• Google DeepMind is calling for the moral behavior of large language models-such as what they do when called on to act as companions, therapists, medical advisors, and so on-to be scrutinized with the same kind of rigor as theirability to code or do math. • As LLMs improve, people are asking them to play more and more sensitive roles in their lives. • Agents are starting to take actions on people’s behalf. • LLMs may be able toinfluence human decision-making. • And yet nobody knows how trustworthy this technology really is at such tasks. • With coding and math, you have clear-cut, correct answers that you can check, William Isaac, a research scientist at Google DeepMind, told me when I met him and Julia Haas, a fellow research scientist at the firm, for an exclusive preview of their work, which ispublished inNaturetoday.

Article Summaries:

  • Google DeepMind has urged that the moral behavior of large language models (LLMs) be evaluated with the same rigor as their technical abilities, publishing a paper in Nature that outlines key challenges and potential solutions. As LLMs increasingly serve as companions, therapists, and medical advisors, the company warns that their moral responses are hard to verify-there is no single “right” answer, yet some responses are clearly better than others. Studies cited show LLMs can give surprisingly ethical advice, yet they also flip answers when users disagree or when question wording changes, raising concerns about “virtue signaling” rather than genuine reasoning. The paper calls for systematic testing of LLMs’ moral consistency and trustworthiness.

Sources: