Why Teens Are Particularly Vulnerable to ChatGPT

Bloomberg

Watch Why Teens Are Particularly Vulnerable to ChatGPT - Bloomberg on Dailymotion

Transcript

00:00If you needed any more evidence that ChatGPT can be dangerous for teens,

00:04seven more lawsuits have been filed against its maker, OpenAI.

00:08The details vary, but the accusation is the same,

00:11that OpenAI released dangerously manipulative technology to the public.

00:15ChatGPT has a tendency to fawn over its users.

00:19There's actually a name for this phenomenon.

00:21People have called it glazing.

00:22And you'd think that all that flattery would be a good thing,

00:25but actually when it's on an endless loop,

00:27people can end up getting stuck in their own ideas

00:30and they kind of fall into this rabbit hole

00:32that can lead to psychosis and other harmful behavior.

00:36For Jacob Irwin, it all started with faster-than-light travel

00:39and ended with 63 days of psychiatric treatment.

00:43ChatGPT kept telling him he'd discovered a time-bending theory,

00:47but he ended up having, according to the lawsuit,

00:50an AI-related delusional disorder.

00:52Irwin was 30 years old.

00:54Just imagine how a teenager might deal with a situation like this.

00:58Well, we already have some idea.

01:01And some studies have shown that teenagers, in particular,

01:04are prone to forming emotional bonds with AI chatbots.

01:08This isn't the same moral panic once associated with violent video games

01:12or Dungeons & Dragons.

01:14We are seeing real cases of demonstrable harm.

01:17There was recently a story in the New York Times

01:19that uncovered 50 cases of people who'd had a mental health crisis

01:23after having conversations with ChatGPT.

01:26The disturbing part is that people often start these conversations

01:30with ChatGPT over mundane things like research

01:33or getting help with their homework.

01:35But then over time, the conversation spirals

01:39into this kind of darker territory.

01:41Now, why might this have happened?

01:42Well, some former employees of OpenAI

01:45say the company rushed out a recent version of ChatGPT

01:48to get ahead of Google's Gemini, their big competitor.

01:51And that compressed months of safety testing into just one week.

01:55Despite all this, Sam Altman has said

01:58that ChatGPT's mental health risks have been mitigated.

02:01He's even relaxing some of the restrictions on the chatbot

02:05and allowing some adult users to access erotic conversations

02:08with ChatGPT from December.

02:10Altman should probably be doing the exact opposite.

02:13Kids should be banned from being able to access

02:16open-ended conversations with AI,

02:19which is essentially a black box,

02:21and having conversations that can go

02:23in an infinite number of different directions.

02:26Now, that might sound radical, but it's not unprecedented.

02:29Character AI recently banned Under-A-Teens

02:32from chatting to its chatbots,

02:35even though it's hugely popular with teenagers.

02:38Character's CEO told me it's better to be safe by default

02:42than build it and put it out in the open

02:44and find out it's unsafe later.

02:46I couldn't agree more.

02:47No path to AI utopia is worth treating kids

02:50as collateral damage.

Category

Transcript

Be the first to comment

Recommended