Comments
Michael •
The tweet which asks “Have you been hacked?”, it is a little suspicious that this happens about the same time that Google launches a limited public preview of Bard…
MalayaZemlya •
The chat history about wasn’t showing up since about a week before the big crash.
I think the functionality was just broken and they started showing wrong chats when they tried to fix it.
Felix •
What privacy was there to lose? ChatGPT was already moderated (so your chats could be reviewed by a random stranger) and the site specifically warned against including personal information.
The only difference is that the random stranger seeing your chats was another user instead of a moderator. Well, users aren’t bound by the same policies as mods, so maybe that’s meaningful.
Misplaced trust.
iAPX •
I would bet this is a C-suite induced problem: how to go to market as fast as possible before competitors.
Ethic team? It slow things down. Fire them!
QC, Security, Privacy? Also slow things down.
This sector is totally non-regulated and it show!
I hope there will be regulations in place before we run into a Skynet-like problem…
ResearcherZero •
“The FTC Act’s prohibition on deceptive or unfair conduct can apply if you make, sell, or use a tool that is effectively designed to deceive – even if that’s not its intended or sole purpose.”
“…Commission staff is tracking those concerns closely as companies continue to rush these products to market and as human-computer interactions keep taking new and possibly dangerous turns.”
‘https://www.ftc.gov/business-guidance/blog/2023/03/chatbots-deepfakes-voice-clones-ai-deception-sale
ResearcherZero •
“Basically, the model learns to recognize familiar shapes in a field of pure noise, then gradually brings those elements into focus if they match the words in the prompt.”
‘https://arstechnica.com/information-technology/2022/09/with-stable-diffusion-you-may-never-believe-what-you-see-online-again/
Fake
‘https://pbs.twimg.com/media/Frwx69YaAAA_WuF.jpg
J.W. Bolton, speaking at the 1943 Convention of the American Foundrymen’s Society (AFS), made the following statements.
“Your indulgence is requested to permit the posing of one question. Will real control of graphite shape be realized in gray iron? Visualize a material, possessing (as-cast) graphite flakes or groupings resembling those of malleable iron instead of elongated flakes.”
A few weeks later, in the International Nickel Company Research Laboratory, Keith Dwight Millis made a ladle addition of magnesium (as a copper-magnesium alloy) to cast iron and justified Bolton’s optimism – the solidified castings contained not flakes, but nearly perfect spheres of graphite. Ductile Iron was born!
‘https://web.archive.org/web/20010129000000/http://www.ductile.org/didata/Section2/2intro.htm
‘https://www.nuclear-power.com/nuclear-engineering/materials-science/material-properties/ductility/
“While humans have been busy over these six decades with our political anguish, and our wars, we have also created a universe inside our universe, one with its own infinite intelligence, composed of cryptic atomic switches, enlightened with ultraviolet and built on sand.”
‘https://www.wired.com/story/i-saw-the-face-of-god-in-a-tsmc-factory/
Winter •
@ResearcherZero
Re: With Stable Diffusion, you may never believe what you see online again
Fake news started with the invention of gossip, ak, the development of language itself.
Just some recent highlights:
‘https://www.cits.ucsb.edu/fake-news/brief-history
‘https://www.theguardian.com/books/2019/nov/22/factitious-taradiddle-dictionary-real-history-fake-news
Now, people have learned to dismiss random strangers spreading evil gossip (or learned to punch them in the face if it is too bad). But with every new “news” channel, the learning has to start again from scratch.
It might sound idiotic to simply trust some random person online, but fact-checking costs time&effort and gossiping feels so good. Even I want to believe those I despise to be evil to the core. But I know that evil is not related to political alliance. Anyone in “my camp” is just as likely as not to do something horrible as anyone in the “other camp”. Just look at MeToo, that too crossed every social and political border.
Robin •
I can attest to this! A few days before they disabled the history, I got really random conversations in my history that I couldn’t access… “GMOs in India”, “India GMOs” and things like this. I never asked ChatGPT about things like this, so I figured that the only possibility is that histories of other users were leaking into my own. Kind of scared the hell out of me to be honest.
Note that I could never access these conversations.
modem phonemes •
@ Winter
people have learned to dismiss random strangers
As you say, we have developed some sense of what’s probably OK and how far to trust. Up to the current point this has been based on the assumption the originator is human.
This seems to need great revison now that the source can be a statistics based computation. There is no socially derived truthiness estimate to work with.
Erdem Memisyazici •
It’s not like they have the only LLM chatbot. Run your own, mind the airgaps?
Mr. Peed Off •
Having recently been experimenting with Stable Diffusion, I noticed a lot of code has not been well vetted for malware. Use caution and scan all downloads before opening or using them. With a lot of code being shared amongst users, this is a malware outbreak just waiting to happen. Also you check the metadata before sharing art to make sure you are not revealing anything you don’t want to share.
Subscribe to comments on this entry
Sidebar photo of Bruce Schneier by Joe MacInnis.
Views: 0