You’ve probably heard about AlphaFold, the AI system that’s made huge breakthroughs in protein folding. But what does it really mean to say something is ‘AI’? In reality, AlphaFold uses a super clever machine learning technique to predict how a string of amino acids will fold into a complex 3D shape. This is not the same as the artificial intelligence we typically think of, which is all about automating and replacing human thinking in a generalized way. AlphaFold will never compete in the Turing test, and it won’t replace anyone’s job. It’s a very narrow machine learning technique that’s simply amazing in its own right. So, let’s stop conflating LLMs and AlphaFold. It’s time to give these two concepts the respect they deserve – or rather, the respect they don’t deserve to be lumped together with. Here’s why:
* AlphaFold uses multi-layer perceptrons to predict protein folding, which is a very specific task.
* This technique is not generalizable to other tasks, unlike true AI.
* AlphaFold will never replace human intelligence or decision-making.
In short, AlphaFold is a remarkable achievement in machine learning, but it’s not AI. Let’s appreciate its unique capabilities without confusing it with the broader field of artificial intelligence.