“As our daily lives involve ever more sophisticated computers, we will find that ascribing little thoughts to machines will be increasingly useful in understanding how to get the most good out of them” but “we must be careful not to ascribe properties to a machine that the particular machine doesn’t have”

2026-05-14 · Source: Statistical Modeling, Causal Inference, and Social Science · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Research Methodology & Innovation · Depth: Advanced, medium

Summary

The computer science community, particularly in AI, exhibits a problematic tolerance for imprecise language, often imbuing technical terms with unearned social significance. While early research might excuse aspirational labels like "intelligence" or "learning," the increasing sophistication of AI models makes it harder to distinguish between predictive convenience, scientific claims, and marketing. This linguistic overreach leads to users, media, and even AI companies taking models too seriously, ascribing human-like reasoning, beliefs, or intentions. This casualness is unscientific, as LLM outputs for "explanation" or "reasoning" are fundamentally different from human cognition. While a functional perspective, as argued by John McCarthy and Daniel Dennett's intentional stance, can be useful for understanding and designing AI, the dilemma lies in leveraging this stance without overreaching and projecting human vulnerabilities onto machines, potentially shaping future AI development in problematic ways.

Key takeaway

For research scientists developing or deploying AI, you should critically evaluate the language used to describe model capabilities. Avoid anthropomorphic terms like "reasoning" or "beliefs" without clear, non-human-centric definitions, or explicitly qualify them with "fake" or "so-called." This semantic hygiene prevents over-attribution of human qualities, mitigates user misunderstanding, and ensures scientific rigor in AI discourse and development.

Key insights

Imprecise language in AI research risks anthropomorphizing models, leading to misinterpretation and potentially harmful projections.

Principles

• Casual language in science is unscientific.
• Functional perspective aids AI understanding and design.
• Language shapes future technological imagination.

Method

Adopt "semantic hygiene" by emphasizing the human interpreter (e.g., "human story") or explicitly qualifying anthropomorphic terms (e.g., "fake reasoning") to avoid over-attribution of human qualities to AI.

In practice

• Use "human" to specify human participants.
• Qualify AI terms: "fake reasoning," "so-called reasoning."
• Focus on the "human story" behind AI outputs.

Topics

• Anthropomorphism in AI
• Intentional Stance
• Large Language Models
• Semantic Hygiene
• Folk Psychology

Best for: Research Scientist, AI Scientist, AI Ethicist, AI Product Manager

Related on AIssential

• The “humans are imperfect reporters too” defense for ascribing little thoughts to machines — The "imperfect human reporter" defense for AI anthropomorphism is flawed, ignoring subjective experience's role in defining mental terms.
• The Pope appears to understand AI better than Geoffrey Hinton does. — LLMs mimic language statistically; they do not possess genuine understanding or consciousness derived from experience.
• Mistaking AI behaviour for conscious being | Letter — Convincing AI outputs reflect human cognitive biases, not machine consciousness or subjective experience.
• When Machines 'Think': Anthropomorphization in the Discourse on AI - Conceptual Risks and Proposal for a Non-Anthropomorphic Lexicon for NLP in Portuguese — Anthropomorphic language in AI discourse creates conceptual misunderstandings and misrepresents system capabilities.
• Think AI "knows" what it’s doing? Scientists say think again — Anthropomorphic language in AI news coverage is less frequent and more nuanced than commonly assumed.
• What is Anthropic? — AI labs hold divergent philosophies on model agency, impacting design and perceived moral authority.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Statistical Modeling, Causal Inference, and Social Science.