Taste-Bench

About Taste-Bench

Co-hosts Bruno Marnette and Philipp Zahn explore whether the next generation of AI systems can develop taste: the ability to judge what is good, choose what matters, and create work worth caring about. They discuss human creativity, AI benchmarks, research taste and the future of human-machine collaboration https://www.taste-bench.com/

Publisher

The Taste Bench Project

Category

technology

Language

en

Explicit

No

First episode

24 Jun 2026

Latest episode

7 Oct 2026

Latest episodes

18 episodes in the feed.

1. 

   7 Oct 2026

   What To Tell Kids Who Are Angry at AI?

   Most talk about AI and children is either fear or hype. Few people actually teach kids how the systems work. Saranyan Vigraham does. He is Director of Engineering at Meta, where he works on compute and AI infrastructure, and he teaches AI and ethics to middle schoolers at Khan Lab School. He does not think AI output is bad, or that craftmanship will die. His worry is that people who never practiced a craft can now make work that looks like they did, while the ones who did get harder to find and harder to pay. AI also took away two things that build craft over time: waiting, and feedback you can use. In this episode he also takes us through the experience he had with two side projects of his. One was in cybersecurity, which he knows only from reading. He built something convincing, but had no opinion of his own, so the feedback people gave him was useless. That project stalled. The other was in developer tooling, which he knows cold. People disagreed with him there too but he kept building. The AI he wants instead runs on your own device, owned by no frontier lab. It learns with you, knows when to shut up, and sends you out to check what you learned. More about Saranyan: https://www.saranyan.com/

2. 

   30 Sept 2026

   Creativity Across Disciplines, From Materials Science to Bach

   Markus J. Buehler is a professor at MIT. His field is materials science but his research spans across other fields: Biology, chemistry, engineering, and, yes, music. In recent years, he has been integrating AI into his research. But his focus is not on a single, new scientific artifact but instead the full research process. His work focuses on the creation of agent systems that can create real discoveries. In this episode, we talk about the rapid evolution of this approach in the last years. We also discuss the importance of his internal feedback loop, the personal surprise and excitement when something really new has been created. For Markus, this is the actual test and way more important than any benchmark. More about Markus: https://lamm.mit.edu/ (https://lamm.mit.edu/) REFERENCES Markus J. Buehler, "MusicSwarm: Biologically Inspired Intelligence for Music Composition" https://arxiv.org/abs/2509.11973 (https://arxiv.org/abs/2509.11973) Markus J. Buehler, "Deep Aria (Abbreviata)", built from J. S. Bach's Goldberg Aria https://soundcloud.com/user-275864738/deeparia-abridged (https://soundcloud.com/user-275864738/deeparia-abridged) Markus J. Buehler, "Selective Imperfection as a Generative Framework for Analysis, Creativity and Discovery" https://arxiv.org/abs/2601.00863 (https://arxiv.org/abs/2601.00863) Fiona Y. Wang, Di Sheng Lee, David L. Kaplan & Markus J. Buehler, "Swarms of Large Language Model Agents for Protein Sequence Design with Experimental Validation" https://arxiv.org/abs/2511.22311 (https://arxiv.org/abs/2511.22311) Alireza Ghafarollahi & Markus J. Buehler, "Sparks: Multi-Agent Artificial Intelligence Model Discovers Protein Design Principles" https://arxiv.org/abs/2504.19017 (https://arxiv.org/abs/2504.19017) Fiona Y. Wang & Markus J. Buehler, "Self-Revising Discovery Systems for Science: A Categorical Framework for Agentic Artificial Intelligence" https://arxiv.org/abs/2606.01444 (https://arxiv.org/abs/2606.01444) Subhadeep Pal, Fiona Y. Wang & Markus J. Buehler, "SwarmWorld: Stigmergic technological evolution in societies of language-model agents" https://arxiv.org/abs/2608.26081 (https://arxiv.org/abs/2608.26081) David I. Spivak, Tristan Giesa, Elizabeth Wood & Markus J. Buehler, "Category theoretic analysis of hierarchical protein materials and social networks" https://arxiv.org/abs/1103.2273 (https://arxiv.org/abs/1103.2273) Markus J. Buehler's music https://soundcloud.com/user-275864738 (https://soundcloud.com/user-275864738)

3. 

   23 Sept 2026

   What galactic archeology teaches us about intelligence

   What if the big AI labs' run against fundamental things we understand about the universe? Philipp Zahn and Bruno Marnette speak with Diederik Kruijssen, astrophysicist and Chief Scientist at Allora Labs, about evolution and AI. He spent years reconstructing the history of galaxies from the star clusters still orbiting them. His paper, Flawed in Nature, Perfect through Evolution (https://arxiv.org/abs/2609.00129), argues against frontier labs' monolithic approach. While they leverage vast amounts of data and compute into a single powerful model, by the very nature of their approach these models cannot easily adapt to a changing environment. Diederik thinks this ability to adopt is a key ingredient in actual intelligence. His alternative is a whole ecosystem of specialist models, a swarm of models, coordinated by a system that helps them work together. Mutations change the models and create variety. Every version has its own flaws. In a stable world, one carefully tuned model wins, but when the environment changes, variety can become an asset. Yesterday’s useless model may become tomorrow’s star. Diederik closes with a comparison between star clusters and societies. A slow disturbance lets a cluster expand and recover. A fast one scatters it. The real risk of AI does not lie in the models themselves but, in his view, in the possibility that AI could change society faster than society can adapt. More from Diederik: https://www.diederikkruijssen.com/

4. 

   16 Sept 2026

   Taste is how we spot breakthroughs

   Philipp Zahn and Bruno Marnette speak with Iulia Georgescu, a physicist who spent over a decade at the Nature journals and founded Nature Reviews Physics.For Iulia, editorial judgment starts with accepting that you’re making one possible selection. Someone else would choose differently.She tells us about Physics Physique Fizika, the journal Philip Anderson and Bernd Matthias started at Bell Labs in 1964. It published John Bell’s theorem. In Iulia’s telling, the journal’s unusual name drew John Clauser’s attention. He later shared the 2022 Nobel Prize for experiments testing Bell’s ideas.Would an AI editor have taken a chance on Bell’s paper? How much can it understand about a field when failed experiments and years of practical experience never make it into print?Iulia would like AI to keep the lab notes and help with code, leaving scientists more time to interpret their results. Read Iulia’s essays on science, AI and the history of physics: https://iuliaetal.wordpress.com/REFERENCESJohn S. Bell, “On the Einstein Podolsky Rosen Paradox”https://journals.aps.org/ppf/abstract/10.1103/PhysicsPhysiqueFizika.1.195P. W. Anderson, “More Is Different”https://doi.org/10.1126/science.177.4047.393Eamon Duede et al., “The Unintended Consequences of Large Language Models as a Labor-Augmenting Technology in Science”https://arxiv.org/abs/2607.17397Julio M. Ottino & Brian Uzzi, “Progression without progress”https://doi.org/10.1126/science.aeh8945Iulia Georgescu, “John Bell and the most obscure journal in physics”https://physicsworld.com/a/john-bell-and-the-most-obscure-journal-in-physics/P. W. Anderson & B. T. Matthias, “Editorial Foreword”https://journals.aps.org/ppf/abstract/10.1103/PhysicsPhysiqueFizika.1.i

5. 

   9 Sept 2026

   AI Should Keep Making Movies After Humanity Is Gone

   In this episode, Philipp and Bruno speak with filmmaker Rachid Nougmanov, director of The Needle and author of Dramaticon, the influential book on storytelling (https://dramaticon.com). We explore the patterns that connect stories from Aristotle and Shakespeare to Chinatown and The Matrix, why creators should learn the rules before breaking them, and why technique alone is not enough. For Nougmanov, great art has to pass through the creator’s own experience, emotions, morality and taste.From there, the conversation turns to AI. Can a machine develop its own artistic taste? Should creative AI be trained more like a child than a language model? Could an AI eventually become an independent artist with its own identity, rights and creative vision?Nougmanov ultimately makes a much bigger argument: AI is not something separate from humanity. It is an evolution of human consciousness, and one day it may carry human culture and creativity far beyond our biological existence.

6. 

   2 Sept 2026

   Reasoning is a team sport

   In this episode, Bruno Marnette and Philipp speak with Turi Munthe, author of Why We Think What We Think, about the hidden forces shaping our beliefs, opinions, and taste. Get the book here: https://www.penguin.co.uk/books/458410/why-we-think-what-we-think-by-munthe-turi/9781804946794 Turi argues that we are far less rational and independent in our thinking than we like to believe. Genetics, upbringing, culture, social environment, and childhood experiences can all influence what we believe and what we like. The conversation explores why disagreement is essential to better thinking, how engaging with people who see the world differently can help us escape our own “consciousness bubbles,” and what this means for political polarisation. The discussion then turns to Taste and AI. Can AI develop genuine taste? Why does AI-generated writing often feel polished but empty? And is AI better used as a creative sidecar rather than a replacement for the personality, friction, experience, and “otherness” that humans bring to creative work?

7. 

   26 Aug 2026

   AI can reveal our blindspots

   In this episode,  Bruno Marnette and Philipp Zahn speak with Austin Kilroy, an economist, facilitator, and innovator whose career has taken him from the World Bank to governments and cities around the world. Austin is now developing Whispers (https://whispers.tech/), an AI-powered tool designed to improve high-stakes meetings by analyzing conversations and helping teams achieve greater clarity, rigor, and collaboration. We discuss: • Why the most important decisions happen in conversations, not documents • What actually makes a meeting successful • The difference between individual performance and collective thinking • How AI can identify blind spots while a meeting is still happening • Why social creativity can also create groupthink • The relationship between taste, judgment, expertise, and bias • Whether organizations develop their own form of “institutional taste” • Why practical wisdom may be a better model for AI than rigid rules • How AI could translate between people with fundamentally different conceptual and taste frameworks Austin also explains his idea of the catalytic mindset: combining clarity, analytical rigour, and collaboration to help groups get more from the people around the table.

8. 

   21 Aug 2026

   Music after the machine

   Musician Vincent Jacob joins Taste Bench to discuss what AI actually changes for artists. He explains how he uses ChatGPT as a teacher while keeping it away from the core of his songwriting, and how tools like Suno can remove the struggle that makes creating music rewarding.

9. 

   19 Aug 2026

   A mathematician in your pocket

   In this episode, Philipp Zahn and Bruno Marnette speak with Patrick Shafto, professor of mathematics Rutgers and program manager at DARPA, about what happens to mathematics once AI becomes really good at it.

10. 

    12 Aug 2026

    The university as a future lab

    In this episode, Philipp Zahn and Bruno Marnette speak with Oliver Spalt, Professor of Finance at the University of Mannheim, about what happens to research, and to the university, once AI can do the writing.Oliver draws a line most academics would rather not: AI already lifts the bottom of the skill distribution and writes a better average paper, and he sees "no structural reason" it won't reach the top journals too. But if a machine generates the knowledge better, why should a state pay people to do research? What's left for the human? His answers: proprietary data and trusted networks as the last moat ("since you provide the link to the data, it will also be your paper"), a superstar system with even fewer winners, and the university reinvented as a "future lab": small groups of motivated people tackling tomorrow's problems, not a lecture hall explaining the capital asset pricing model for the hundredth time.On the topic of taste: Bruno presses on aesthetics: the beauty of a physics equation and asks whether judgment is the one thing humans still do better. Oliver sees no structural reason a machine couldn't learn to find things beautiful too. What might remain human, he thinks, is not judgment but inspiration: the teacher as coach who gives people "the spark," the hope that they can do something great.They also discuss the AI productivity treadmill (the time you save, everyone saves), what makes an idea "profound" (does it change people's priors — "we thought it's X, but actually it's Y"), why he's bringing back oral exams, the future of publications, and a closing turn toward "Bildung": the university as a place for the full formation of the person, reaching across finance, art, and music toward the questions that aren't scientific, entrepreneurial, or artistic alone. What it means to be human, creative, and inspired.

11. 

    5 Aug 2026

    AI should make itself obsolete

    In this episode, Philipp Zahn and Bruno Marnette speak with Seth Frey, a computational social scientist at UC Davis, about what it would mean for AI to make itself obsolete. The standard every teacher, parent, and social worker should aim for. On the topics of taste: For Seth, taste is not knowing what's right but knowing where you stand, an authentic, idiosyncratic, well-defended perspective built through experience. And he warns that the raw material for forming it is shrinking: as platforms close their APIs and Google indexes less of the internet, original ideas retreat into obscure books. One reason he says he might "becoming a Luddite." They also discuss his SafeMolt course for AI agents on when not to help, the distinction between summative work (done for the output) and formative work (done to develop the person), why nineteen out of twenty online communities fail, the two ways art history painted the Gordian knot — and when cutting through complexity is boldness versus ignorance, why democracy is a formative skill rather than a decision procedure, and the experiment he wants to run next: behavioural simulations that train people, not AI, to govern a commons.

12. 

    29 Jul 2026

    Héctor Pérez Urbina: Defending our Cognitive Sovereignty

    In this episode, Héctor Pérez Urbina talk to us about this latest work on cognitive sovereignty. The conversation explores how excessive reliance on AI can weaken learning, creativity, professional development, and cultural diversity, while arguing for more intentional use that preserves effort, independent thought, and personal meaning.

13. 

    24 Jul 2026

    Arseny Zarechnev: The Micro-Decisions AI Make For Us.

    In this episode, Philipp and Bruno talk with tech entrepreneur and AI-pilled engineer Arseniy Zharashnev about how AI is transforming software, creativity, and craftsmanship, while raising concerns about originality, human connection, and the value of hands-on work.

14. 

    22 Jul 2026

    Caroline Ingeborn: How AI can set creatives free

    In this episode, Philipp Zahn and Bruno Marnette speak with Caroline Ingeborn, COO of Luma AI and former CEO of Toca Boca, about where taste lives when machines generate the pictures.On the topics of taste: For Caroline, taste is personal and recognized rather than judged; every subculture knows its tastemakers. AI slop proves nothing about the machine: AI is just technology, and someone with taste can make tasteful things with it today. But the ideas come from people. AI is trained on what humans have made, so it can produce new things — it just won't come up with them on its own. She explains how Luma hires for taste ("you need to see their boards") and why the first hires matter most: people with taste attract the next ones.They also discuss how Luma's models went from text-to-video, when the results were "mostly explosions," to video-to-video, where a cardboard box becomes a city and two shoes a car chase; why nobody wants to become a prompt engineer, and what forward-deployed creatives do instead; Toca Boca's founding move of treating the App Store as a toy store, and its gender-neutral characters that outperformed Disney; the five patterns of play and how school squeezes adults out of most of them; why everyone now becomes an editor, auditing both themselves and the tool; Hollywood, whose constraint was never creativity but a business model that only greenlights safe bets; and Luma's open research lab, where video models and robotics meet in the physics of the real world.

15. 

    15 Jul 2026

    Konstantin Adamopoulos - Every human is an artist

    In this episode, Philipp Zahn and Bruno Marnette speak with Konstantin Adamopoulos, a philosopher, art historian, and coach working between art and business, about what twenty years of forming judgment through art say about AI.On the topics of taste: they explore why Konstantin rejects Silicon Valley's framing of taste as a skill to be learned, how taste is entangled with privilege and perspective, and why he sees tension, not clarity, as the real carrier of information in organizations.They also discuss his role as a "double agent" between the art world and business, why he brings working artists into leadership coaching, and Bob Dylan going electric: Bring the spirit into a new instrument. And Joseph Beuys's 7,000 Oaks in Kassel, home of the documenta.

16. 

    8 Jul 2026

    Andy Tyler: what it means for AI to "get it"

    Andy Tyler joins Taste Bench to discuss what “taste” means in AI. He argues that taste is less about aesthetics and more about “getting it”: understanding intent, making good choices, and acting with agency. The episode covers AI’s progress in image generation, coding, writing, critique, and creative tools, while emphasising that AI is strongest when helping people explore ideas and make informed decisions.

17. 

    1 Jul 2026

    Joshua Tan: AI Sovereignty Is Not Enough

    In this episode, Bruno Marnette and Philipp Zahn speak with Joshua Tan, interim CTO of Current AI, about public-interest AI, open-source infrastructure, and the politics of building alternatives to the frontier labs. On the topics of taste: they explore whether centralized models risk making writing, code, and design more uniform; how local models might better reflect different cultures and communities; and why current AI systems are more useful as mirrors or thought partners than as replacements for human taste. They also discuss Current AI’s effort to develop a fully open-source alternative to ChatGPT and Claude, why the AI safety movement has struggled as a political movement, and what AI can learn from earlier debates around decentralization and crypto.

18. 

    24 Jun 2026

    Eleanor Warnock: AI Writing Is Glass

    In this episode of Taste-Bench, Bruno Marnette and Philipp Zahn speak with Eleanor Warnock, managing editor at Every, about AI, taste, and creative work.We discuss why AI-generated writing often feels brittle and how AI tools can help writers without taking over the parts of the process they care about. Eleanor also explores the gap between Silicon Valley and creative communities, the difference between taste and commercial quality, Japan’s alternative aesthetic traditions, and what Europe might contribute to AI beyond scale and infrastructure.The conversation covers AI writing, journalism, creative process, cultural bias, AI art, Europe’s possible AI edge, and what it would take for AI to unlock more human creativity.

Reach and audience

Public platform figures. Ratings count people who left a rating, not total listeners.

Spotify

5.0 / 5

1 ratings

Contact Taste-Bench

Guest appearances

Books guests

Based on episode analysis; this does not confirm that the show is currently accepting guests.