How is ChatGPT like my dog?

My LinkedIn feed is flooded with ChatGPT posts, but what is the intelligence of the technology? The article below does a good job of making the technical details accessible. The upshot is that ChatGPT is using probabilities to construct language that sounds right to us. This begs the question, to what extent are we projecting a higher intelligence on something than we should, because it is behaving like we behave but is not actually executing the underlying complex processing that produces that behavior?

I have a great dog, and she's pretty smart, and she often acts a little human. And sometimes we find ourselves thinking she understands things that she probably does not understand. Really, we are human, and we've intentionally and unintentionally trained her over years and years with treats and food and praise to behave in ways that we understand and please us. But under the surface she is not doing the complex processing that would, in humans, produce the same behaviors we're seeing from her.

We are human and we trained ChatGPT on lots of our language so it can sound like us. Understanding what ChatGPT is actually doing prevents us from thinking it has a higher intelligence than it does. We can therefore use the tech in the ways that will be most beneficial, and importantly not rely on it to do things it cannot do.

Give the article a read and let me know what you think.

https://lnkd.in/gRzfm7h9

Sources:
[Link]