There’s a really interesting research paper recently published by Apple’s engineers looking at how well LLMs (Large Language Models), the heart of AI systems like ChatGPT, and asking about their ability to deliver mathematical reasoning. They conclude that they are not ready for prime time right now. To illustrate the issue, TechCrunch gave this example: Let’s say…

The post AI may be good with words, but when it comes to numbers, duh? appeared first on Chris Skinner’s blog.