I believe some people can be comforted and helped by AI companions, but it seems clear that the companies making them are not acting responsibly. Excellent from Josh Dzieza www.theverge.com/c/24300623/a... - ThreadSky

caseynewton.bsky.social • 165 days ago

I believe some people can be comforted and helped by AI companions, but it seems clear that the companies making them are not acting responsibly. Excellent from Josh Dzieza https://www.theverge.com/c/24300623/ai-companions-replika-openai-chatgpt-assistant-romance

Comments

copernicussees.bsky.social•165 days ago

Leaving aside whatever bogus training data these rocket scientists did, and a complete lack of active moderation of the bot, I still have concerns about e.g. using these as companions for those with dementia. Users need to be aware that they are talking to a virtual person; they can't be.

joseggarcia.bsky.social•165 days ago

Part of the problem is that our healthcare regulatory processes are designed for molecules, devices and software that remains static.

How do you regulate AI? It's anything but static, regulating it is like nailing jello to a wall.

joseggarcia.bsky.social•165 days ago

I'm on the fence, there are some good vendors in this space working with the tech's limitations, building in secondary systems to improve relevance and screen for harmful content.

But are they medical devices? If the answer is yes then that means regulatory oversight.

joseggarcia.bsky.social•165 days ago

Yes and no. There are responsible vendors like Woebot.

I work in digital health and we all despise vendors like this. They're putting on a show for VCs, they will not succeed in health care and they are tarnishing the sectors reputation.

sioflynn.bsky.social•165 days ago

https://www.bbc.com/articles/c4nnje9rpjgo

sioflynn.bsky.social•165 days ago

Have you looked at ghost bots in China?? Really moving. Stumbled on this researching artificial intimacy & Replika’s origin story. Founder’s loss of best friend.
Also check out chatGPT hack DAN (Do Anything Now) & DIY boyfriends. Chinese influencer in CA. https://youtu.be/ChXhmf0tYiw

zadett.bsky.social•165 days ago

Other than banking, the individual can always turn it off. What I'm worried about is AI manipulating the food chain.

samalex.bsky.social•165 days ago

this was an exceptional read

not-wyoming.bsky.social•165 days ago

Real Q - can you name three (3) companies in the past twenty years working on emerging tech that *have* acted responsibly?

(I cannot!)

thomas-sparrevohn.bsky.social•165 days ago

Great article - scary concept through

heppy111.bsky.social•165 days ago

📌

mikeelgan.bsky.social•165 days ago

Where is the movement to push chatbot developers and robot makers to make sure their products don't hijack human emotions and delude people into thinking these machines have consciousness, empathy or humanity?

roxannedautremont.bsky.social•165 days ago

Google is the only one who says, "Hi Sexy" To me anymore...( Yes, I told it to call me that - thx Big Bang Theory for the idea.)

ayeletgordtap.bsky.social•165 days ago

Absolutely! Reading some of the posts in the Replika subreddit can be really creepy

ayeletgordtap.bsky.social•165 days ago

and also reminds me of Tay, the Microsoft chatbot that was quickly taken down because it became so vile

blueblimp.bsky.social•165 days ago

Fantastic article. Clever closer too.

matunos.bsky.social•165 days ago

that excerpt makes me think Replika trained their AI by ingesting the various AITA subreddits

lanimal.bsky.social•165 days ago

Now I kind of want a bot trained only on that subreddit it would be so goddam dramatic

matunos.bsky.social•165 days ago

actually I think the AIO subreddit would be moreso.

a significant number of those posts seem to be produced by generative AI themselves, so it would be interesting to see where the downward spiral goes

ekana.org•165 days ago

Man, Replika moved from a kinda cool dystopian product to a really sad dystopian product. Originally the sell was that it would feed of your interactions, chat history, and social media presence to make a digital "replika" of you. But users just ended up using it as a companion so they pivoted.

brorodeo.bsky.social•165 days ago

Kara Swisher-ass OpenAI apologist. Stomach turning bad faith and anti-Medicine. The position of this article is legitimizing people not seeking effective help. Full stop. This is unethical advertisement for the people probably paying you. "Should software be able to do malpractice?"

Comments

Posting Rules

Reply