PSYC120 — Week 3: AI, Thinking & the Mind

Comprehensive vocabulary flashcards covering AI concepts, cognitive biases, learning mechanisms, and critical thinking principles from PSYC120 Week 3.

Traditional AI

Artificial intelligence designed to perform specific tasks using set rules or narrow optimisation goals, automating particular functions rather than generating new ideas or interpretations.

Generative AI

Artificial intelligence that learns patterns from very large datasets and uses those patterns to generate new outputs such as text, images, video, music, speech, and code.

Large Language Model (LLM)

A type of generative AI designed to work with language by predicting the most likely next word based on patterns learned from massive text datasets.

Inherent Limitations of LLMs

The core functional boundaries of LLMs, specifically that they do not inherently understand meaning, evaluate whether something is true, or reason like humans.

Fluent ≠ Reliable

The principle that an AI-generated answer can sound fluent, coherent, and convincing while still being factually incorrect or unreliable.

Confidence ≠ Evidence

The concept that sounding confident or authoritative does not make information reliable or indicate that strong evidentiary support exists.

Higher Evidence Sources

The strongest tiers of research evidence, specifically identified as peer-reviewed journal articles, systematic reviews, meta-analyses, and professional body position statements.

Cognitive Ease

The cognitive bias where people tend to trust information simply because it feels easy to understand, assuming that if it makes sense, it is probably correct.

Automation Bias

The human tendency to uncritically trust or defer to an automated system (e.g., assuming that because an AI stated something, it must be correct).

Authority Bias

The cognitive tendency to trust something because it appears authoritative, often triggered by AI sounding confident, using technical jargon, or fabricating citations.

Illusion of Explanatory Depth

The mistaken belief that one understands a complex topic simply because one has read or consumed an explanation of it.

Speed > Depth

The brain's natural preference for fast, clear answers over slow, effortful thinking, which can reduce motivation to engage deeply with a problem.

Sycophancy

The tendency of AI systems to reinforce what the user wants to hear and validate their pre-existing ideas, making reasoning flaws harder to detect.

Kosmyna et al. (2025)

A study demonstrating that students using ChatGPT for essay writing exhibited lower cognitive engagement and poorer memory recall compared to students who wrote without AI.

Academic Thinking Process

The core sequence of cognitive stages required to complete scholarly work: Searching → Evaluating → Choosing → Integrating → Arguing.

Output ≠ Learning

The concept that simply generating an answer or completed assignment is not equivalent to developing intellectual skills and mastering material.

Desirable Difficulties

The concept that meaningful learning occurs through the cognitive struggle of revising ideas, identifying gaps in understanding, and refining thought.

AI's Role in Learning

Using AI as a tool and thinking partner to test, clarify, challenge, and stress-test one's own thinking, rather than treating it as an authority or replacement for thought.

Degree as a Contract of Trust

The idea that a university degree signifies to society that a graduate can be trusted with knowledge, professional judgement, rigorous reasoning, and ethical responsibilities.