Enlarge (credit: Aurich Lawson | Getty Images)

As far back as 2016, work on AI-based chatbots revealed that they have a disturbing tendency to reflect some of the worst biases of the society that trained them. But as large language models have become ever larger and subjected to more sophisticated training, a lot of that problematic behavior has been ironed out. For example, I asked the current iteration of ChatGPT for five words it associated with African Americans, and it responded with things like "resilience" and "creativity."

But a lot of research has turned up examples where implicit biases can persist in people long after outward behavior has changed. So some researchers decided to test whether the same might be true of LLMs. And was it ever.

By interacting with a series of LLMs using examples of the African American English sociolect, they found that the AI's had an extremely negative view of its speakers—something that wasn't true of speakers of another American English variant. And that bias bled over into decisions the LLMs were asked to make about those who use African American English.

Enlarge / MDMA pills up close. (credit: Getty | Universal History Archive)

After a remarkably sensational regulatory drug review, advisers for the Food and Drug Administration on Tuesday voted overwhelmingly against approving midomafetamine (MDMA, aka ecstasy or molly) as part of therapy for post-traumatic stress disorder.

In a response to the question, "Do the available data show that the drug is effective in patients with posttraumatic stress disorder?" advisers voted 9 "no" to 2 "yes" against MDMA.

The votes come after a full-day meeting in which the independent expert advisers heard detailed reviews of clinical trial data from the company that proposed the therapy—Lykos, formerly MAPS—as well as internal FDA reviews.

Enlarge / MDMA is now in the FDA's hands. (credit: Getty | PYMCA/Avalon)

The safety and efficacy data on the use of MDMA (aka ecstasy) for post-traumatic stress disorder therapy is "challenging to interpret," the Food and Drug Administration said in a briefing document posted Friday. The agency noted significant flaws in the design of the underlying clinical trials as well as safety concerns for the drug, particularly cardiovascular harms.

On Tuesday, June 4, the FDA will convene an advisory committee that will review the evidence and vote on MDMA's efficacy and whether its benefits outweigh its risks. The FDA does not have to follow the committee's recommendations, but it often does. If the FDA subsequently approves MDMA as part of treatment for PTSD, it would mark a significant shift in the federal government's stance on MDMA, as well as psychedelics, generally. Currently, the US Drug Enforcement Administration considers MDMA a Schedule I drug, defined as one with "no currently accepted medical use and a high potential for abuse." It would also offer a new treatment option for patients with PTSD, a disabling psychiatric condition with few treatment options currently.

As Ars has reported previously, the submission of MDMA for approval is based on two clinical trials. The first trial, published in Nature Medicine in 2021, involved 90 participants with moderate PTSD and found that MDMA-assisted psychotherapy significantly improved Clinician-Administered PTSD Scale for DSM-5 (CAPS-5) scores compared with participants who were given psychotherapy along with a placebo. In the second study, published in September in Nature Medicine, the finding held up among 104 participants with moderate or severe PTSD (73 percent had severe PTSD).