There’s a lot of talk about AI doing the rounds at the moment. For some people, it’s the best thing since sliced bread – it even knows what sliced bread is, which is kind of weird – for others, however, it’s an apocalyptic disaster just waiting to happen.
Now we’re not sure whether AI is going to be the human-killing tech that some people say it is – heck, it’s only just learned about sliced bread, baby steps, people – but a lot of the concern surrounding AI’s negative influences are warranted, and they are something we should eventually talk about.
But not yet.
The Concept Of True AI
Why do we say this? Because, technically, AI doesn’t actually exist. That is to say, true AI doesn’t exist. Right now, the AI that we’re seeing is a marvel. It can help sift through vast amounts of data for a healthcare company, or it can help detect fraudulent activities when playing live casino.
Every industry around the world can benefit from the technology, but the technology itself is not AI as we think it is. Take ChatGPT, for instance. Since the invention of the computer in the 1940s, this has been the single biggest achievement in the road to AI, but what does it actually do?
Following The Script
Well, it does exactly what your phone has been doing for the last couple of decades. Predictive text is essentially technology that knows what you want to say based on your prompts – it’s prompt completion, on a small scale. ChatGPT does exactly the same thing, albeit on a far larger scale. The true marvel behind ChatGPT is that it can generate text that reads like a human, but it is generating that text from its database – the world wide web.
It doesn’t understand that text. It doesn’t umm and ahh about what response is best. ChatGPT can offer automated translation and text generation, but it has no conscious understanding of the words themselves. In this way, they’re no different from any other computer tool that is used to analyse and classify data.
The Pursuit For AGI
The AI that humans aim for is known as AGI – artificial intelligence with a human level of cognition – and this is something that has not been achieved yet. Certainly not with ChatGPT, which can only be described as “weak AI” – AI that can only function within its set parameters.
Indeed, the AI that humans strive for should exist not as an LLM program, but a robotic entity that can think for itself, learn, and assist. Discovering how to achieve that is a long way in the future, and it may never even be achieved.
Understanding Sliced Bread
It’s baffling, even, that a platform like ChatGPT has caused such an explosion in the first place. There are plenty of other AI technologies that could have had the same impact, but didn’t, such as the AI used in gaming and world-building. Again, this is AI that is not at all intelligent beyond its programming. It is AI that is only intelligent in its tasks, predefined rules, patterns and programming.
So in response to those excited about AI and those who are concerned – conserve your energy. While ChatGPT is enormously impressive, in terms of the worldwide pursuit for true AI, it might know about sliced bread, but it doesn’t understand it. If something we deem intelligent doesn’t understand what sliced bread is, that tells us that there’s still some way to go.
