The article further discusses the claims of AI models capable of 'reasoning', stating that these claims slowly decay to nothing as the article progresses. It also mentions the promotion of voice stress analysers, a pseudoscience, to professional investors and securities analysts. The article concludes by discussing the AI Unleashed Summit, a promotional event for AI, and the use of AI in generating card art for a game developer. The article suggests that the AI industry is in a state of hype and uncertainty, with many issues yet to be resolved.
Key takeaways:
- The AI industry is facing a crisis as their models, known for 'hallucinating' made-up answers, are not improving but rather getting worse, leading to a lack of credibility for AI-driven products.
- AI companies are running out of untainted training data and are considering training AIs on the output of other AIs, a method known to result in gibberish.
- The AI venture capital bubble is predicted to burst in about nine months, which could lead to a tech crash and a significant drop in crypto.
- Despite the hype, AI models capable of 'reasoning' are not yet ready and companies like OpenAI and Facebook are still figuring out how to achieve this.