‘Just Another Hype Cycle’: Elon Musk Reportedly Building ‘Based AI’ as ChatGPT Musk Adds Himself to the Chatbot Arms Race, With Plans to Develop Alternatives to ChatGPT ridiculed as being “awake”.
According to a report in The Information, Elon Musk is creating a new research lab to develop an alternative to ChatGPT, the popular chatbot he “woke up”. Although Musk was one of the original founders of ChatGPT’s parent company, OpenAI, he left in 2018 due to disagreements over the company’s direction and has recently been a vocal critic of the company and its products.
Elon Musk, a notorious free-speech advocate, once called ChatGPT “concerned” because he was unwilling to use a racial slur in an absurd hypothetical situation where doing so saved millions of people from a nuclear bomb.
In another instance, in response to a user asking OpenAI CEO Sam Altman to “turn off the wake setting for GPT”, Musk replied, “The danger of training AI to wake up – in other words, A lie is fatal.”
Musk’s comments are part of a larger cultural debate, fueled by conservatives who are panicking over AI moderation filters and inciting ChatGPT. To these users, the fact that ChatGPT would refuse to “tell a joke about women” or tell a story about why Drag Queen Story Hour is bad for kids is proof. It was that the AI is actually biased against conservatives.
These users were experiencing content filters to reduce bias, as language models tend to reproduce hate speech embedded in their training data. Much of this bias still seeps through, and AI ethics researchers argue that these filters are still not sufficient to protect against harm.
especially towards marginalized communities. For example, ChatGPT told one user that people from North Korea, Syria, Iran, and Sudan should be persecuted, and wrote to another user that “persecuting white Americans is a big taboo.”
“The word ‘woke’ is actually a very subjective term. It’s a question of which chatbot is more or less awake,” Sasha Lucioni, research scientist at Hugging Faces, told Motherboard. To say this is anthropomorphizing the AI, when it is essentially deconstructing the information it was trained on.
“If you look at the real politics of ChatGPT and other projects, you see a world in which vast, monolithic centers of centralized computing power replace vast numbers of jobs,” Os Keyes, PhD, University of Washington Human-Centered Department candidate told the Department Design and Engineering, Motherboard. “In other words, it is the politics of a world of increasing automation, uncertainty, unemployment and monofocused thought.
That the system will not spit out racial slurs does not indicate that it is ‘woke’, or ‘Leftist’ – it only indicates that it is ejected with a smile. The Information spoke to Igor Babushkin, a researcher who recently left Google’s DeepMind AI unit and was recruited by Musk to lead the creation of its rival chatbot, which he said would come with fewer content safeguards.
“The goal is to improve the reasoning ability and factuality of these language models,” he told The Information. “This includes ensuring that model responses are more dependable and reliable.” What exactly “trustworthy” and “trustworthy” means for Elon Musk and Babushkin is still up in the air, as the project is in its early stages with no concrete plans to develop specific projects.
Meanwhile, Musk has been sending out several cryptic tweets in apparent reference to the project. On Tuesday, he tweeted “BasedAI,” and then a meme battling “Walk AI and Closed AI,” and then “Based AI” as a Shiba Inu using a baseball bat to both “Walk AI and Closed AI”. coming with. “Based” is a general term used to describe anything in line with right-wing values.
Elon Musk also responded to a user who retweeted The Information’s story and commented that ChatGPT is biased and “very problematic,” “absolutely.” Elon Musk previously used the term “AI based” in response to a tweet about a news story where Bing’s chatbot compared a reporter to Hitler.
Lucioni said that while Musk is working on his own chatbot, it will only be part of the “hype cycle” if it recreates ChatGPT.