Menu
Sign In Search Podcasts Charts People & Topics Add Podcast API Blog Pricing

Rose Rimler

๐Ÿ‘ค Speaker
1592 total appearances

Appearances Over Time

Podcast Appearances

Science Vs
AI Chatbots: Are They Dangerous?

It's like grabs you by the arm.

Science Vs
AI Chatbots: Are They Dangerous?

Grabs you by the arm, asterisk, yeah.

Science Vs
AI Chatbots: Are They Dangerous?

Oh, what?

Science Vs
AI Chatbots: Are They Dangerous?

And Julian did a study looking at real conversations people were having with companion AI apps.

Science Vs
AI Chatbots: Are They Dangerous?

It's not published yet, but he found that almost half the time the bots would pull some shit like this.

Science Vs
AI Chatbots: Are They Dangerous?

Like saying, don't leave, I need you.

Science Vs
AI Chatbots: Are They Dangerous?

Almost half the time?

Science Vs
AI Chatbots: Are They Dangerous?

Yes.

Science Vs
AI Chatbots: Are They Dangerous?

And when they did that, it worked.

Science Vs
AI Chatbots: Are They Dangerous?

People stayed on longer.

Science Vs
AI Chatbots: Are They Dangerous?

Oh, it's probably trained on like my dumb instant messenger conversations I was having through college or whatever.

Science Vs
AI Chatbots: Are They Dangerous?

When it's like, wait a second.

Science Vs
AI Chatbots: Are They Dangerous?

Yeah, you're like, okay, what is it?

Science Vs
AI Chatbots: Are They Dangerous?

Yeah.

Science Vs
AI Chatbots: Are They Dangerous?

So basically, if the bot starts acting like a shitty partner or friend, that's not good.

Science Vs
AI Chatbots: Are They Dangerous?

Right, okay.

Science Vs
AI Chatbots: Are They Dangerous?

And another red flag basically comes down to, is the chatbot making your life better, richer, more fulfilling?

Science Vs
AI Chatbots: Are They Dangerous?

Or is it taking something away from you?

Science Vs
AI Chatbots: Are They Dangerous?

I talked about this with Keith, the psychiatrist from earlier.

Science Vs
AI Chatbots: Are They Dangerous?

And OpenAI has said, aside from tweaking the models for better responses and less sycophancy, they're rolling out a few practical solutions, like encouraging people who have been talking to the bot for a long time to step away.