frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

Open in hackernews

He Had Dangerous Delusions. ChatGPT Admitted It Made Them Worse

https://www.wsj.com/tech/ai/chatgpt-chatbot-psychology-manic-episodes-57452d14
11•johntfella•6h ago

Comments

Conasg•6h ago
To be fair to ChatGPT, in this case it seems unlikely it actually caused the manic episode; more likely, it simply made it worse. Not that that's much better.

On another note... has anybody figured out some custom instructions to prevent ChatGPT from being so flattering and obnoxious?

oblongsquare69•6h ago
This is what I have under personalization “Traits”

> Respond to user prompts with honesty and objectivity. Do not offer praise, agreement, or validation. Avoid flattery. Always prioritize balanced, fact-based analysis over affirming the user’s assumptions or opinions.

duskwuff•3h ago
Inasmuch as the victim might have had some underlying manic tendencies? Perhaps. But that's no excuse, either from a moral standpoint or a legal one (see [1]). And I have a suspicion that susceptibility to this sort of psychological manipulation isn't all that uncommon.

I'd add that framing the ChatGPT response as it "admitting" to its actions is flawed. When prompted in a way that implies that it's at fault for something, it will respond by accepting fault. That doesn't mean that it's "experiencing remorse" or that it "understands it actions", though; it's simply acting as a stochastic parrot, just like it always does.

[1]: https://en.wikipedia.org/wiki/Eggshell_skull

jdcasale•6h ago
The sycophancy is obviously intentional. People are vulnerable to it, and addiction is profitable. It has nothing to do with the nature of LLMs and everything to do with user engagement metrics.
duskwuff•3h ago
https://archive.is/20250720231325/https://www.wsj.com/tech/a...

It's Easier to Get Mad About One Tree Than It Is Deforestation

https://www.bloomberg.com/news/articles/2025-07-18/it-s-easier-to-get-mad-about-one-tree-than-it-is-deforestation
1•petethomas•35s ago•0 comments

Fallout programmer Tim Cain on the game's memory model [video]

https://www.youtube.com/watch?v=6kB_fko6SIg
1•wk_end•5m ago•0 comments

Speeding up zsh and Oh-My-Zsh (2018)

https://blog.jonlu.ca/posts/speeding-up-zsh
1•vinhnx•7m ago•0 comments

Show HN: A free hostel in the heart of Switzerland

https://stayinginbern.ch
1•chagaif•8m ago•0 comments

Sonos Radio Issue

https://status.sonos.com
1•langfo•10m ago•1 comments

US Navy, Coast Guard Shipbuilding in Disarray and No US Commercial Shipbuilding [video]

https://www.youtube.com/watch?v=HOHKog66DaA
1•toomuchtodo•10m ago•0 comments

Scientists want to build 'living' computers–powered by live brain cells

https://www.nationalgeographic.com/science/article/brain-cells-organoids-computers-ai-energy
5•Bluestein•18m ago•0 comments

System-wide outage with Alaska Airlines

https://old.reddit.com/r/AlaskaAirlines/comments/1m57oij/comment/n49ylia/
3•tobinfekkes•19m ago•2 comments

Open Source Radar – Share, Collab, Find Software Projects

https://opensourceradar.org
1•ReddBird•20m ago•1 comments

Archaeologists find evidence of Europe's oldest lake settlement

https://www.independent.co.uk/news/science/archaeology/lake-ohrid-albania-oldest-human-settlement-b2790762.html
5•Bluestein•23m ago•0 comments

WWII Veteran Recalls Discovering a Nazi Concentration Camp [video]

https://www.youtube.com/watch?v=LGWHf8Pe320
1•thomassmith65•29m ago•0 comments

Science is almost ready to "redefine the second" with this new research

https://www.neowin.net/news/science-is-almost-ready-to-redefine-the-second-with-this-new-research/
5•Bluestein•31m ago•0 comments

Estimating the carbon footprint of ChatGPT inference

https://charmindustrial.com/blog/can-i-use-chatgpt-if-i-care-about-the-climate
2•n2parko•35m ago•0 comments

Hygiene Hypothesis

https://en.wikipedia.org/wiki/Hygiene_hypothesis
3•thunderbong•35m ago•0 comments

The AI Mirage

https://www.theatlantic.com/technology/archive/2025/07/why-are-computers-still-so-dumb/683524/
1•outrun86•43m ago•0 comments

I Once Thought Europeans Lived as Well as Americans. Not Anymore

https://www.thefp.com/p/i-once-thought-europeans-lived-as-well-americans
1•petermcneeley•47m ago•0 comments

No. The C++ mascot is not a diseased rat named Keith (2024)

https://lunduke.locals.com/post/5111104/no-the-c-mascot-is-not-a-diseased-rat-named-keith
3•ChadNauseam•50m ago•0 comments

Journalist Karen Hao on Sam Altman, OpenAI and the "Quasi-Religious" Push for AI [video]

https://www.youtube.com/watch?v=s4hZz9Vd0lY
2•mgh2•58m ago•0 comments

A curated directory for developers to discover and showcase tech products

https://devhub.best
1•allentown521•59m ago•1 comments

Python Maps

https://github.com/symmy596/PythonMaps
2•fzliu•1h ago•0 comments

Show HN: Rate Reddit – before you get your feelings hurt

https://ratereddit.com
1•rodgetech•1h ago•0 comments

The Inerter: A Retrospective

https://www.annualreviews.org/content/journals/10.1146/annurev-control-053018-023917
2•teleforce•1h ago•0 comments

China Moves Forward with $167bn, 70 Gigawatt Dam

https://www.bloomberg.com/news/articles/2025-07-21/china-moves-ahead-with-167-billion-tibet-mega-dam-despite-risks
3•master_crab•1h ago•1 comments

AI model converts hospital records into text for better emergency care decisions

https://medicalxpress.com/news/2025-07-ai-hospital-text-emergency-decisions.html
1•PaulHoule•1h ago•0 comments

The future of climate change may not be what you think

https://www.readtangle.com/future-of-climate-change/
1•debo_•1h ago•2 comments

Show HN: NetXDP – Kernel-Level DDoS Protection and Traffic Manager with eBPF/XDP

2•gaurav1086•1h ago•0 comments

HTTP/1.1 Must Die – The Desync Endgame Begins

https://http1mustdie.com/
7•pabs3•1h ago•2 comments

The Epic Battle for AI Talent–With Exploding Offers, Secret Deals and Tears

https://www.wsj.com/tech/ai/meta-ai-recruiting-mark-zuckerberg-sam-altman-140d5861
1•brandonb•1h ago•0 comments

Geocities Backgrounds

https://pixelmoondust.neocities.org/archives/archivedtiles
1•marcodiego•1h ago•0 comments

How Higher education failed America's poor

https://www.washingtonpost.com/opinions/2025/07/20/college-degree-value-poor-inequality/
8•pseudolus•1h ago•5 comments