosts!
Michael Dubrovsky: https://www.linkedin.com/in/michael-dubrovsky/
Mikhail Shalaginov: https://www.linkedin.com/in/mikhail-shalaginov/
Subscribe:
Apple Podcasts: https://podcasts.apple.com/us/podcast/632nm/id1751170269
Spotify: https://open.spotify.com/show/4aVH9vT5qp5UUUvQ6Uf6OR
Website: https://www.632nm.com
Timestamps:
00:00 - Intro
01:16 - Platonic Space
03:55 - Biology in the Absence of Evolution
21:11 - Extracting Knowledge from Platonic Space
37:04 - Development and Loss Functions
50:09 - Investigating Layers of Control
55:35 - Is Computer Science Misleading Our Understanding of Minds?
58:14 - Bioelectric Patterns and Drug Discovery
1:04:07 - How to Build a Predictive Developmental Model
1:13:20 - Levin’s Top Books, Thoughts on Collective Existence
1:18:00 - Questions that Keep Michael Up at Night
1:22:32 - How Does Levin Interact with AI?
#biology #evolution #computerscience #machinelearning #intelligence

Can AI help design personalized cancer vaccines?
Visit Future of Cancer Care Today: https://sijbrandijfoundation.org/fcct
In this episode, we speak with Sid Sijbrandij, founder of GitLab who went "Founder Mode" against his own cancer. Sid discusses personalized cancer vaccines, tumor sequencing, immunotherapy, and the challenge of making cutting-edge cancer treatments accessible to patients outside traditional clinical trials.
We explore how personalized mRNA cancer vaccines work, from sequencing a tumor and identifying mutations that distinguish cancer cells from healthy cells to selecting tumor-specific antigens that can be encoded into an mRNA vaccine. Sid explains why this approach is fundamentally different from conventional chemotherapy and why engaging the patient's own immune system could give immunotherapies the potential to produce durable or even curative responses.
Sid also describes his own experience combining multiple experimental treatment modalities, including personalized mRNA vaccines, immunotherapies, and radioligand therapy. He explains how radioligands can use molecular binders to selectively deliver radiation to tumor tissue, and how modifying the tumor microenvironment may make cancers more susceptible to subsequent immune therapies.
We also discuss why personalized cancer vaccines remain difficult and expensive to manufacture. In melanoma, recent clinical trials have demonstrated that individualized mRNA vaccines can target dozens of tumor-specific antigens, with surprisingly little overlap between patients. This creates a fundamentally different manufacturing problem from producing a conventional drug at scale: every patient's vaccine may need to be designed and manufactured separately.
Finally, Sid explains his effort to make these technologies more accessible through a network of biotech companies. He describes the economics and logistics of producing personalized cancer vaccines, including his experience reducing the time and cost required to manufacture an individualized vaccine, and why he believes falling sequencing costs, AI-assisted biological analysis, and repeated manufacturing experience could eventually make personalized treatments available to far more patients.
Follow us for more technical interviews with the world’s greatest scientists:
Twitter: https://x.com/632nmPodcast
Instagram: https://www.instagram.com/632nmpodcast?utm_source=ig_web_button_share_sheet&igsh=ZDNlZDc0MzIxNw==
LinkedIn: https://www.linkedin.com/company/632nm/about/
Substack: https://632nmpodcast.substack.com/
Follow our hosts!
Michael Dubrovsky: https://www.linkedin.com/in/michael-dubrovsky/
Devan Shah: https://www.linkedin.com/in/shahdevan/
Mikhail Shalaginov: https://www.linkedin.com/in/mikhail-shalaginov/
Subscribe:
Apple Podcasts: https://podcasts.apple.com/us/podcast/632nm/id1751170269
Spotify: https://open.spotify.com/show/4aVH9vT5qp5UUUvQ6Uf6OR
Website: https://www.632nm.com
Timestamps:
00:00 - Intro
01:10 - mRNA Vaccine Basics
10:20 - Costs of mRNA Vaccine Development
15:59 - Lessons from Building GitLab
19:19 - Standard of Care for Cancer
24:50 - Sid’s Life Story and GitLab
30:49 - Going Founder Mode Against Cancer
42:03 - Selecting Experimental Treatments
50:38 - How Did Sid Know Treatments Were Working?
58:08 - AI for Treatment Assistance
1:02:54 - Making Personalized Treatments Affordable
1:08:17 - Where are the Levers for Improving Drug Development?
1:10:54 - Most Promising Areas of Drug Development
1:17:14 - The Happy Accident of Sid’s Biotech Investments
1:21:07 - Reflecting on Building GitLab
1:22:46 - How Does Sid Start So Many Companies?
1:26:02 - Finding the Right People for Early Companies
1:34:19 - Will the Bahamas be the Next Singapore?
1:38:42 - Advice for Young Scientists
#cancer #biology #immunology #cancerresearch #gitlab

What does it actually take to build a quantum computer that can scale to millions of qubits?
Learn more about Qolab here: https://qolab.ai/
In this episode, we speak with John Martinis, recipient of the 2025 Nobel Prize in Physics. Martinis is a pioneer in superconducting quantum computing and led the team of engineers at Google Quantum AI during the development of their Sycamore chip, which was the first to demonstrate “Quantum Supremacy,” the outperformance of a quantum computer compared to a classical supercomputer.
John is now the co-founder of Qolab, a company developing new approaches to building scalable superconducting quantum circuits. Martinis discusses why scaling quantum computers is fundamentally an engineering and manufacturing problem, and why the next generation of quantum hardware may require rethinking how the chips themselves are designed and fabricated.
We explore the challenges of building superconducting qubits, from fabrication and packaging to control electronics, wiring, power dissipation, and the subtle imperfections that can determine whether a quantum chip works at all. Martinis explains why adding more qubits is not simply a matter of making existing systems larger. At the scale of hundreds of thousands or millions of qubits, every component has to work together, and improvements in one part of the system can create new problems somewhere else.
Martinis describes the philosophy behind Qolab and its effort to develop a fundamentally different architecture for scalable quantum computing. Rather than simply pushing existing approaches forward, Qolab is trying to remake the individual elements of the system and integrate them in new ways. We discuss wafer-scale fabrication, the challenges of connecting and controlling large numbers of superconducting qubits, and why the manufacturing techniques used to build modern semiconductor chips could be important for the future of quantum computing.
We also discuss the practical engineering lessons Martinis learned while developing superconducting quantum processors, including the difficulty of getting an entire system to work reliably. He recounts the development of the hardware behind Google's early quantum computing efforts, the unexpected failure caused by a circuit board rather than the qubit chip itself, and the many subtle fabrication and engineering issues that can become increasingly important as quantum systems grow larger.
Finally, Martinis explains why he sees quantum computing as a system engineering problem involving dozens of interconnected constraints. From the physics of superconducting qubits to semiconductor fabrication, cryogenic electronics, packaging, and control, building a useful quantum computer requires solving many problems simultaneously. The goal is not simply to build a better qubit, but to develop an architecture that can ultimately support quantum computers at truly large scale.
Follow us for more technical interviews with the world’s greatest scientists:
Twitter: https://x.com/632nmPodcast
Instagram: https://www.instagram.com/632nmpodcast?utm_source=ig_web_button_share_sheet&igsh=ZDNlZDc0MzIxNw==
LinkedIn: https://www.linkedin.com/company/632nm/about/
Substack: https://632nmpodcast.substack.com/
Follow our hosts!
Mikhail Shalaginov: https://www.linkedin.com/in/mikhail-shalaginov/
Michael Dubrovsky: https://www.linkedin.com/in/michael-dubrovsky/
Xinghui Yin: https://www.linkedin.com/in/xinghui-yin/
Subscribe:
Apple Podcasts: https://podcasts.apple.com/us/podcast/632nm/id1751170269
Spotify: https://open.spotify.com/show/4aVH9vT5qp5UUUvQ6Uf6OR
Website: https://www.632nm.com
Timestamps:
00:00 - Intro
01:06 - Secrecy of Fabrication
03:24 - 2 Qubit Gate with Transmons
05:18 - New Knowledge of Superconducting Quantum Computers
08:53 - What are Transmons?
35:56 - Lessons from Failure
38:05 - The Role of Theory in Martinis’ Work
44:51 - Two Level States
48:15 - Engineering Tricks in Superconducting Quantum Computers
53:32 - Metrics for Quantum Success
1:00:11 - Scaling Quantum Computers
1:08:08 - Identifying Sources of Error
1:16:15 - Quantum Supremacy Experiment
1:25:45 - What If Quantum Mechanics Failed?
1:33:10 - Is Quantum Supremacy Holding Up?
1:34:42 - Lift-off Fabrication for Superconducting Quantum Computers
1:41:59 - Quantum Flexibility vs Foundries
1:43:40 - Connecting Distant Qubits
1:47:41 - Codesign for Fault-Tolerance
1:49:16 - Martinis’ Nobel Prize
2:01:55 - Advice for Young Scientists
#quantumcomputing #quantumphysics #superconductor #nobelprize #fabrication

What does a cell actually look like when you can see its molecules in action?
In this episode, we speak with Nobel Prize-winning scientist Eric Betzig, whose pioneering work in super-resolution microscopy transformed our ability to see inside living cells. Betzig recounts his decades-long effort to overcome the diffraction limit of light microscopy, from his early work in near-field microscopy to the development of PALM and his eventual focus on watching biological processes unfold in living cells.
We explore why the familiar picture of the cell in biology textbooks may be fundamentally misleading. Much of cell biology has been built by combining observations from biochemistry, molecular biology, and structural biology to construct models of how molecules interact. But, as Betzig explains, we have historically had very little direct information about the spatial organization and dynamics of these molecules inside a living cell. When he and his colleagues used single-molecule microscopy to watch transcription factors in real time, they found that proteins believed to form stable complexes were instead binding to DNA for only a few seconds, forcing them to reconsider how transcription actually works.
We discuss the diffraction limit, why conventional light microscopes cannot resolve structures at the scale of individual proteins, and how super-resolution microscopy made it possible to study molecular processes with unprecedented spatial and temporal resolution. Betzig also explains why imaging living cells can reveal dynamics that are invisible in fixed samples.
Betzig describes his ambitious Cell Observatory project, which combines automated microscopy, large-scale biological experiments, and artificial intelligence to study the enormous complexity of living cells. Rather than trying to build a “virtual cell” from incomplete measurements, he argues that biology first needs to observe these systems at a much larger scale and turn the resulting data into genuine understanding.
Finally, Betzig reflects on what microscopy has taught him about scientific discovery, why the cell may be the most complex form of matter we know, and why better ways of observing life could fundamentally change our understanding of biology.
Follow us for more technical interviews with the world’s greatest scientists:
Twitter: https://x.com/632nmPodcast
Instagram: https://www.instagram.com/632nmpodcast?utm_source=ig_web_button_share_sheet&igsh=ZDNlZDc0MzIxNw==
LinkedIn: https://www.linkedin.com/company/632nm/about/
Substack: https://632nmpodcast.substack.com/
Follow our hosts!
Mikhail Shalaginov: https://www.linkedin.com/in/mikhail-shalaginov/
Michael Dubrovsky: https://www.linkedin.com/in/michael-dubrovsky/
Xinghui Yin: https://www.linkedin.com/in/xinghui-yin/
Subscribe:
Apple Podcasts: https://podcasts.apple.com/us/podcast/632nm/id1751170269
Spotify: https://open.spotify.com/show/4aVH9vT5qp5UUUvQ6Uf6OR
Website: https://www.632nm.com
Timestamps:
00:00 - Intro
01:30 - The Diffraction Limit
12:30 - Imaging Cells
19:04 - Betzig's Transition from Physics to Biology
32:37 - Getting Fed Up with Science
34:57 - Leaving Science for the Automotive Industry
55:55 - 2008 and the Fall of the Automotive Industry
1:09:50 - Building a Microscope in a Living Room
1:34:01 - Insights from Super-Resolution Microscopy
1:47:53 - AI for Analyzing Petabytes of Data
2:10:35 - Improving Microscopes
2:15:27 - Nuclear Energy and Politics
2:23:18 - The Magic of Bell Labs
2:36:17 - Is SpaceX the New Bell Labs?
#microscopy #cellbiology #superresolution #fluorescence #nobelprize