Professor to Advise OpenAI on Improving Well-Being Through ChatGPT
A Georgia Tech interactive computing professor has been selected to advise one of the largest artificial intelligence (AI) companies on making its agents safer and healthier to use.
OpenAI, the developer of the text-to-text generative artificial intelligence tool ChatGPT, announced in October the establishment of an Expert Council on Well-Being and AI.
Professor Munmun De Choudhury is one of eight experts serving on the initial council. OpenAI invited De Choudhury for her expertise on how digital resources affect youth mental health.
OpenAI has raised awareness of improved safety features in recent months after facing criticism for how it protects its users, particularly minors.
The Federal Trade Commission has probed some tech companies, including OpenAI, about the potential negative effects that chatbots like ChatGPT may have on children and teenagers.
The company is also facing a wrongful death lawsuit from a family claiming ChatGPT pushed their son to take his own life.
“The goal is to help OpenAI understand healthy interactions with AI as well as those that might be harmful,” De Choudhury said. “I’ll advise them with domain knowledge about how they should think about model behavior, how to protect users seeking help, and what safety protocols should be used in crisis-linked discourse.”
De Choudhury, who works in the School of Interactive Computing, said she will serve a six-month appointment on the council and then reassess before committing to further involvement.
“They acknowledged there are issues of harm that need to be addressed,” she said. “They seem open to feedback and want to do the right thing. I think it’s valuable that we as academics can help shape something that has become a global technology, and hopefully we can make these models safer.”
According to a press release on OpenAI’s website, the company will work with the council to explore:
- How AI should behave in complex or sensitive situations.
- Guardrails that can support ChatGPT users.
- How ChatGPT can contribute to well-being.
“Hopefully, more AI companies will feel the need to put together expert councils like this,” De Choudhury said. “That would help move the needle a lot in this space. It’s easy in academia to point out what is lacking, but it’s hard to achieve anything if you don’t have a way in.
“These technologies are impacting the lives of people in positive and negative ways. The sooner the companies acknowledge the harms and the need to act, the better for everyone.”
