Large Language Models: AI ≠ LLM

      Bottom line: So called ‘artificial intelligence’ isn’t intelligent. Don’t call it that.

      Unless you’ve been living under a rock, I’m sure you’ve heard of ‘AI’. Or at least you think you have. What you’ve heard about is called ‘artificial intelligence’, but it’s really ‘large language models‘. I think the distinction is an important one.

      ‘Artificial’ is the obvious part. It says ‘not natural’, ‘man made’. I’ve got no problem with that, we make alot of things. Such as whatever device you are reading this on. ‘Intelligence’, on the other hand, cause worrisome implications here.

      Something that is intelligent is able to make decisions, reason things out, offer advice, and so on. If they have more information then you, wouldn’t that make them more intelligent then you? And if they are more intelligent then you, maybe you should be following their direction. After all, isn’t that what we’ve been told for so many years? That we live in a meritocracy, where those who are wiser and smarter make the decisions. And if the greatest intelligence happens to be artificial? Why should that change anything. Trust the expert, even if it’s artificial.

      I’ve already interacted with people thinking this way. Who believe that ‘AI’ is able to magically solve all their problems. Which is not ideal.

      But a model, regardless of how large it is, or how much language it has been fed, is just a model.

      We’ve all dealt with models. Model cars, planes, buildings, etc. Even a map is really just a model of a location. We wouldn’t confuse a model for the real thing. Even as models can be incredibly useful in certain situations, we don’t outsource our thinking to them. At least not in the same way. We expect to do some error checking with the results, or use the model knowing full well it’s only ‘close enough’.

      Perhaps someday these ‘large language models’ will reach the point of intelligence. But until then, it behooves us to remember they are models. Tools to be used, where they can be useful. When not useful? Then we have to rely on our natural intelligence.

      Or as a wise man once said ‘you must be smarter then the tools you are using’. Because if you aren’t? Then the tools may well end up using you.