OpenAI updating ChatGPT to encourage healthier use – 9to5Mac
OpenAI is updating how ChatGPT works to encourage healthier use and avoid unintended consequences. One change that OpenAI says ChatGPT users will see “starting today” is a technique used by other digital services.
Much like video streaming services and social networks, OpenAI is adding a gentle break reminder for users during prolonged chat sessions.
Starting today, you’ll see gentle reminders during long sessions to encourage breaks. We’ll keep tuning when and how they show up so they feel natural and helpful.
According to the blog post, OpenAI says it measures success in return visits, not session duration. This subtle nudge to spend less time in ChatGPT sounds like it will change over time, but we should see the first examples in actual use from today.
OpenAI provides this example of how usage break reminders will look to start:
OpenAI also says it’s tuning ChatGPT to be less absolute when a user asks for actionable advice. For example, if a user asks if they should end a relationship, OpenAI says ChatGPT shouldn’t provide a yes or no response. Instead, ChatGPT should respond with prompts that encourage reflection and guide users to think through problems on their own.
The company also says it has “worked with over 90 physicians across over 30 countries—psychiatrists, pediatricians, and general practitioners — to build custom rubrics for evaluating complex, multi-turn conversations,” and it is “convening an advisory group of experts” to help it further improve healthy ChatGPT usage. You can read the full announcement post here.
At a high level, avoiding absolute responses for life advice sounds like the right direction for ChatGPT.
The reality is that you can convince ChatGPT to answer any inquiry with the one answer or another. Don’t like what ChatGPT has to say? Just prompt it again to get a different response.
For that reason alone, OpenAI should avoid absolute responses and strive for a more consistent experience that encourages critical thinking rather than being a substitute for decision-making.
As for the subtle reminders to take breaks from ChatGPT use, I’m more skeptical of how effective these prompts are without seeing actual data. Both changes give OpenAI some cover from criticism over how ChatGPT usage can be unhealthy or dangerous without proper guardrails.
FTC: We use income earning auto affiliate links. More.
Check out 9to5Mac on YouTube for more Apple news:
Zac covers Apple news, hosts the 9to5Mac Happy Hour podcast, and created SpaceExplored.com.
