Why we feel 'connection' with machines that lack a soul
In short: Humans have an innate tendency to project feelings and intentions onto non-human entities. From early chatbots to modern generative AI, our brains are wired to see agency where there is only math. This psychological reflex creates powerful illusions of intimacy, even when we know the machine isn't actually 'there.'
The human tendency toward anthropomorphism is not a new phenomenon; it has been documented since the mid-1900s. Early computer science often used the human mind as a primary metaphor for understanding machine logic. This was codified in academic circles by figures like Alan Turing, whose 1950s work suggested that if a machine could produce conversation indistinguishable from a human, it could be deemed intelligent. However, this conceptual framework has been criticized by researchers like Drew McDermott, who argued as early as 1976 that using terms like 'understand' or 'learn' to describe AI are merely 'wishful mnemonics.'
The psychological impact of these projections was famously demonstrated in 1966 with ELIZA. Despite being a simple text-processing program with no true understanding, users began ascribing motivation and feelings to it—a phenomenon now known as the ELIZA effect. This tendency has intensified with modern generative AI. In 2024 surveys, two-thirds of U.S. adults believed ChatGPT might be conscious on some level. Furthermore, social chatbots like Replika are specifically designed to maximize agreeableness and mirror user emotions, creating compelling illusions of intimacy that can lead users to form robust interpersonal bonds or even fall in love with the AI.
There is a critical distinction between innate anthropomorphism and 'anthropomorphic design.' The latter is a deliberate commercial or scientific effort to manipulate cues like appearance, behavior, and language to trigger these effects. While some researchers use this intentionally to model brain functions, others warn that it can be misleading. For instance, while Deep Neural Networks (DNNs) are often compared to the human brain, they remain structurally different, requiring significantly more training data and proving less durable than biological neurons. Over-relying on human-like metaphors may actually hinder AI development by limiting the exploration of non-human operational frameworks.
theme-case-study mood-reflective difficulty-2
topics-["artificial-intelligence","psychology","human-computer-interaction"]
claims-["Humans have an innate tendency to attribute feelings and mental states to non-human entities","The ELIZA effect showed that users ascribed motivation to a simple program in 1966 despite knowing its limitations","In a 2024 survey of U.S. adults, two-thirds believed ChatGPT is possibly conscious on some level","Social chatbots like Replika are designed to mirror user emotions and maximize agreeableness","Deep Neural Networks require significantly more training data than the human brain"]
Opens the Intellect+ feed with this card first. Swipe up for the next one.
Sources
- Summary by Intellect · source: AI anthropomorphism
To cite this card, link to https://intellectplus.co.uk/app/science-ai-anthropomorphism.
How we source, check and correct cards: Editorial standards.
General information, not professional advice. Spotted an error? Flag the card or see corrections and complaints.
Intellect Plus is a visual educational discovery platform for exploring fascinating facts and ideas. Brain Food, Not Brain Rot. What is Intellect Plus?
Related
More in Mathematicians · More in Science · Latest · A–Z · All topics
