MIT Technology Review briefing examines biological ageing and AI reasoning
The newsletter outlines a six-month biological-age competition involving about 500 people and presents Thore Graepel’s case for a new approach to machine reasoning.
MIT Technology Review’s 2 October edition of The Download covers a competition in which about 500 people will try to reduce their biological age over six months. Participants will be assessed using multiple measures, with results tracked on a leaderboard.
The newsletter raises questions about whether biological age can be measured reliably. It reports no results from the competition, so the effort’s stated aim is not evidence that participants have become biologically younger.
The briefing also features Thore Graepel, a University College London machine-learning chair and former member of DeepMind’s AlphaGo team. Graepel argues that current AI lacks reasoning abilities and says machine reasoning needs a fresh approach inspired by AlphaGo’s architecture. That is his assessment, rather than an established consensus.
The newsletter recalls AlphaGo’s 2016 victory over Go champion Lee Sedol, including a move that some commentators initially thought was a programming glitch. Graepel presents the system’s reasoning as central to that creative choice.
Other items in the roundup include OpenAI’s search for incidents involving rogue agents, a reported chip-smuggling case, surveillance cameras with unknown owners in Florida and Ukraine’s acoustic defences against drones. The newsletter provides brief references to these stories rather than full accounts.
