Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

MacRumors

macrumors bot
Original poster
Apr 12, 2001
68,714
39,655


OpenAI is making several changes to the way that ChatGPT works, with the aim of helping people use the chatbot in a healthier way.

chatgpt-logo.jpg

Starting today, ChatGPT will provide users with "gentle reminders" about how long they've been using the service, with suggestions for breaks. OpenAI says that it isn't measuring success by time spent on ChatGPT, but by ensuring that people leave the product after finishing what they came for.
We build ChatGPT to help you thrive in all the ways you want. To make progress, learn something new, or solve a problem -- and then get back to your life. Our goal isn't to hold your attention, but to help you use it well.
OpenAI is working to improve ChatGPT's responses to certain kinds of questions. Rather than providing a direct answer to a high-stakes, sensitive question, ChatGPT should help users weigh pros and cons and think it through. OpenAI is updating ChatGPT's behavior for important personal decisions soon, so users will see more helpful responses for questions like "Should I break up with my boyfriend?"

chatgpt-time-check-in.jpg

OpenAI is still improving its models to better detect signs of mental or emotional distress, to ensure that ChatGPT does not feed into signs of delusion or emotional dependency and can direct users to helpful, evidence-based resources.

The company says that it has worked with more than 90 physicians across 30 countries to build custom rubrics for complex, multi-turn conversations. Human-computer interaction researchers have been recruited to give feedback on how ChatGPT has identified concerning behaviors and to stress-test product safeguards, plus OpenAI is putting together an advisory group of experts in mental health, youth development, and human-computer interaction.

Article Link: OpenAI Adds Break Reminders and Mental Health Features to ChatGPT
 
Having worked in this AI field, I can say: I will NEVER EVER let my kids use ChatGPT. NOT EVEN 30 minutes.
Steve Jobs never let his daugther use iPad longer than 30 minutes.
I know tons of people working at Meta. And even they are very strict to their kids and don´t let them use Instagram.

I see more and more GenZ users who rely on ChatGPT as a therapist. And everybody who works in this field knows that it is a real danger and Instagram or other SNS platforms are "nothing", because we observe that such faceless ChatGPT bots could be "best friends" for users that they could get depressed more than ever.

Nobody could stop Meta. I don´t think that our nation is ever interested in regulating OpenAI and co. in this case.
 


OpenAI is making several changes to the way that ChatGPT works, with the aim of helping people use the chatbot in a healthier way.

chatgpt-logo.jpg

Starting today, ChatGPT will provide users with "gentle reminders" about how long they've been using the service, with suggestions for breaks. OpenAI says that it isn't measuring success by time spent on ChatGPT, but by ensuring that people leave the product after finishing what they came for.

OpenAI is working to improve ChatGPT's responses to certain kinds of questions. Rather than providing a direct answer to a high-stakes, sensitive question, ChatGPT should help users weigh pros and cons and think it through. OpenAI is updating ChatGPT's behavior for important personal decisions soon, so users will see more helpful responses for questions like "Should I break up with my boyfriend?"

chatgpt-time-check-in.jpg

OpenAI is still improving its models to better detect signs of mental or emotional distress, to ensure that ChatGPT does not feed into signs of delusion or emotional dependency and can direct users to helpful, evidence-based resources.

The company says that it has worked with more than 90 physicians across 30 countries to build custom rubrics for complex, multi-turn conversations. Human-computer interaction researchers have been recruited to give feedback on how ChatGPT has identified concerning behaviors and to stress-test product safeguards, plus OpenAI is putting together an advisory group of experts in mental health, youth development, and human-computer interaction.

Article Link: OpenAI Adds Break Reminders and Mental Health Features to ChatGPT
This is a joke, right?
 
  • Like
Reactions: The Ensorcerer
ChatGPT is an excellent foreign language tutor FYI.
Yeah, I've had good success with Grok3 voice mode for this as well, it's one of the best use cases for it IMO since you can invoke command in different languages without context switching and it can detect them perfectly or correct mispronunciations within reason.

Grok isn't good enough at sourcing articles directly e.g. it will actually get a list of research papers and have them in the referenced docs but if you ask it for the URL it always gets it wrong, due to that I'm cancelling until they improve the model but it was a fun experiment overall.

I haven't used voice mode with anything else since OpenAI is retaining all data right now due to the NYT lawsuit but once that is settled I look forward to trying out that use case there too.
 
Was just thinking of the related topic of digital wellbeing and multiple users drafting:

multiple users is good for privacy - but it could be improved by adding both digital wellbeing and AI to a top level menu


idea - The ability to *hide* / set time limits for a user space instead of just deleting it outright. This way, if we want to visually remove it, we could. A user space takes a long time to finalize.

And if it's for digital wellbeing, maybe add in a bit of AI chat to talk about even going to that user space. Something like "are you sure you want to go to this user space" and we could have a conversation for a minute or two on whether or not we really actually want to unlock the user space

Pixel supports multiple users and should add some of these options I think. If they don't, i would like to see more marketplace competition from Apple and Samsung to add multiple users but to enhance the capabilities with some of the digital well-being ideas above.


I made a concept / mockup for some options that Google could add to mix digital wellbeing and AI with the multiple users functionality and/or something that Apple could do as well, when they implement multiple users.
 

Attachments

  • 1754182183669~5.jpg
    1754182183669~5.jpg
    109.4 KB · Views: 11
Last edited:
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.