There is lots of buzz recently about the soul of AI machines. Yet while this discussion elucidates the cleverness of AI applications, it ignores one key aspect of thought: intentionality. The soulful person’s thought is intentional. There is some driving impetus behind it, a problem to be solved, an observation that worth exploring, an emotion that needs to be expressed. The AI machine, by contrast, follows instructions and “thinks” on command. That is what machines can do. There is no denying that some applications are really clever and that their responses are uncannily human-like. But if you look at them closely, those responses are all remixes of things the machine was taught in the past. And the human interlocutors who induce these responses are skillful in their questioning to enhance the human-like character of these responses. Consider the oft-cited “painting by Grant Wood of an astronaut couple “American Gothic”-style.” On the face of it, this is hugely creative. Grant Wood never drew astronauts so this is surely original content. However, there are many images of astronauts that machines can use to learn what astronauts look like, and there are many signal elements of Grant Wood paintings that can be learned. The task is made even easier by naming a specific painting which has a layout and other features that can be learned. The machine takes three learned models and mashes them up. Nothing new to see here folks, but a very clever output nonetheless.
The real test of the presence of AI soul would be evidence of intentionality on the part of the machine. What does it produce if the instruction is simply “draw” or even “start?” What kind of conversations would it have if the human user simply said “let’s chat” and only responded with comments like “I see” or “interesting” or “hmm?” Would it ever initiate a conversation or draw because it felt like it? Absent this internal intent these outputs are soulless. The cynic in me would say we are confused by this issue because so many humans live their lives without intentionality. Their conversations are echoes of what they have heard, cleverness stands in for thoughtfulness. They would enjoy a cocktail party with AI-driven robots. The optimist in me would say we should be really excited by just how clever these outputs are and how impressive the humans who created them are and what promise they show for conversational technology in well-defined situations. The philosopher in me says we do not yet have reason to worry about the soul of a machine.