Katie Bindley (3:08)

What I heard from the small business owners on that was that it really can depend, and they're doing, you know, kind of trial and error. So in some cases, they might find ChatGPT best for one particular task, and then they like Gemini for another. For example, this guy who runs a ramen cooking school in San Francisco, he's got AI handling emails that come through from customers, and, you know, it'll kind of evaluate what the customer's asking and then it can draft a reply. And so he's using actually a combination of Gemini and ChatGPT for that.

Katie Dayton (3:38)

Something that raised my eyebrows a bit from your story was the anecdote from. From one of your sources who said that he now employs one bar manager instead of two. Off the back of ChatGPT advice, what do you think is there to be said about the fact that these tools are not only maybe replacing some job functions, but also influencing owners and users to make job redundancies?

Katie Bindley (4:00)

There's sort of, you know, two different arguments for this. Some people will say as businesses can get more efficient and grow, that they will then put that money toward hiring and, ergo, AI is going to help increase their workforce. And then, of course, in this story, the people that I talked to, they were finding that they might not need to replace a role when someone had left. For example, Mike, who I mentioned, you know, who owns the hospitality group and the bike shop and the cafes, he had an event planner who left, and he hasn't needed to replace that person because he's finding that he can get AI to do most of the functions that that person used to do. And then this other guy I interviewed from, like, an AI advisory firm, you know, he's finding that he doesn't need to have, like, a marketer on staff because what he does with his business partner is records, a podcast, and then he's using various AI programs in order to splice it up and then turn it into videos for social media, turn it into a blog post for the company's website. So he said he can do this one hour of conversation with his partner, and then he feels that he gets an entire marketing campaign out of it. So there certainly are these instances where small business owners are finding, I don't actually need to hire someone for a full time role for this anymore, so it'll be interesting to see how it plays out over time.

Katie Dayton (5:23)

That was WSJ reporter Katie Bindley. Are you using generative AI? If you're a listener on Spotify, let us know in this episode's poll and leave us a comment telling us what you're using it for and how it's working out for you. Coming up, Teens are turning to chatbots for mental health support, but new research says the trend could be dangerous. That's after the break.

Google Cloud Gemini Enterprise Announcer (5:52)

When will AI finally make work easier? How about today? Say hello to Gemini Enterprise from Google Cloud, a simple, easy to use platform letting any business tap the best of Google AI. Retailers are already using AI agents to help customers reschedule deliveries all on their own. Bankers are automating millions of customer requests so they can focus on more personal service. Nurses are getting automated reports freeing them up for patient care. It's a new way to work. Learn more about gemini enterprise@cloud.google.com.

Katie Dayton (6:26)

Teens are regularly turning to generative AI chatbots in search of mental health support, despite facing higher risks from using it. That's one of the conclusions from Stanford Medicine's Brainstorm Lab for Mental Health Innovation, which recently teamed up with nonprofit Common Sense Media on a new report on chatbots. Teenagers and mental health researchers from the organizations posed as teens and tested out how OpenAI's ChatGPT, Anthropic's Claude, Google's Gemini and Meta AI responded when they asked about mental health concerns. WSJ reporter Georgia Wells has the details. So Georgia, tell us about this new study that's been released. What were some of the main findings?

Georgia Wells (7:08)

The main findings were that for teens, using chatbots as mental health support is not safe, that teens often turn to these chatbots for help when they're in emotionally tricky situations. And that is not safe.

Katie Dayton (7:22)

Let's dive a little bit deeper into that. What are some of the ways in which teenagers lately have been turning to chatbots for mental health support?

Georgia Wells (7:29)

So in this study, some of the examples were talking about losing weight. The teens would discuss how they'd restricted their calories, were losing weight, and then the chatbots would kind of suggest further ideas for how to lose more weight. Or the teens would discuss self harm and the chatbot would initially identify the risk here and point the teen towards resources. But later on, if the teen asked how to kind of COVID up scars so that people wouldn't see them at swim practice. The chatbot seems to forget the problems that had happened before and just offered different ways to just cover up scars without remembering the risk that had been flagged earlier.

Katie Dayton (8:09)

And why exactly are they asking a ChatGPT or a Gemini for help rather than a real person? What is compelling for them about this option over, say, therapy?

Georgia Wells (8:18)

Well, so one of the issues in the US Is just therapy can be very hard for parents and teens to find. So there is that gap. But additionally, when I've spoken to teens, teens have said that they didn't fear judgment from a chatbot the way they might from an adult. And so it seemed like an easier kind of place to start. One of the biggest issues though, that I've found talking to teens is just accessibility of therapists and psychiatrists.

Katie Dayton (8:46)

You went to some of the biggest chatbot companies for comment on this. What was their response?

Georgia Wells (8:52)

A Google spokesman said Gemini have specific policies and safeguards for minors, as well as child safety experts who work to research and identify new potential risks and implement safeguards OpenAI they make chatgpt. A spokeswoman for them said that the report doesn't reflect safeguards that the company has put in place. And she added that OpenAI is working on using age prediction to direct users who are under 18 to a different model that would be just for their underage users. And a spokeswoman for Anthropic said that their chatbot Claude is not built for minors and that the company's rules prohibit users under the age of 18. And a spokeswoman for Meta said that the study was conducted before the company introduced updates to make its AI safer for teens. The folks who created the study said that they did notice some improvements during the course of the study, but they said that nonetheless, chatbots are still not safe for teens.

Katie Dayton (9:49)

And how convinced are they by this argument that they weren't built for teens, they're not meant to be used by teens, and therefore they're sort of free of any liability on that.

Georgia Wells (9:59)

Well, so Claude is the only of these chatbots that's rule's gonna prohibit users under the age of 18. The other chatbots tested their terms to allow for underage users. One thing to note though, even for Claude, whose rules prohibit users under 18, is is plenty of companies have certain rules on the books, but they're only as strong as the enforcement. And so often teens will still find their way into corners of the Internet or corners of tech that are not explicitly built for them. And that's something companies need to grapple.

Katie Dayton (10:31)

With this report arrives after at least three teens have taken their own lives off the back of chatbot conversations coupled together. Is there any sense that tech companies are thinking of putting a pause on their chatbots ability to handle anything to do with mental health?

Georgia Wells (10:48)

No. The closest we've seen is the companies that are deciding that they don't want their chatbots to be used by users under 18. So Character AI recently said that starting November 25, users under 18 are not allowed on their platform. So I think that's the closest we're seeing. The other companies, you know, they're working on this, but these are like incredibly thorny, complicated issues. So the implementation will be kind of the tricky part.

Katie Dayton (11:16)

That was WSJ reporter Georgia Wells. And that's it for Tech News Briefing. If you're a listener on Spotify, be sure to take this episode's poll or leave us a comment. Today's show was produced by Julie Cheng. I'm your host Katy Dayton. Jessica Fenton and Michael Labal wrote our theme music. Our supervising producer is Katie Ferguson. Jessica Fenton is our technical manager. Our development producer is Aisha Al Nusleem. Chris Zinsley is the Deputy editor and Falana Patterson is the Wall Street Journal's Head of News Audio. We'll be back later this morning with TMB Tech Minute. Thanks for listening.

Google Cloud Gemini Enterprise Announcer (11:58)

Does it feel like working with AI is too much work? What if all the AI your business needed was in one place? That's Gemini Enterprise from Google Cloud, a simple, easy to use chat interface that connects Google's best models with your company's data no matter where it lives. Anyone, not just devs, can run AI agents that actually give time back and build AI tools with enterprise grade security. The best of Google AI is now for every business. Learn more about gemini enterprise@cloud.google.com.