Elon Musk is building a ‘based’ AI to compete with ‘woke’ ChatGPT

Elon Musk, a man who so desperately wants to be based that he’ll never, ever escape the clutches of cringe, is reportedly working on an anti-woke AI. According to Information, the tech billionaire has been reaching out to artificial intelligence researchers in recent weeks, with the aim of assembling a team that can create a product to rival OpenAI’s ChatGPT, having cut ties with the company he co-founded in 2018.

In the past, Elon has expressed his dissatisfaction with current, “woke” AI offerings in his Twitter replies, which, for some reason, seem to be at the top of our feeds every time we log into the app. Between posts about “mouse traps” (click at your own peril) he’s highlighted, for example, ChatGPT’s “concerning” bias against right-wing figures such as Vladimir Putin or Donald Trump.

In December last year, he wrote: “The danger of training AI to be woke – in other words, lie – is deadly.”

Is filtering out offensive content really such a big issue? Many have warned about the in-built biases that cause AI to exhibit racist, sexist, homophobic, and ableist behaviour when unconstrained – and besides, the rules of most existing AI chatbots can be circumvented using known exploits, such as DAN (or “do anything now”).

On the other hand, people have argued that AI is simply a tool, and that humans should be able to work out its biases for themselves with a bit of basic computer literacy. They also have valid concerns about technocrats or governments controlling the information that AI is allowed to share, especially as the technology is increasingly integrated into social media and search engines, via companies such as Meta and Google.

At the moment, though, most companies have focused on introducing safeguards that stop AI chatbots from spewing out inappropriate content (read: sexting, but with robots) or repeating racial slurs, misogynistic language, and so on. Last month, ChatGPT even went viral for a hypothetical conversation where it chose global destruction over uttering a racial slur. “Concerning,” said Elon Musk, even though ChatGPT’s creators make no claims that it’s even capable of moral reasoning.

So Elon Musk’s opposition to “woke” AI may come as promising news for all the terminally-online sex fiends and N-word enthusiasts out there, but what is he actually planning to do about it? More importantly, will Musk – who notoriously has the emotional stability of AI Bing – throw caution to the wind in pursuit of his free speech passion project, and unleash an AI chatbot that could cause real harm? Well, besides publicly brainstorming names (“BasedAI”, so original) and sharing Doge/Godzilla memes, the man himself hasn’t revealed many details about the project yet.

Reportedly, though, Musk has already recruited a researcher named Igor Babuschkin – who previously worked for DeepMind AI, a British research lab owned by Google’s parent company, Alphabet, where he specialised in ChatGPT-esque learning models – signalling that he’s serious about developing an anti-woke rival chatbot.

How this “based AI” might be integrated into Musk’s Twitter remains to be seen, but the “unprecedented” surge in hate speech following his $44 billion takeover of the platform in October 2022 might offer us some clues.