We Need to Talk About How We Talk About 'AI' (www.techpolicy.press)

In a recent article, Emily M. Bender and Nanna Inie critically examine the prevalent practice of anthropomorphizing artificial intelligence (AI) in public discourse. They argue that labeling AI systems with human-like attributes, such as "reasoning", "understanding", or even "friend", can obscure their actual limitations and capabilities. This misleading language risks fostering misplaced trust and accountability, particularly as people may develop emotional ties to these technologies, mistakenly attributing human characteristics to machines that simply process data probabilistically. The authors emphasize the need for precise language when discussing AI to promote better public understanding and accountability. They advocate for shifting from anthropomorphic terms to clear descriptions of functionalities and roles of AI systems, focusing on what these systems do rather than what they supposedly possess in terms of intelligence. This shift in narrative is crucial not only for improving AI literacy but also for safeguarding vulnerable populations who may be more susceptible to the deception inherent in anthropomorphic language. By refining the conversation around AI, the authors suggest we can empower users and encourage more responsible technology development and usage.