Imagine a coffee company trying to optimize its supply chain. The company sources beans from three suppliers, roasts them at two facilities into either dark or light coffee, and then ships the roasted coffee to three retail locations. The suppliers have different fixed capacity, and roasting costs and shipping costs vary from place to place. Wouldn't it be easier for the company to just ask ChatGPT to come up with an optimal plan' In fact, for all their incredible capabilities, large language models (LLMs) often perform poorly when tasked with directly solving such complicated planning problems on their own. Rather than trying to change the model to make an LLM a better planner, MIT researchers took a different approach. They introduced a framework that guides an LLM to break down the problem like a human would, and then automatically solve it using a powerful software tool. A user only needs to describe the problem in natural language ' no task-specific examples are needed to train or prompt the LLM. The model encodes a user's text prompt into a format that can be unraveled by an optimization solver designed to efficiently crack extremely tough planning challenges....

Pattie Maes, the Germeshausen Professor of Media Arts and Sciences at MIT and head of the Fluid Interfaces research group within the MIT Media Lab, has been awarded the 2025 ACM SIGCHI Lifetime Research Award. She will accept the award at CHI 2025 in Yokohama, Japan this April. The Lifetime Research Award is given to individuals whose research in human-computer interaction (HCI) is considered both fundamental and influential to the field. Recipients are selected based on their cumulative contributions, influence on the work of others, new research developments, and being an active participant in the Association for Computing Machinery's Special Interest Group on Computer-Human Interaction (ACM SIGCHI) community. Her nomination recognizes her advocacy to place human agency at the center of HCI and artificial intelligence research. Rather than AI replacing human capabilities, Maes has advocated for ways in which human capabilities can be supported or enhanced by the integration of AI. Pioneering the concept of software agents in the 1990s, Maes' work has always been situated at the intersection of human-computer interaction and artificial intelligence and has helped lay the foundations for today's online experience. Her article 'Social information filtering: algorithms for automating 'word of mouth'' from CHI 95, co-authored with graduate student Upendra Shardanand, is the second-most-cited paper from ACM SIGCHI....

You can't approach AI ethics with only formal procedures, say leaders at global retailer H&M Group. So to help build its collective moral compass, the company has built a culture of AI ethics based on experimentation. Its AI ethics training emphasizes concrete business examples and principles that teach people what to ask. H&M Group's example shows that organizations can practice AI ethics and become better at it despite not knowing what 'best' looks like. Artificial intelligence changes how organizations work, and that's one of the reasons why it challenges our ethics: Who should take responsibility for automated decisions and actions' How much agency should algorithms have' How should we organize interactions between minds and machines' How does technology affect the workforce' Where are biases built into the system' Companies, regulators, and policy makers search for steadfast ethical principles to help them navigate these moral mazes. They follow what feels like a logical strategy: First, identify universal values (such as transparency, fairness, human autonomy, or explainability), then define applications of the values (such as decision-making or AI-supported recruitment processes), and, finally, formalize them in codes of conduct. The idea is that codes of conduct for AI should override the computational code of AI....

Researchers enhanced the device ' known as a brain'computer interface (BCI) ' with artificial intelligence (AI) algorithms that decoded sentences as the woman thought of them, and then spoke them out loud using a synthetic voice. Unlike previous efforts, which could produce sounds only after users finished an entire sentence, the current approach can simultaneously detect words and turn them into speech within three seconds. Older speech-generating BCIs are similar to 'a WhatsApp conversation', says Christian Herff, a computational neuroscientist at Maastricht University, the Netherlands, who was not involved with the work. 'I write a sentence, you write a sentence and you need some time to write a sentence again' It just doesn't flow like a normal conversation.' The study participant, Ann, lost her ability to speak after a stroke in her brainstem in 2005. Some 18 years later, she underwent a surgery to place a paper-thin rectangle containing 253 electrodes on the surface on her brain cortex. The implant can record the combined activity of thousands of neurons at the same time....