xAI to unveil its chatbot to a limited group on Saturday

Elon Musk the head of xAI has announced via X (formerly Twitter) that the company will launch its generative AI chatbot to a limited audience on Saturday. We will have to wait for more specific details as Musk only divulged that “in some important respects, [the AI] is the best that currently exists.”

Back in July, Elon Musk told an X Space that xAI is creating a chatbot which will be called “TruthGPT”. Musk said that xAI would try to develop a chatbot that is more trustworthy than both ChatGPT and Bard - achieving this, however, would require overcoming hallucination where an AI just makes stuff up but makes its sound convincing at the same time.

By launching to a limited audience, any serious issues can be ironed out. As an increasingly polarizing figure some people may also try to derail the launch of xAI’s chatbot and be overly critical simply because they don’t like Elon Musk’s tirades online, launching to a small group can prevent this at first.

Before announcing the opening up of xAI’s chatbot, Musk also shared that there was a new AI-powered feature to see similar posts that was rolling out now on X. It’s not clear if any of the xAI technology is being used in this new X feature but it wouldn’t be surprising if this was the case because xAI has said it will work closely with X and Musk has previously stated that X will be an everything app.

In recent days, Elon Musk has been attending the first AI summit in the UK where he sat down for an hour-long interview with UK PM Rishi Sunak. He reposted a clip of this interview online where he said that he wants X to be a platform that is as truthful as possible - even if some people don’t like it.

He also posted a cartoon on his feed criticizing countries attending the AI summit. The cartoon showed a lion representing the UK, a woman with a Europa bandanna on, a dragon representing China, and Uncle Sam representing the US.

Outloud they’re shown as saying “We declare that AI poses a potentially catastrophic risk to humankind” but then secretly thinking “And I cannot wait to develop it first” showing that Musk is skeptical of the talk around AI safety.

Each of the different large language models that have come online so far have varying levels of censorship. Out of them all, Bard seems to be the most prudish and bots accessed through Poe seem to be the most willing to give an answer, even on sensitive subjects.

Given Musk’s attitude to AI safety, it’s possible that xAI’s chatbot could be a bit more liberal in what it will generate rather than worrying about offending any sensitive souls out there.