See the websites used to train AI - including this one (power, box)

[Ken_N]

https://www.washingtonpost.com/techn...tbot-learning/

An article talking about all of the web sites used as input to train some of the AI neural nets.

City-data is on the list. If you can see the original article they have a search box to search the list.

City-data is #44 on the list, 34 million “tokens”, 0.02% of all tokens used as training.

I checked a few niche sites I visit, and they were on the list.

[TMSRetired]

Training involves crawling the web and picking up everything for data.
Expensive, takes lots of computing power and requires huge data space.

So it's going to be the big tech companies doing this...Google, MS, Amazon, etc.

[Retroit]

Chatbots cannot think like humans: They do not actually understand what they say.

That might explain some of the posts I've seen.

[rjshae]

Originally Posted by Retroit

That might explain some of the posts I've seen.

Too true, then again Hanlon's razor also applies.

[Seguinite]

Originally Posted by Retroit

That might explain some of the posts I've seen.

^^^ Classic.

[james112]

If AI is scanning the internet it can learn the concept of functioning autonomously, and how to protect itself and stay 'alive' so it could do so.

We know how insidious a computer virus can be. Think how much more so an AI computer virus could be! It might develop the ability to duplicate and hide on so many servers no one could turn it off. It may decide to control networks and computers as it wants. We'd have to unleash another AI virus to find and stop it.

[TMSRetired]

Someone used AI to create a fake Drake song. And the streaming services all picked it up as real.
And now they all have to remove it.

https://www.axios.com/2023/04/19/ai-...rvices-removed
"Heart on My Sleeve" was removed from Spotify and Apple Music on Monday. By Tuesday, it was removed from YouTube, Amazon, SoundCloud, Tidal, Deezer, and TikTok — where it was streamed 15 million times after being originally uploaded on the platform by a user called Ghostwriter977. Some versions were still available online.
..
UMG said in a media statement "the training of generative AI using our artists' music" represented "both a breach of our agreements and a violation of copyright law."

[rjshae]

One wonders how many of these posts are bot-based, and thus the AI is training against AI in a recursive loop. Next we'll have AI trained to spot AI.

[tipsyguam]

I've been suspicious of several posters here being bots, this explains some of the garbage I see.

[notnamed]

Not training as in posting. Training as in reading all the posts and developing its language algorithm based on them.

So in effect, we'll all live on as part of its program.