AIs are not as powerful as people think they are. And this is a problem.

\-----<br><br>The problem I see with AIs is not that they are smart. They aren't. The problem is that they are perceived to be super smart, when they are not.<br><br>The reason I bring this up is that when I see someone meet an AI with a certain amount of skepticism, their first reaction is to try to convince me that AIs are super smart. And this goes on for a while. Eventually, this leads to an acknowledgment that either AIs aren't that smart, or that I'm not very smart. But I think the first reaction should be to look at why we have this mismatch in perception.<br><br>AIs are not that smart. They are not human. They are not general-purpose intelligence. They are not creative, they don't have a sense of humor, they don't have empathy, they don't have common sense, they don't really understand language. They are an algorithm that works in a certain set of contexts to produce certain outputs.<br><br>I'm sure you know someone who thinks otherwise. The writers of The Futurist think AIs are so smart that we can't compete with them anymore. The CEO of Google thinks AIs are basically general-purpose intelligence and this makes him excited. The CEO of Amazon thinks AIs are already as smart as humans and now it's time for the next phase of merging AIs with humans. As if this is a debate in which different opinions can coexist, and the CEO of Google's opinion is just as valid as anyone else's. AIs are not general-purpose intelligence, they are not as smart as humans, and they certainly don't need to be merged with us. They are not as smart as the average high schooler. Some people know this. But the majority of people seem to have the perception that AIs are way smarter than humans, and this perception is crazy.<br><br>I think basically every other problem that we see today stems from this. That The Futurist thinks AIs are so smart that we can't compete anymore. That Google is so excited about OpenAI that it basically copied all of their model and spent tens of billions of dollars on it. That Meta put all of its eggs in the Llama basket and didn't even bother to train it to be accurate. That Facebook fired all of its journalistic integrity team and replaced them with AIs. That Amazon's CEO thinks it's okay to send a poorly trained AI model to people's homes and essentially replace their natural language with a commercialized version.<br><br>I think this basically comes down to a lack of understanding of how AIs work, and how they don't work. I'm sure every one of you has had a debate with a friend about how AIs are not that smart, and the friend just says that they have seen evidence that shows that AIs are really smart. As if this is a debate that needs to be won by more evidence. The problem is not that we have differing opinions on AI's intelligence. The problem is that we have differing perceptions of reality.