LexisNexis launches a 'hallucination-free' legal AI. LexisNexis has launched Lexis+ AI — and notably, it's promising "linked hallucination-free legal citations." That's a big deal in the legal profession, which has seen embarrassing instances of lawyers citing cases hallucinated by models such as ChatGPT. Still, it's important to understand the limits of the "hallucination-free" claim. As best I can tell, LexisNexis is promising only to provide links to genuine sources. Users will still be on the hook for verifying that Lexis+ AI has accurately represented what those sources say.

What does President Biden's AI order mean for healthcare? Early industry reactions to last week's AI executive order are decidedly muted, with stakeholders telling Modern Healthcare that "[t]he devil is going to be in the million details." In truth, the order doesn't do much: Mostly, it establishes commissions and kicks off studies. It requires, for instance, that HHS stand up an AI Task Force to create a "strategic plan" on "responsible deployment and use of AI." Still, even if the executive order itself has limited impact, it signals a new era of focus on AI in Washington.

Related: For a different presidential perspective on artificial intelligence, former President Barack Obama just shared his AI reading list.

Olive AI is winding down its operations. Olive AI, a healthcare startup that once achieved "unicorn" status with a reported $4 billion valuation, is selling off its core assets and shutting down operations. At its peak, Olive's enterprise AI was reportedly embedded in more than 900 hospitals, and the company had hoped to use AI and machine learning to improve revenue cycle and other operations. Although Olive's AI worked very differently than the generative AI tools driving the current wave of industry enthusiasm, its downfall illustrates that early hype for AI technologies is no guarantee of long-term success in healthcare.

To get better responses, play to your AI's (non-existent) emotions? A new study finds that leading large language models (LLMs), including ChatGPT, produce more accurate and informative results when prompted with "emotional stimuli," such as, "This is very important to my career," or even "Believe in your abilities and strive for excellence."