Will AI machines think freely?
Well, do humans think freely?
If someone asks you to describe a chair, how would you do it? You and I might describe it differently, but we’d know what each was talking about.
AI can also describe a chair, and it does a great job. Here is the beginning of how Google’s Bard describes a chair:
“A chair is a type of furniture with a raised surface supported by one or more legs, typically used to seat a single person.”
Google Bard
(Then it goes on to describe different types of chairs, uses, materials, settings, etc.)
But would AI know a chair if it saw one?
You and I would even if the chair wasn’t typical. We might come across a ledge, a tree stump, or a car bumper shaped and sized just right for us to park our butts on. A chair, if not in the traditional sense.
Because we understand “chairness.”
If we want to sit, we scan all of the objects and formations around us, looking for something to serve as a chair, if not a specific-built chair. If we don’t see a human-made chair, we think about what we could use as chair. Our thoughts are guided by our intention.
If an AI wanted to sit, could it recognize chairness in the objects and formations around it? Could it think about what could be used as a chair?
Would an AI ever think to itself, “I’d really like to sit down.”
