Apple researchers have found “fundamental limitations” in cutting-edge artificial intelligence models, in a paper raising doubts about the technology industry’s race to develop ever more powerful systems.
Apple said in a paper (PDF download) published at the weekend that large reasoning models (LRMs) – an advanced form of AI – faced a “complete accuracy collapse” when presented with highly complex problems.
It found that standard AI models outperformed LRMs in low-complexity tasks, while both types of model suffered “complete collapse” with high-complexity tasks. Large reasoning models attempt to solve complex queries by generating detailed thinking processes that break down the problem into smaller steps.
No, they definitely are AI. ChatGPT for example is a generative pretrained transformer (GPT) is a transformer model is a deep learning model is a machine learning model is AI.
It’s just that the general public has no fucking idea what “AI” is due to being swamped in marketing about a field they have zero background in and have been led to believe is some kind of general intellect on the level of a human or smarter. In reality, a perceptron with one weight and one bias is machine learning is AI.
Since the start, what “AI” is has been fairly arbitrary; it’s just the ability for a machine to perform tasks we’d associate with human intelligence. It doesn’t even need to be machine learning; that’s just one branch. The game Video Checkers (1980) for the Atari 2600 running on 128 bytes of RAM has AI that you play against. The bar isn’t high at all.