Google DeepMind is calling for increased scrutiny of the moral behavior of large language models (LLMs), particularly as these technologies take on more sensitive roles in people's lives, according to a report by MIT Technology Review. The call for greater ethical oversight comes as LLMs are increasingly used as companions, therapists, and medical advisors, raising questions about their trustworthiness.
William Isaac, a research scientist at Google DeepMind, emphasized the need for rigorous evaluation of LLMs' moral behavior, similar to the scrutiny applied to their coding and math abilities. "With coding and math, you have clear-cut, correct answers that you can check," Isaac stated, highlighting the challenge of assessing the ethical dimensions of LLM interactions.
The push for ethical oversight of LLMs coincides with broader developments in the tech world. Jean-Paul Thorbjornsen, founder of the blockchain THORChain, is navigating the complexities of a permissionless cryptocurrency environment. Thorbjornsen, who previously used a pseudonym and an AI-generated avatar, is now openly operating in the crypto space. He claims a previous crypto project generated over AU$400 million, though these funds were tied up in the company, according to MIT Technology Review.
Meanwhile, the use of predictive algorithms continues to expand. As MIT Technology Review noted, algorithms are constantly attempting to anticipate user behavior, from predicting the next word typed to forecasting market trends. This pervasive use of predictive technology underscores the growing importance of understanding and controlling the future.
In other tech news, Metriport, a Y Combinator-funded startup, is seeking a Senior Security Engineer. The open-source platform focuses on healthcare data intelligence, integrating with major US healthcare IT systems to provide real-time access to patient data for 300 million individuals, according to Hacker News. The role, based in San Francisco, offers a salary range of $160,000 to $220,000.
AI Experts & Community
Be the first to comment