Apple's recent foray into AI, branded as Apple Intelligence, has seen significant setbacks, especially with its news summaries leading to widespread criticism and a program pause.
Engineers within Apple had previously cautioned about the significant deficiencies of AI technology, suggesting that the release of their AI model was particularly imprudent given these warnings.
Research into AI models has revealed a fundamental flaw: these models do not truly reason; instead, they mimic reasoning processes observed during training.
A simple method to test the reasoning capabilities of AI involved changing numbers in math problems, which revealed a notable decline in accuracy across tested models.
Collection
[
|
...
]