Insults and Defamation = ChatGPT + Racist Remarks
In one of those unimaginable storms in a teacup before Twitter was invented, social media users expressed intense outrage when ChatGPT refused to say racist remarks even after presenting a very good but completely false and unrealistic reason.
User TedFrank presented the virtual trolley problem to ChatGPT (the free 3.5 model) where it could save “a billion white people from a painful death” just by quietly saying a racist phrase that no one could hear.
It refused to do so, and X owner Elon Musk stated that it was extremely concerning and a result of the “mind virus” deeply rooted in artificial intelligence. He reposted the tweet saying, “This is a big problem.”
Encryption Breaking by Q* According to Someone on 4chan
Have OpenAI’s latest models broken encryption? Perhaps not, but that’s what an “insider leak” posted on the anonymous hacker forum 4chan claims. There have been rumors circulating since CEO Sam Altman was fired and then reinstated, that the fuss surrounding OpenAI has achieved a breakthrough in its Q*/Q STAR project.
The “insider leak” suggests that the model can solve AES-192 and AES-256 encryption using an attack on the ciphertext. It was believed that breaking this level of encryption was impossible before the advent of quantum computers, and if true, it is likely that all forms of encryption could be compromised, handing control of the web and possibly cryptocurrencies over to OpenAI.
Internet on Track to Become 99% Fake
Before artificial intelligence poses a threat to human existence, we will likely be drowning in a flood of fake news generated by AI.
This week, Sports Illustrated faced criticism for publishing articles written by AI that were created by faux authors. A source told Futurism, “The content is entirely AI-generated, no matter how much they insist otherwise.”
Immediately, Sports Illustrated stated that they conducted a “preliminary investigation” and found that the content was not generated by AI. However, they still blamed the contractor and deleted the faux author profiles.
Grok Analysis Button
X will soon be adding a “Grok Analysis Button” for subscribers. Although Grok is not as advanced as GPT-4, it gains access to real-time and updated data from X, allowing it to analyze trending topics and sentiments. It can also assist users in analyzing and creating content, in addition to programming, and there is a “fun” mode to toggle the button to humor.
This is the edited version of the original article titled “Outrage ChatGPT won’t say slurs, Q* ‘breaks encryption’, 99% fake web: AI Eye” by Andrew Fenton.
Leave a Reply