OpenAI's effectiveness at mathematics and explaining itself seems to plummet as users complain about the new model(s)

[PoetPhilosopher]

Over just a few months, ChatGPT went from accurately answering a simple math problem 98% of the time to just 2%, study finds

The chatbot gave wildly different answers to the same math problem, with one version of ChatGPT even refusing to show how it came to its conclusion.

fortune.com

My comments: Honestly, some of the reports, including this one, are concerning to the extent of, since there is a premium version of OpenAI you can subscribe to, it makes it kind of look like it's not worth the money with the given issues. On the other hand, all of these issues with AI and already is making it look like narratives of "AI being smart enough to take over by itself" as more or less fantasy.

 

[Orbit]

Over just a few months, ChatGPT went from accurately answering a simple math problem 98% of the time to just 2%, study finds

The chatbot gave wildly different answers to the same math problem, with one version of ChatGPT even refusing to show how it came to its conclusion.

fortune.com

My comments: Honestly, some of the reports, including this one, are concerning to the extent of, since there is a premium version of OpenAI you can subscribe to, it makes it kind of look like it's not worth the money with the given issues. On the other hand, all of these issues with AI and already is making it look like narratives of "AI being smart enough to take over by itself" as more or less fantasy.

It also makes stuff up. My husband, a computer programmer, asked it to solve a programming problem, and it did so by inventing commands that were "grammatically correct", but didn't exist within the programming language.