BABL AI is excited to announce that Ali Hasan, BABL AI’s Senior Advisor, has authored a fascinating blog for the American Philosophical Association. The blog, titled “Are You Anthropomorphizing AI?,” explores the increasing tendency to anthropomorphize advanced AI systems like ChatGPT, attributing human-like understanding and intentions to machines that, in reality, merely mimic meaningful communication through statistical patterns.

In his article, Hasan says while this phenomenon isn’t new, the development of more advanced AI systems has intensified it. In previous decades, computer systems lacked the capabilities to compete with human intelligence outside specific tasks like calculations or games. As a result, people generally recognized these systems as mere tools despite using personifying language for convenience. However, with the advent of generative AI systems like ChatGPT, the boundary between machines and perceived human-like behavior has become blurred.

The article emphasizes that while AI like ChatGPT doesn’t have true understanding, sentience, or consciousness, it is very good at mimicking meaningful human communication through “meaning-semblant” behavior. This creates a strong but misleading impression that AI systems have cognitive states or intentions similar to humans. Even people who intellectually recognize that AI lacks genuine understanding may find themselves using language that implicitly anthropomorphizes these systems, describing them as “thinking,” “understanding,” or “interpreting” information. 

Hasan began teaching anthropomorphism in his Ethics and Technology course a few years ago. Since then, he’s observed that certain aspects of anthropomorphism and AI personification are often overlooked or not carefully distinguished. When asked what the average person, with limited education or understanding of the subject, can do, he emphasized the importance of education and communication. 

“Ordinary people need to better learn, in an accessible way, how these systems work and how they can fail to work, and what the potential risks to them are. Relatedly, we need to communicate more clearly and effectively about these systems, and design them in ways less likely to mislead or confuse us about their abilities, and their personal or moral status,” said Hasan.

You can read the full article HERE.