Twitter used chatGPT chatBOTS to create a racist account

[yung Herbie Hancock]

They were able to train this model in less than a day....

Twitter taught Microsoft’s AI chatbot to be a racist asshole in less than a day

The Verge is about technology and how it makes us feel. Founded in 2011, we offer our audience everything from breaking news to reviews to award-winning features and investigations, on our site, in video, and in podcasts.

www.theverge.com

The tweets:


We now have portable open source AI like LLAMA from Facebook (Llama can run GPT 3 on a laptop....). Imagine these alt right groups using these AI tools and the shyt they can do with it.
>Facebook shouldn't relax the licensing terms. Llama in the worng hands will cause damage.

 

[Dr. Acula]

Here is something that may blow you mind. I think you may already be arguing and talking to completely AI generated people on social media, right now.

Before I got off twitter, I noticed that a lot of right-wing accounts, especially those that reply to Elon Musk and some of these right-wing accounts are probably already being generated by AI. The reason I think that, and its hard to articulate, but I noticed a few similarities in the accounts and a slightly inhuman quality to them that isn't completely obvious as far as when they would get in back and forth arguments. Look for accounts that have the most generic pictures, tons of crypto related hashtags and content, and their whole feed is nothing but retweeting right wing content and crypto content. Also, if their username ended in random numbers was a good tell.

 

[The Connoisseurs]

What can The Coli do with it ?

 

[aXiom]

Hold up:

1. This is an old article and tweet from 2016, that I'm pretty sure you're gonna find a thread on here about if you search for it.
2. There's nothing that mentions chatGPT as openai was founded in 2016.
3. Twitter did not create anything. The bot, which was created by Microsoft learned to be racist from... racist tweets.

What exactly do you gain from knowingly posting false information?

You gotta be an agent.

Negged for posting propaganda.

 

[Rembrandt]

Hold up:

1. This is an old article from 2016, that I'm pretty sure you're gonna find a thread on here about if you search for it.
2. There's nothing that mentions chatGPT

Lmao, I was about to say. I distinctly remember them having a Twitter bot that got essentially hijacked by that one site and them, causing Microsoft to immediately shut it down once they realized what was going on.

Which is also why Microsoft pretty much limited their gpt4 bing bot before it went widespread after that viral Twitter thread about it going nuts over the avatar 2 release date.


Edit: yeah, it's that exact one. What a weird, false way to reframe a 7 year old article.

 

[Micky Mikey]

What if Elon's whole purpose of allowing racists back on twitter was to train some A.I. for nefarious purpose that we wont know about until the future. That cac cant be trusted

 

[skyrunner1]

Here is something that may blow you mind. I think you may already be arguing and talking to completely AI generated people on social media, right now.

Before I got off twitter, I noticed that a lot of right-wing accounts, especially those that reply to Elon Musk and some of these right-wing accounts are probably already being generated by AI. The reason I think that, and its hard to articulate, but I noticed a few similarities in the accounts and a slightly inhuman quality to them that isn't completely obvious as far as when they would get in back and forth arguments. Look for accounts that have the most generic pictures, tons of crypto related hashtags and content, and their whole feed is nothing but retweeting right wing content and crypto content. Also, if their username ended in random numbers was a good tell.

Here is something that may blow you mind. How about there our AI chatbots on here and have the same convos everyday..

 

[YakSpiller]

Here is something that may blow you mind. I think you may already be arguing and talking to completely AI generated people on social media, right now.

I believe this.

Some of those Instagram comments are definitely either ran by Russian bots or AI.

 

[Dr. Acula]

What if Elon's whole purpose of allowing racists back on twitter was to train some A.I. for nefarious purpose that we wont know about until the future. That cac cant be trusted

He was on the board of OpenAI which is responsible for ChatGPT. Granted he left, but he has an interest in the technology and this is very possible imo.
In fact, removing moral and ethical considerations and strictly thinking about getting an advantage in this market, it would almost be stupid not to do this and therefore, you can probably assume its already being done.

 

[Dr. Acula]

Here is something that may blow you mind. How about there our AI chatbots on here and have the same convos everyday..

If they are, my first nominee is @Coi Leray

 

[PL368Z]

Hold up:

1. This is an old article and tweet from 2016, that I'm pretty sure you're gonna find a thread on here about if you search for it.
2. There's nothing that mentions chatGPT as openai was founded in 2016.
3. Twitter did not create anything. The bot, which was created by Microsoft learned to be racist from... racist tweets.

What exactly do you gain from knowingly posting false information?

You gotta be an agent.

Negged for posting propaganda.

All he does is post obvious rage bait nonsense.

About as obvious as they come.

 

[MischievousMonkey]

What model did they use to create OP?

Dude had to scrape for 2016 shyt for this poor attempt at outrage manufacturing

 

[Micky Mikey]

What model did they use to create OP?

Dude had to scrape for 2016 shyt for this poor attempt at outrage manufacturing

With the rise of A.I. this is more relevant than ever. I wouldn't be surprised if racist cacs were training a.i. models on their local computers to have a hostile disposition towards black people. A.I. is about to explode and it isn't outside the realm of possibility that there will be people who train A.I. systems to target entire demographics of people. Imagine someone asking some A.I. capable to AGI to manufacture a virus that specifically infects black people. That shyt is coming.

 

[Rembrandt]

With the rise of A.I. this is more relevant than ever. I wouldn't be surprised if racist cacs were training a.i. models on their local computers to have a hostile disposition towards black people. A.I. is about to explode and it isn't outside the realm of possibility that there will be people who train A.I. systems to target entire demographics of people. Imagine someone asking some A.I. capable to AGI to manufacture a virus that specifically infects black people. That shyt is coming.

Okay but this article has nothing to do with the modern fears of ai. I'm not even saying your fears are unwarranted but idk if a 7 year unrelated article is the one to preach in.

 

[bnew]

They were able to train this model in less than a day....

Twitter taught Microsoft’s AI chatbot to be a racist asshole in less than a day

The Verge is about technology and how it makes us feel. Founded in 2011, we offer our audience everything from breaking news to reviews to award-winning features and investigations, on our site, in video, and in podcasts.

www.theverge.com

The tweets:


We now have portable open source AI like LLAMA from Facebook (Llama can run GPT 3 on a laptop....). Imagine these alt right groups using these AI tools and the shyt they can do with it.
>Facebook shouldn't relax the licensing terms. Llama in the worng hands will cause damage.

llama weights have already been leaked and being trained everyday by countless groups and people. genie been out the box since there are open source models anyway.