Tote Bag Productions
110 episodes listed below
PRESS PLAY
Talking Machines is your window into the world of machine learning. Your hosts, Katherine Gorman and Neil Lawrence, bring you clear conversations with experts in the field, insightful discussions of industry news, and useful answers to your questions. Machine learning is changing the questions we can ask of the world around us, here we explore how to ask the best questions and what to do with the answers. Hosted on Acast. See acast.com/privacy for more information.
Based on 6 episodes we listened to · September 2026
Talking Machines connects machine learning's technical debates to their consequences for research, work, institutions, and public decision-making.
Talking Machines treats machine learning as both a technical discipline and a force acting on institutions, work, and public life. Katherine Gorman and Neil Lawrence typically begin by testing an idea together, then widen the conversation through an interview.
AI meets labor economics; your future-of-work hot take could use a boulder of salt.
Press play if you want
Short reviews from the PlayNext desk, based on the episodes we processed.
A grounded debate on open access, journal costs, and the medical image data behind machine learning.
Talking Machines places a debate over machine-learning publishing next to a quieter but equally practical question: what medical image data allows researchers to see.
A thoughtful episode on AI language, research culture, and the systems around the field.
Talking Machines looks at the language people use around AI, with Michael Jordan's distinction between artificial intelligence and intelligence infrastructure giving the episode its center.
A clear bridge from European machine learning policy to who gets heard as work changes.
Economies, Work and AI links machine learning policy to the larger question of how technology reshapes work and who gets a voice in that process.
Explainability becomes a matter of method, judgment, and who gets heard.
Talking Machines turns a technical question about making machine-learning systems intelligible into a careful discussion of method, judgment, and trust.
A GDPR setup gives way to Andrew's warm account of lasers, MIT, and an unexpected path toward AI.
Talking Machines begins with Katherine Gorman and Neil Lawrence framing GDPR around the movement of personal data outside the EU and the protections being built around it.
A grounded AI conversation that expands into work, inequality, and prediction.
Talking Machines finds Katherine Gorman and Neil Lawrence working through the distance between natural and artificial intelligence, starting with Gorman's memory of a TEDxExeter talk as Mind and Machine.
A clear, cross-disciplinary look at explainability, law, and black box decisions.
Talking Machines spends this episode in the awkward space between machine learning and legal accountability. Neil Lawrence and Katherine Gorman discuss counterfactual explanations under the GDPR, and the useful tension is that the model can stay opaque while still having to…
Markets, macro, AI, and crypto in motion.
Daily AI news through the lens of trust, infrastructure, and real-world deployment.
The hosts discuss the importance of reproducibility in machine learning, focusing on recent initiatives at conferences like NRIPS to incorporate code submission and other measures to enhance research transparency.
This episode features an interview with Madhulika Shrikumar from the Partnership on AI, discussing her background in law and technology regulation, focusing on challenges like cross-border data access in criminal cases and responsible AI publication.
This episode discusses the International Conference on Learning Representations (ICLR), its growth from a workshop to a major virtual conference, and its significance in the machine learning ecosystem.
The hosts discuss technical challenges in managing the large-scale ICML conference proceedings for PMLR, including GitHub timeouts and surname disambiguation issues.
This episode features PhD student Devin Guillory discussing his journey into robotics and computer vision research, his industry experience, and his current work on learning with less labels at UC Berkeley.
This episode of Talking Machines features Michael Lippman discussing his work with the Humanity-Centered Robotics Initiative at Brown University and his experiences as a communications chair for NeurIPS.
The hosts celebrate their 100th episode while discussing the COVID-19 pandemic's impact in the UK and Italy, noting varying levels of public concern and social distancing behaviors.
This episode features an interview with Dr.
This episode of Talking Machines features hosts Katherine Gorman and Michael Littman interviewing Neil and Rabbi Laura Jenner-Klausner about their backgrounds in machine learning and interfaith dialogue.
A physician shares a personal story from the emergency room about a dying patient to introduce the ethical implications of using machine learning to predict mortality in healthcare.
The hosts discuss their first live broadcast episode featuring an interview with Elaine Nsoisi from Boston University about modeling viruses and public health during social distancing.
This episode of Talking Machines explores Gaussian processes as a Bayesian framework for modeling distributions over random functions in machine learning.
This episode discusses a best paper award-winning study at ICML that examines the delayed impact of fair machine learning, emphasizing the need for longitudinal analysis to truly assess fairness in algorithms like loan awarding systems.
This segment introduces a TEDx Boston talk by Hugo La Rochelle on foundational deep learning concepts, shared in preparation for the ICLR conference.
This segment introduces the concept of data trusts, tracing its origins to Lillian Edwards' 2004 paper and discussing how it aims to realign incentives between data holders and subjects through a feudal ecosystem analogy.
Hosts Catherine Gorman and Neil Lawrence discuss the potential name change of the NIPS conference, addressing concerns over racial slurs and anatomical references that affect diversity.
The hosts discuss the AISTATS conference in Japan, highlighting a paper on Fisher Information and Natural Gradient Learning in Random Deep Networks by Amari and colleagues.
Neil Lawrence discusses his experience at Data Science Africa in Addis Ababa, the expansion to two annual meetings including events in West Africa, and challenges in scaling such conferences while considering local economic impacts.
This segment discusses the upcoming ICLR conference in Addis Ababa in 2020, focusing on its significance for Africa and what international attendees might learn from the experience.
Neil Lawrence discusses a recent paper by Edwin Fong and Chris Holmes on marginal likelihood and cross-validation within the MOPEN framework for Bayesian modeling, highlighting the challenges of probabilistic models in capturing real-world complexity.
The hosts discuss the announcement of NeurIPS 2019 workshops, highlighting key insights from Jen Wartman-Vaughn's blog post on diversity, innovation, and application trends.
Professor Barbara Englehart discusses using machine learning as a detective-like tool to solve mysteries in genomics, biology, and medicine, introducing Leo Breiman's concept of the two cultures in statistical modeling.
Neil Lawrence discusses the 'great AI fallacy'—the misconception that AI will be the first wave of automation to adapt to humans rather than requiring humans to adapt to it, contrasting it with past automation like factories and computers.
Neil Lawrence discusses his experience at the Mars conference, highlighting Vivian Say's talk on energy-efficient hardware implementations for machine learning that optimize processor layouts based on algorithm data access patterns.
Neil Lawrence discusses his experience at the Stu Hunter Conference, highlighting its unique format focused on industrial statistics and debates around statistics versus analytics and machine learning.
This episode discusses the origins and uses of Jupyter Notebooks, tracing their development from the IPython project and comparing them to tools like Mathematica for data analysis and machine learning.
This episode explores the intersection of machine learning with real-world causality, discussing challenges like counterfactual reasoning and the limitations of randomized studies in fields like personalized medicine.
This episode discusses the 3D process for machine learning systems design, emphasizing how algorithms are composed into useful systems and the importance of software engineering and transparency.
The hosts discuss Tom Griffiths' question about conducting research in an era of hype and reflect on changes in the academic community.
This episode discusses a critical paper titled 'Troubling Trends in Machine Learning Scholarship' that analyzes problematic patterns in how machine learning research is conducted and presented.
The hosts discuss Rich Sutton's blog post 'The Bitter Lesson', exploring how general-purpose methods like search and learning scale with computation and data in machine learning.
This episode explores the use of simulation in machine learning, particularly reinforcement learning, and discusses its potential limitations and ethical implications in real-world applications.
The hosts discuss Mike Tipping's request for five key machine learning papers from recent years, reflecting on his career and the challenge of selecting papers in today's rapidly evolving field.
Sela Novo discusses her journey from early volunteering experiences to realizing she could use her mathematical and computer science skills to create positive impact at scale in her career at Google Research.
Neil Lawrence recounts his experience speaking at an Amazon Machine Learning Conference in 2016, where Jeff Bezos was unexpectedly in the audience, leading to an amusing anecdote about the paradox of Bezos's unassuming presence.
This episode of Talking Machines discusses the book 'Bit by Bit' by Matthew Selgenick, exploring how it bridges machine learning and social science for research and ethical applications.
This episode features a TEDx talk by Harvard Professor Finale Doshi-Velez exploring the possibility of explanation in artificial intelligence, particularly in healthcare applications.
This episode explores the concept of distributed versus centralized intelligence through examples like swarm robots and the human immune system, drawing from a Royal Society panel discussion on AI.
This episode discusses the importance of moving beyond mere publication in machine learning research to focus on actionable insights, data curation, and answering the 'so what' factor for real-world impact.
The hosts discuss reflecting on the past decade's machine learning advancements and predicting how work methodologies will evolve in the coming decade.
Neil Lawrence discusses his upcoming debate with IBM's Project Debater at the Cambridge Union, exploring how the AI debating system works and the context of parliamentary-style debates.
This episode features a TEDxBoston talk by law professor Kenneth Anderson discussing the need for legal frameworks to regulate AI and robotics, challenging simplistic solutions like Asimov's Three Laws.
Michael Littman uses anecdotes about his children's different learning styles to illustrate contrasting approaches to developing intelligent systems in artificial intelligence.
This episode features an interview with Jennifer Dy, a professor at Northeastern and program chair of ICML 2018, discussing her career path from the Philippines to machine learning research.
This episode of Talking Machines explores AI ethics with a paper on counterfactual explanations under GDPR, discusses the stochastic nature of the NIPS reviewing process, and features an interview with Sven Strohband on practical AI startup strategies in robotics and agriculture.
Talking Machines examines the announced Nature Machine Intelligence journal through Neil Lawrence's criticism of subscription publishing and Maithra Phagu's explanation of neural-network interpretability.
Katherine Gorman and Neil Lawrence open with Michael Jordan’s Medium post "artificial intelligence, the revolution hasn’t happened yet" and Rodney Brooks’s essay, using an ultrasound false-positive story and the UK Q risk score to argue that AI needs intelligence infrastructure.
Talking Machines hosts Katherine Gorman and Neil Lawrence examine ELIS, the European Lab for Learning and Intelligent Systems, and the UK artificial intelligence sector deal as competing approaches to machine-learning investment.
Neil Lawrence connects AI debates about singularity, immortality, and omnipotence with religious imagery, then shifts attention toward privacy, freedom, and human responsibility.
Talking Machines examines GDPR as a formal attempt to regulate personal-data use, including its 1995 directive predecessor, 1998 drafting history, worldwide-revenue fines, and unresolved meaning of algorithmic explanations.
Peter Donnelly says AI is more likely to enable jobs than replace them, but the job mix may still change.
"100 bits" versus "a gigabit" becomes Neil Lawrence's case for an "embodiment factor" in intelligence.
Ali Rahimi said machine learning is "alchemy," and Ryan framed the real question as "what do we actually do about it?"
Katherine Gorman and Neil Lawrence open Talking Machines season four by discussing a December 13 blog post from Christiane Lum after NIPS in Long Beach.