AI reasoning models aren’t as smart as they were cracked up to be, Apple study claims

A new study by Apple has ignited controversy in the AI field by showing how reasoning models undergo 'complete accuracy collapse' when overloaded with complex problems.