Sorry, ChatGPT but you've got Zero Personality

… and that will always be the case. Yes, we’re living during a time when unprecedented resources are being devoted to research in artificial intelligence. Somehow I still remain skeptical about the goals we’re attempting to achieve. I’m particularly skeptical regarding claims about Artificial General Intelligence, or anyone claiming to build AI that’s as smart as (or smarter) than a human. Maybe I’m being stubborn, or maybe this attitude of mine is due to my education in philosophy. I know that I’m forming this opinion in direct opposition to some of the leading philosophers of our time. It’s just how my beliefs happen to line up. Hopefully, by the end of this article, you’ll understand why.

THOUGHT EXPERIMENT

To help illustrate my perspective, I’ll ask you to engage in a thought experiment. This is something philosophers love to do. Don’t worry, thought experiments are free! You don’t have to go anywhere, or install a new app. You don’t even have to sign-up for an account. I do recommend that you not operate any heavy machinery while engaging in philosophical thought experiments. Are you ready? OK, picture a fruit fly. You know when you leave fruit like bananas, pears, or grapes lying out for a while and some little bug starts flying around and around it? That’s the one I’m talking about. Its main goal in life is finding fruit and eating it. Relative to humans, these flies are pretty small and yet they have brains. Even though it seems like they don’t do much, their brains are incredibly complex. (If you want to know exactly how complex, then check out this paper published by eLife.)

You’ve got this image in your mind, or maybe you’ve actually got the real thing somewhere nearby. There’s some little fly buzzing around a piece of fruit, using its incredibly complex brain to determine the best approach. Does this fruit fly have a personality? Honestly, I’m kind of agnostic regarding whether or not fruit flies have personality. For the purposes of this thought experiment, we’ll imagine in the affirmative. What kind of personality could a fruit fly possibly have? I think it’s safe to assume that the fruit fly’s incredibly complex brain would serve as the nexus for any kind of personality. Whatever biological hardware happens to be responsible for a given fruit fly’s personality, we’ll also assume that we can measure aspects of this personality via the fruit fly’s behaviour. For example, a specific fruit fly might use a particular strategy of finding fruit based on some genetic predisposition combined with the unique conditions of its current environment. If fruit flies have personalities at all, then it makes sense that such a personality would provide the fruit fly with an advantage in survival.

Alright, we’ve got some idea of the rudimentary personality that we’re willing to grant a fruit fly. It’s measurable according to behaviour patterns that depend on beneficial, physiological fruit fly traits. Now imagine one of your favourite humans, with all of the deep complexity that shapes their own unique personality. Picture how this human’s personality has formed over the course of their lifetime. Think about how this human’s personality allows them to respond to their environment. Think about your own personality and how you form patterns of behaviour in response to your environment. What would happen if your physical environment was completely changed, or removed all together? We know that changes in our environment affect our behaviour. After all, our behaviour exists to help us better cope with changes in our environment. If personality is a collection of behaviour patterns, then it makes sense that these patterns are affected by environmental changes.

What happens when we try to imagine a personality that doesn’t respond to its physical environment? It’s difficult to consider because it’s as though we’re expecting the personality to exist outside of time and space. If there’s anything we’ve learned so far about personalities, it’s that each one is attached to a body that exists in a very specific time and space. When you remove either a personality’s environment or its body, we're not too sure what we’re left with…

Maybe you’re thinking, “The problem is with a human personality. Those are really complex. Lets go back to the fruit fly!“ Yes, maybe if the personality (or patterns of behaviour) are a little simpler then it’ll be easier to imagine them existing without an environment. We’ll return to our old friend the fruit fly, buzzing around and looking for fruit. Then we’ll remove the fruit, and the physical buzzing around… Hmm… What happens if we take this thought experiment all the way and imagine a tiny robotic fruit fly. It’s got incredible hardware that matches every single function of the fruit fly brain. We can set it loose in a fruit fly environment and say, “Hey! We’ve built an AI with personality!“ Unfortunately, some philosopher will probably come along and point out that our robot fruit fly isn’t actually alive. If that robot isn’t alive, then how can our intricately constructed personality serve to convey an advantage for its survival? At the most, we might have built a personality that provides an advantage for our robot’s continued existence. That existence would likely not depend on the consumption of fruit, which was our original fruit fly’s whole goal in life. Now we’re stuck with a robotic fruit fly personality that doesn’t depend on its environment in the same way that the real fruit fly does.

IN CONCLUSION

Allow me to summarize the journey we’ve just taken. We started with a fruit fly whose entire existence depends on its ability to survive in a given environment. We granted the fruit fly a personality, exhibited via behaviour patterns that exist to increase its probability of survival. Then we decided to build a model of this personality by constructing a tiny, robotic fruit fly. At the end of the day, we’d be stuck with a model of the physiology required for the fruit fly’s continued evolution. Such a model has no need for survival. In other words, our robotic fruit fly will continue to exist regardless of whether it’s successful at survival. This non-living property of machines is fundamentally at odds with the very existence of personalities.

We can remove the environment in which living personalities typically operate, but then it’s difficult to imagine how artificial personalities would exhibit behaviour. They would have to do so in an ageless way that isn’t specific to any particular time or place. That’s not how personalities work! On the other hand, if we try to build a robot that exhibits fruit fly behaviour in a real environment then we’re expecting our robot to depend on that environment for survival. Once again, we’re stuck with something that doesn’t look much like a personality. We seem to be at an impasse with our conception of machines that have personality.

Lets take a page out of Kate Linebaugh’s playbook, and ask ChatGPT about its own capacity to exhibit human-like behaviour. When I did that, here’s the response ChatGPT gave me: “ChatGPT has limitations, including… a lack of true understanding or consciousness… it’s important to recognize the inherent challenges in creating machines that fully replicate human intelligence and understanding.“

In conclusion, ChatGPT’s distinct lack of personality precludes any necessity for me to owe it an apology. I hope you’ve enjoyed this philosophical journey into thought experiments. May your personality continue to benefit from reading my articles.