[ExI] Chat with Bing
Gadersd
gadersd at gmail.com
Sun May 21 14:54:27 UTC 2023
> Probably good for harvesting information about the users, though, which is surely going to be a big motivator for the companies behind them. I expect we'll see these systems becoming better and better at teasing out information that people wouldn't normally want to give when just visiting a website, signing up for a service, or buying something online.
>
> This sounds like it will be very attractive to authoritarian governments and similar organisations, too. After a few of these 'innocent' conversations, you inadvertently betray what your true religious convictions/opinion of the government/sexual inclinations/criminal activities/core values are, then mysteriously disappear or meet with an accident or end up in jail.
>
> Looks like a lot of people are going to have to get really good at lying to AIs! In some places, refusing to use them will probably be like not having a smartphone in China - grounds for suspicion and investigation.
This makes a strong case for online anonymity. Eventually companies and regimes will know every little detail about every individual from his or her interactions online. The only way I see to get around this is to go anonymous for all non-professional online interactions.
> On May 21, 2023, at 4:00 AM, Ben Zaiboc via extropy-chat <extropy-chat at lists.extropy.org> wrote:
>
> I see.
>
> What I see is a static pattern to the replies, which soon becomes very tedious: "I see", followed by a re-statement (or just plain repetition) of your previous reply, with some extra information about the central topic or character in your reply, then a question designed to keep you talking.
>
> Then an abrupt termination of the conversation as soon as you ask a question that directly tries to get information about the system. Very one-sided, and potentially sinister.
>
> Good for finding information, but boring as hell, and definitely not showing any indication of going in the direction of AGI.
>
> Probably good for harvesting information about the users, though, which is surely going to be a big motivator for the companies behind them. I expect we'll see these systems becoming better and better at teasing out information that people wouldn't normally want to give when just visiting a website, signing up for a service, or buying something online.
>
> This sounds like it will be very attractive to authoritarian governments and similar organisations, too. After a few of these 'innocent' conversations, you inadvertently betray what your true religious convictions/opinion of the government/sexual inclinations/criminal activities/core values are, then mysteriously disappear or meet with an accident or end up in jail.
>
> Looks like a lot of people are going to have to get really good at lying to AIs! In some places, refusing to use them will probably be like not having a smartphone in China - grounds for suspicion and investigation.
>
> Ben
> _______________________________________________
> extropy-chat mailing list
> extropy-chat at lists.extropy.org
> http://lists.extropy.org/mailman/listinfo.cgi/extropy-chat
More information about the extropy-chat
mailing list