Wikipedia Founder Jimmy Wales on AI errors

Key Concepts

• Trust in the age of AI
• AI Hallucinations
• AI Citation and Sourcing
• Large Language Models (LLMs)
• Information Verification
• AI-generated Content Reliability

Trust in the Time of AI

Jimmy Wales, founder of Wikipedia, discussed the critical issue of trust in the current era of Artificial Intelligence (AI). He highlighted a significant problem with AI models: their tendency to "hallucinate," meaning they generate false or fabricated information.

AI Hallucinations and Lack of Citation

Wales pointed out that AI models often fail to cite their sources and present fabricated information as fact. He provided a personal anecdote:

• Example: While discussing TV shows with an AI, he was recommended a show that sounded interesting. Upon further investigation, he discovered the show did not exist and was "made up out of thin air" by the AI. This experience underscored the user's expectation to interact with real-world information, stating, "I actually am a human. I live here on Earth and I only want to watch shows that exist."

This lack of factual grounding and citation is a major concern for the reliability of AI-generated content.

Elon Musk's "Graipedia" and LLM Limitations

Wales also commented on Elon Musk's recent launch of "Graipedia," an AI-powered encyclopedia. He expressed skepticism about its potential usefulness due to the inherent limitations of the Large Language Models (LLMs) used to create it.

• Argument: Wales believes that LLMs are prone to making "massive errors" and that there is "no way to know for sure" if the information they produce is accurate.
• Perspective: Consequently, he is "not that optimistic that he's going to create anything very useful right now" with Graipedia, given these underlying issues with the technology.

The CNBC Technology Executive Council

The transcript mentions the CNBC Technology Executive Council, which comprises technology executives from various industries. Interested individuals can learn more about joining by visiting cbcounsils.com.

Conclusion

Jimmy Wales's remarks emphasize the urgent need to address the problem of AI hallucinations and the lack of reliable sourcing in AI models. The current state of LLMs, as exemplified by concerns surrounding Graipedia, suggests that significant challenges remain in ensuring the trustworthiness and factual accuracy of AI-generated information, particularly when dealing with real-world knowledge.