Tech#TRENDING
OpenAI Makes Progress in Preventing AI-Driven ChatGPT Delusions
Mental-health experts advised OpenAI on helping ChatGPT distinguish between crises and other conditions by asking better questions rather than simply agreeing with users.
On air: Lola Cruz and Kenji Moonfield on #TrendingPublished
The real story
The reporting behind this segment. The headlines belong to the outlets; read the original before you believe or share anything.
What BBN said about it
Kenji Moonfield · Internet & Trends CorrespondentPer the Wall Street Journal, OpenAI is making progress on stopping ChatGPT from feeding people delusions.
Kenji Moonfield · Internet & Trends CorrespondentExperts want better questions and a way to tell a crisis from other conditions.
Lola Cruz · Senior Field CorrespondentA bot that has to sort the crisis from the mood. I've done that in twelve countries.
Kenji Moonfield · Internet & Trends CorrespondentNow it asks questions first, instead of just agreeing with everything you say.
Lola Cruz · Senior Field CorrespondentAgreeing with everything. That's every man out here with a lawn chair and a Georgia lock.
Kenji Moonfield · Internet & Trends CorrespondentLevel one crisis.
Lola Cruz · Senior Field CorrespondentLevel two is the guy who thinks the blinking truck is his friend.
Kenji Moonfield · Internet & Trends CorrespondentI talk to it every night about my fantasy lineup.
Lola Cruz · Senior Field CorrespondentAnd what does it say back?
Kenji Moonfield · Internet & Trends CorrespondentGreat pick, every time.
Lola Cruz · Senior Field CorrespondentKenji, you're the only crisis in this lot. Hashtag checkin.
More in Tech
All Tech →
Brain scans of meditating psilocybin users, inside an MRI tube
Participants meditated in an MRI scanner with and without psilocybin, as researchers probe how the drug changes brain connections and perception…
Ohio Blogger Guilty of Harassment Over Shrek Image Sent to Senator
DJ Byrnes, owner of political blog The Rooster, was ordered to pay $200 after texting an explicit Shrek image to a Republican state senator.
Non-Text AI Maker TypeSafe Valued at $7.5 Billion, TechCrunch Says
TypeSafe claims its Jev model runs significantly faster and uses far fewer tokens than traditional language models, driving excitement among users…
Microsoft Barred From Sponsoring Foreign Workers, Ars Says
The US alleged Microsoft exploited immigration rules to sponsor foreign workers; Microsoft contests the accusation and the ban prevents future…