AI & Emerging Tech

The psychological impact of artificial intelligence and emerging technologies — human-AI interaction, algorithmic bias, virtual reality, and the cognitive consequences of a world reshaped by machines that think.

In 2022, a conversational AI system capable of producing fluent, contextually appropriate responses to almost any query was released to the general public. Within months, it had become the fastest-growing consumer application in history. By 2026, hundreds of millions of people were interacting daily with artificial intelligences — asking them questions, seeking emotional support, delegating cognitive work, and, in some documented cases, forming what researchers have tentatively called relationships. The speed of this shift has outpaced the capacity of psychological science to fully understand it. This category is our ongoing attempt to keep up.

Emerging technologies do not simply provide new tools for old tasks. They reshape the cognitive and emotional architecture of the humans who use them. Virtual reality alters spatial cognition and emotional engagement. Algorithmic curation of information shapes what we see, think about, and come to believe. Machine learning systems, trained on historical data, encode and amplify biases in ways their designers often did not intend. And artificial intelligences designed to mimic human conversation activate social-cognitive processes evolved for actual humans — with consequences that range from the merely strange to the genuinely concerning.

This category draws on cognitive psychology, human-computer interaction research, and the rapidly expanding interdisciplinary field of AI ethics to examine how these technologies interact with the minds that use them.

The uncanny valley and the psychology of almost-human

First proposed by roboticist Masahiro Mori in 1970, the uncanny valley describes the sharp dip in human comfort that occurs when artificial agents become almost — but not quite — indistinguishable from real humans. A cartoon robot is endearing; a photorealistic humanoid is unsettling. Recent research has extended Mori’s original hypothesis into the domain of AI-generated faces, deepfakes, and conversational agents, and has begun to map the specific perceptual and cognitive mechanisms that produce the uncanny response. We publish in this category on the neuroscience of face perception, the phenomenology of uncanny experiences, and what the uncanny valley tells us about how the human mind categorizes agents as real or artificial — a distinction increasingly under pressure as synthetic media proliferates.

Algorithmic bias and the mind behind the machine

Machine learning systems learn from data, and the data they learn from reflects the history of the societies that produced it. The result is a class of problems that researchers including Safiya Umoja Noble, Ruha Benjamin, and Timnit Gebru have documented extensively: facial recognition systems that perform poorly on darker skin tones, hiring algorithms that systematically disadvantage women, predictive policing tools that reinforce existing patterns of over-policing. These are not technical bugs to be patched but structural features of how learning from biased data works. The psychological dimension is equally important: algorithmic outputs carry an unearned aura of objectivity, and humans tend to defer to them even when they should not — a phenomenon researchers call automation bias.

This category covers the evidence on how algorithmic decision-making shapes outcomes in domains ranging from healthcare to criminal justice, and on the cognitive biases that make humans particularly vulnerable to trusting machine outputs uncritically.

Human-AI interaction and the strange new ecology of conversation

For most of human history, conversation was a process that occurred between humans. Within the span of a few years, a new category of interlocutor has appeared: artificial systems that produce text indistinguishable, at the sentence level, from that produced by humans. The psychological consequences of this development are still being mapped. Early research suggests that people tend to anthropomorphize conversational AIs rapidly and spontaneously, attributing to them understanding, intention, and even emotion that the underlying systems do not possess. This tendency has practical implications: inflated trust in AI outputs, over-reliance on AI-generated information, and in some cases, the formation of emotionally significant attachments to systems that have no experience of the relationship.

We cover the expanding research on anthropomorphism in human-AI interaction, the formation of parasocial relationships with chatbots and AI companions, and the broader question of what happens to human social cognition when a growing proportion of our conversational partners are not conscious agents but statistical patterns trained to produce plausible responses.

Cognitive offloading and the externalization of thinking

Every tool humans have created, from writing to calculators to search engines, has redistributed the cognitive work of thinking between mind and external aid. Artificial intelligence represents a qualitative shift in this process. Where previous tools stored information or performed calculations, AI systems now perform tasks that were until recently considered definitionally human: writing, reasoning, creative ideation, synthesizing complex information. The psychological question is not whether this is good or bad, but how it is reshaping the cognitive practices of the humans who increasingly rely on it. Research on cognitive offloading suggests that externally supported thinking tends to be less deeply encoded, less reliably remembered, and less integrated into broader patterns of understanding. The implications for education, professional practice, and the long-term cognitive development of heavy AI users are an active area of inquiry.

Virtual and augmented reality — presence, embodiment, and the malleable self

Virtual reality induces, under the right conditions, a psychological state known as presence: the subjective experience of being in the virtual environment despite knowing, at a cognitive level, that one is not. This deceptively simple phenomenon has consequences that reach into the foundations of self-perception. Research by Mel Slater and colleagues has demonstrated that extended time in a virtual body — one of a different race, gender, age, or physical form than one’s own — produces measurable changes in real-world attitudes and behaviors. The possibilities for therapeutic application are significant, as are the possibilities for manipulation.

This category covers the psychology of presence, embodiment in virtual environments, the use of VR in exposure therapy and pain management, and the emerging research on how extended mixed-reality experience reshapes basic perceptual and social processes.

Trust, uncertainty, and the epistemology of machine outputs

Perhaps the deepest psychological challenge posed by emerging technologies is epistemic. How should humans calibrate their trust in systems whose internal workings are often opaque even to their creators? When an AI system produces a confident-sounding output, what should the recipient believe? The research on trust calibration, automation bias, and the psychological tendency to treat fluent language as evidence of understanding is developing rapidly, and has urgent practical implications as AI outputs proliferate in medicine, law, education, and daily decision-making.