Cassidy Diaz
News Editor
The California Legislature has voted to implement Senate Bill 903, would ban artificial intelligence chatbots that provide mental health therapy services and will place regulations when AI is used in health spaces.
The bill, which now awaits approval by Gov. Gavin Newsom in the next month or so, was authored by Sen. Steve Padilla, D-San Diego.
Padilla said in a recent press release that as AI starts to grow in our world and trickle into the mental health spaces, we must place regulations that protect those who use AI for support.
“SB 903 draws a clear line: AI can be a tool in the hands of licensed professionals, but it cannot be the professional itself,” Padilla said in the press release.
SB 903 would assure that mental health services will not allow any use of AI algorithms to provide and advertise AI chatbots as a form of therapy.
The bill would also ensure that mental health services are only performed carefully and ethically by a licensed professional and will prohibit consumers and companies from advertising and offering AI therapy chatbots.
The bill would permit licensed professionals to use AI only for limited administrative or additional support such as transcribing therapy sessions with full transparent disclosure and consent.
But AI would be prohibited from interacting with patients, making therapeutic decisions and generating designed treatments under the pending law.
Andrew Guydish, assistant professor of psychology, who studies conversations and how people interact with each other, explained the importance of having interactions with each other that AI cannot pick up on.
“Things like fluctuations in tones, raising kinds of tones, as well as body language and hand movements help us to communicate more efficiently,” Guydish said.
Conversing is a skill and a part of functioning in life, AI can diminish those communication skills and social interactions, Guydish said.
“When you simply communicate with artificial intelligence or rely on and spend a lot of time communicating with artificial intelligence rather than with others, you lose out on those opportunities and you lose out on that practice,” Guydish said.
AI “therapists” can cause serious problems because of the bots’ tendency to validate behavior in conversation, said Aaron Baker, associate professor of psychology.
“Validating potentially dangerous behavior is probably my biggest concern because those are the things we tend to be most worried about from an ethical standpoint as providers,” Baker said.
Baker added, however, that there is likely a future in which AI will function in the mental health environment because it is a relatively inexpensive way to direct one-on-one intervention.
“These tech companies are not particularly trained or concerned in mental health,” said Baker, who also has a private therapy practice. “I don’t think tech companies have actually put a lot of work into making sure it’s safe, and until they do it’s really dangerous to employ (AI).”
“The language of this bill is good because it starts from the principle that no one should undergo psychological counseling without the supervision of a trained professional,” Jason Neidleman, professor of political science, said. “It puts safeguards on the areas where AI could be most dangerous.”
“As a species, not just California, we need to be on guard against the threat that AI poses to our humanity,” Neidleman added.
Cassidy Diaz can be reached at cassidy.diazgomez@laverne.edu.
Cassidy Diaz is a senior journalism major with a concentration in print and online journalism. She is also news editor for the Campus Times, and previously served as an assistant editor and a staff writer.
