circumstance
From teaching machines to teaching people: an interview with Michael Littman
Welcome back to AI Pioneers, our series of in-depth conversations with leading figures in artificial intelligence. In this interview, we speak with Michael Littman about the evolution of reinforcement learning, the role of creativity and culture in how humans and machines learn, and the challenges generative AI poses for education. We also explore his approach to science communication, his use of music to convey computer science concepts, and his unexpected connection to the first multi-media email attachment . You've been in the field of reinforcement learning for decades. How did you get into it, and which questions have you found most interesting across your career? I was drawn to reinforcement learning when I was a college student in the mid to late '80s. At that time, people were just starting to talk about neural networks, and the machine learning techniques being used were supervised, which is very focused on the input and output. I didn't find that so interesting - I wanted to know how to make them behave. So, I created my own reinforcement-learning system to address the game of noughts and crosses, or tic-tac-toe. It was a small enough game that I could run it on my little computer at home, but it made me address some interesting questions. For example, what does it mean to get feedback? And how do you use feedback at the end of the game to know what you should have done at the beginning of the game? I started to explore this topic without knowing that it was reinforcement learning. It wasn't until 1988, when I got out of college, that I discovered the link. I was working in a research lab with a researcher named David Ackley, who had done work on similar kinds of models. He used to be a student of Geoff Hinton, so he was very familiar with neural networks and very interested in behavior. I explained to him what I was playing with and why I found it so interesting, and he said "oh, that's called reinforcement learning, Rich Sutton just published a paper on this - let's get him to come and give a talk."
Disability campaigners call for Burnham to delay assisted dying bill
Disabled and terminally ill campaigners have written to Prime Minister Andy Burnham urging him to reform social and palliative care before assisted dying legislation is introduced in England and Wales. The Assist Us To Live group says people are currently being failed by society and introducing assisted dying in these circumstances will put sick and disabled people's lives at risk. Burnham believes the funding of palliative and social care should be fixed before assisted dying is considered. The government is likely to remain officially neutral, but the PM's remarks suggest he personally will not back the bill proposed by Labour MP Lauren Edwards. Bill supporters argue improving care services and giving terminally ill people autonomy over their final weeks must go hand in hand.
When Is Next-Token Prediction Useful? Marginalization, Ergodicity, Mixture Identifiability, Local Sufficiency, RAG, Tools, and Programming
Language models trained on observed sequences are often described as learning the conditional distribution of the next token given previous tokens. This description is only conditionally correct. A model trained on realized token trajectories does not observe full conditional laws; it receives sampled continuations. Moreover, real language generation is conditioned not only on previous words but also on non-textual circumstances: facts, events, intentions, goals, beliefs, social context, and task-specific constraints. This paper distinguishes three objects that are often conflated: the full conditional language process conditioned on latent circumstances, the marginal text-only process obtained by integrating those circumstances out, and the model-induced distribution learned from finite observed corpora. The paper argues that interpreting model training as estimating the marginal text-only law requires strong assumptions of stationarity, representativeness, and ergodicity, assumptions that are standard in statistical estimation but problematic when applied to heterogeneous language corpora. Even if these assumptions hold, the marginal text-only law is useful only when the observed prefix is an approximately sufficient statistic for the latent circumstances relevant to continuation. In information-theoretic terms, usefulness requires that the residual conditional mutual information between the next token and the omitted circumstances, given the observed text, be small. The paper then extends this argument to heterogeneous training corpora. Finally, the paper interprets Retrieval Augmented Generation (RAG) and tool use as conditional sufficiency devices.
Court challenge over Met Police's use of live facial recognition thrown out
Court challenge over Met Police's use of live facial recognition thrown out Privacy campaigners have lost a High Court challenge aimed at limiting the Metropolitan Police's use of live facial recognition technology. Youth worker Shaun Thompson, and Silkie Carlo, director of campaign group Big Brother Watch, brought the claim over concerns that facial recognition could be used arbitrarily or in a discriminatory way. Scotland Yard defended the challenge, telling the court that the policy was lawful. The Met Police will continue to use the technology, with commissioner Sir Mark Rowley calling the ruling an important victory for public safety. One of the claimants, Thompson, was misidentified by live facial recognition technology (LFR).
Me, Myself, and AI: The Situational Awareness Dataset (SAD) for LLMs
AI assistants such as ChatGPT are trained to respond to users by saying, I am a large language model".This raises questions. Do such models know'' that they are LLMs and reliably act on this knowledge? Are they aware of their current circumstances, such as being deployed to the public?We refer to a model's knowledge of itself and its circumstances as situational awareness
She Was Given Up by Her Chinese Parents--and Spent 14 Years Trying to Find a Way Back
More and more Chinese adoptees in the US are trying to reunite with their birth parents. For Youxue, it took more than a decade, and a remarkable coincidence. A girl is found on a street in Ma'Anshan, China, in May 1993. Her paternal grandfather, the story goes, set her down and walked away. It's unclear how long she's been outside when somebody arrives and takes her to the orphanage. A white woman adopts the girl and brings her to America in August 1994. She gives her an English name. In spring 2010, when Youxue (her Chinese name) was a high school sophomore in Dallas, Texas, she decided to start searching for her birth parents.
Self-Explaining Deviations for Coordination
Fully cooperative, partially observable multi-agent problems are ubiquitous in the real world. In this paper, we focus on a specific subclass of coordination problems in which humans are able to discover self-explaining deviations (SEDs). SEDs are actions that deviate from the common understanding of what reasonable behavior would be in normal circumstances. They are taken with the intention of causing another agent or other agents to realize, using theory of mind, that the circumstance must be abnormal. We motivate this idea with a real world example and formalize its definition. Next, we introduce an algorithm for improvement maximizing SEDs (IMPROVISED). Lastly, we evaluate IMPROVISED both in an illustrative toy setting and the popular benchmark setting Hanabi, where we show that it can produce so called finesse plays.
"Sirฤt" Is a Harrowing, Exhilarating Dance of Death
At one point, Luis assumes that he and Esteban have been abandoned, only to realize, with a start, that their newfound friends are actually circling back to help. In such moments, we grasp the source of the story's mysterious power: a tough-minded understanding that kindness is rare yet persistent, and quite possibly an affront to the laws of nature. "Sirฤt" is a chain of defiantly compassionate acts--noble human improbabilities that take on, in retrospect, an air of fatalistic inevitability. Laxe, a restless wanderer himself, knows Morocco well. He shot his first feature, "You All Are Captains" (2011), in Tangier, where he'd spent several years working at a shelter for disadvantaged children. Several of these children appeared in the movie--a formally playful collision of fiction and documentary in which Laxe, also making an appearance, slyly interrogated his European outsider-artist role. Next came "Mimosas" (2016), an elusive, arrestingly gorgeous drama about a caravan bearing a dying sheikh across Morocco's Atlas Mountains to his homeland. The film had the beauty of a travelogue and the opacity of a parable. Its most dynamic character was a fiery Muslim preacher who warned his fellow-travellers not to stray, geographically or morally.
"I Sweated So Much I Never Needed to Pee": Life in China's Relentless Gig Economy
In his newly translated memoir, Hu Anyan captures the brutal labor and quiet grace of life at the edge of China's booming ecommerce industry. "Often, sweat was dripping down my back within the first two hours of a shift and would not stop dripping until the next morning," writes Hu Anyan in the new English translation of his bestselling book . "I sweated so much I never once needed to pee." This passage was on my mind as I read his book in Tianjin during one hot, Labubu brainrot summer, during which yet another unprecedented annual heat wave had forced almost everyone inside--except for the tireless couriers and delivery workers, whose services are in higher demand when temperatures soar. Hu's writing first went viral in China five years ago, and he's now a prolific, established author in the country.
Soppia: A Structured Prompting Framework for the Proportional Assessment of Non-Pecuniary Damages in Personal Injury Cases
Applying complex legal rules characterized by multiple, heterogeneously weighted criteria presents a fundamental challenge in judicial decision-making, often hindering the consistent realization of legislative intent. This challenge is particularly evident in the quantification of non-pecuniary damages in personal injury cases. This paper introduces Soppia, a structured prompting framework designed to assist legal professionals in navigating this complexity. By leveraging advanced AI, the system ensures a comprehensive and balanced analysis of all stipulated criteria, fulfilling the legislator's intent that compensation be determined through a holistic assessment of each case. Using the twelve criteria for non-pecuniary damages established in the Brazilian CLT (Art. 223-G) as a case study, we demonstrate how Soppia (System for Ordered Proportional and Pondered Intelligent Assessment) operationalizes nuanced legal commands into a practical, replicable, and transparent methodology. The framework enhances consistency and predictability while providing a versatile and explainable tool adaptable across multi-criteria legal contexts, bridging normative interpretation and computational reasoning toward auditable legal AI.