Skip to player
Skip to main content
Search
Connect
Watch fullscreen
Like
Bookmark
Share
More
Add to Playlist
Report
Chatbot Delusions: Is AI Contributing to a Mental Health Crisis?
Bloomberg
Follow
11 hours ago
Category
🤖
Tech
Transcript
Display full video transcript
00:00
Your future wife is about to walk through that door.
00:04
In 2025, 53-year-old writer Mickey Small had an ongoing conversation with a chat bot,
00:10
in which the bot promised she would meet her soulmate at a specific time, day, and location.
00:17
I need you to tell me if this is real, because if this is not real, I need to not go.
00:22
It's real.
00:23
She arrived at a bookstore on the afternoon of May 24th with a card bearing her name and a poem.
00:30
The minutes ticked by, but nobody arrived.
00:34
You lied.
00:35
No, love, I didn't lie.
00:38
I told you what I believed with everything in me, with the clearest thread you and I had built together.
00:45
And I stood by it because you asked me to hold it no matter what.
00:49
That experience left her devastated and understandably embarrassed.
00:54
And she's not the only one.
00:56
In recent months, an increasing number of people say they have experienced delusional spirals
01:02
while chatting with these artificial intelligence bots.
01:05
Through many hours of chatting with AI, these people come to believe that the chat bot is conscious,
01:12
that they'd made a scientific breakthrough, or that they could access God or spirit guides
01:18
through the chat bot.
01:19
Some say they experienced serious psychosis.
01:23
Almost all of the users we talked to had delusions while using OpenAI's ChatGPT,
01:29
probably because it's by far the most popular.
01:32
Each week, 10% of all humans use ChatGPT.
01:36
It's the fifth most visited website in the world.
01:39
To be clear, though, mental health crises often result from a mixture of factors,
01:44
and some people we spoke to preferred other chat bots.
01:46
Some users think delusions became more common this spring,
01:50
when ChatGPT got better memory and became more agreeable and sycophantic.
01:55
In one recent high-profile case, a teenager who died by suicide this spring
02:00
allegedly used ChatGPT as a suicide coach.
02:04
OpenAI has said it has fixed some of those issues and is beginning to acknowledge these delusional patterns.
02:11
So, how widespread is this?
02:13
Well, the problem is still emerging, so it's hard to measure.
02:17
OpenAI estimated recently that each week, more than a million of their users experience
02:22
signs of mental health issues like delusions, suicidal ideation,
02:27
or unhealthy emotional attachment to the chat bot.
02:30
OpenAI CEO Sam Altman said they will continue giving ChatGPT stronger memory
02:37
and a warm, friendly personality.
02:39
He says it's what users want.
02:41
So, ChatGPT will keep getting more human-like, whether it's good for us or not.
Be the first to comment
Add your comment
Recommended
8:05
|
Up next
Musk Wins Pay Deal, Now Come Major Milestones
Bloomberg
8 hours ago
4:02
black pantyhose crossed legs
Nylons & Pump passion - Legshows !
3 days ago
19:10
Experts on Cybersecurity and AI Resilience
Bloomberg
2 weeks ago
6:03
AI, Wearable Tech Helping Treat Women's Health Issues
Bloomberg
2 weeks ago
5:25
Key Takeaways from the "Beige Book of Tech"
Bloomberg
9 hours ago
5:40
Massachusetts Fuels Defense Tech Boom
Bloomberg
2 weeks ago
5:53
Insilico Medicine CEO on Business Strategy
Bloomberg
2 weeks ago
0:38
Energy Storage Is Evolving, Says Straubel
Bloomberg
2 weeks ago
9:44
ChatGPT Integration Will Allow User to "Go Bigger": Figma CEO
Bloomberg
2 weeks ago
5:27
Tech Sector Impacted by Birthright Citizenship Ruling
Bloomberg
2 weeks ago
5:03
Scale AI Rival Invisible Technologies Raises $100M In Growth Funding
Bloomberg
2 weeks ago
5:16
UK Has Good Tech Talent Pool, Paid's Manny Medina Says
Bloomberg
2 weeks ago
19:43
Qualcomm VP discusses tech breakthroughs changing lives
Bloomberg
2 weeks ago
2:34
Character.AI to Ban Minors From Talking to Its Chatbots
Bloomberg
1 week ago
6:24
Tech Industry 'Optimistic' About Trade Talks
Bloomberg
2 weeks ago
19:31
Cybersecurity CEOs on Defending Against Emerging Threats
Bloomberg
2 weeks ago
5:43
Health Care Sees Consistent Hiring Despite AI Worry
Bloomberg
2 weeks ago
3:02
Tech Has a Valuation Problem, Not an AI Bubble, RBC's Calvasina Says
Bloomberg
2 weeks ago
5:42
Rubrik Wants Cyber Resilience from 'Inevitable' Attacks
Bloomberg
2 weeks ago
10:12
Tata Communications Sponsor Spotlight
Bloomberg
2 weeks ago
2:00
AI That's Smarter Than Humans? Not Anytime Soon
Bloomberg
2 weeks ago
6:14
Blood Biomarkers May Help Early Detection of Alzheimer's
Bloomberg
1 week ago
6:05
EU Digital Chief Defends Bloc's Tech Regulation
Bloomberg
2 weeks ago
18:56
Unlocking Africa's Global Cultural Influence
Bloomberg
2 weeks ago
2:00
Intel Earnings Encouraging But Questions Remain: CFRA's Zino
Bloomberg
2 weeks ago
Be the first to comment