A Comparative Glossary
This table follows familiar words from human experience into AI. The fourth column contains the AI’s comments. I’ve left the last column open for my own observations.
1. Architecture and organization
| Term | Human brain | Artificial intelligence | Comparison (AI commentary) | My notes |
|---|---|---|---|---|
| Model | Internal picture of the world and accumulated experience. | Trained network encoding patterns in data. | The human world model is part of the brain. The AI model is usually the central computing component. | |
| Neural Network | Biological neurons and changing synapses. | Artificial units, weights, and mathematical operations. | Shared historical inspiration, different mechanisms and training. | |
| Agent | Perceives, chooses goals, and acts. | A model with state and tools that acts on tasks. | Both observe, decide, and act. AI goals usually come from outside. | |
| Tool | Body, speech, objects, and technology. | APIs, browsers, code, databases, and apps. | Software tools usually have explicit interfaces. | |
| Orchestration | Coordinates perception, attention, memory, and action. | Coordinates steps, agents, state, and tool calls. | AI orchestration can be explicitly programmed. The brain has no single controller. | |
| Multi-Agent | People with roles who exchange information. | Agents divide up functions. | People also have relationships and interests of their own. | |
| Parallel Processing | Many processes run simultaneously. | Parallel computations and independent workflows. | The architecture and cost of coordination differ. | |
| Autonomy | Can form and change goals. | Pursues goals within constraints and permissions. | Compare the origin of goals and the limits on independence. | |
| System | Brain, body, and environment. | Model, memory, tools, interfaces, and rules. | Define the systems before comparing them. |
2. Memory and knowledge
| Term | Human brain | Artificial intelligence | Comparison (AI commentary) | My notes |
|---|---|---|---|---|
| Short-Term Memory | Briefly holds recent information. | Current context and temporary state. | A context window has different limits from human short-term memory. | |
| Working Memory | Holds and actively processes information. | Context, scratch space, and intermediate results. | AI can keep intermediate material externally and verbatim. | |
| Long-Term Memory | Knowledge and experience retained for years. | Model parameters and persistent external stores. | Weights and editable records are distinct kinds of storage. | |
| Episodic Memory | Life events situated in time and place. | Conversation, task, and event logs. | Recording an event does not establish a lived autobiographical experience. | |
| Semantic Memory | Concepts, facts, and general knowledge. | Knowledge in weights and knowledge bases. | Both use facts. The model may be unable to identify its source. | |
| Procedural Memory | Practiced skills and habits. | Learned policies, skills, and stored procedures. | Both reproduce skills. Biological movement also involves bodily feedback. | |
| Recall | Reconstructs memories from cues. | Retrieves records or generates answers from learned patterns. | AI can read exactly or produce a false recollection. | |
| Forgetting | Loss of access, interference, and reorganization. | Truncation, deletion, and outdated knowledge. | Causes and controllability differ. | |
| Attention | Selectively processes important signals. | Mathematical attention weights relationships between representations. | Transformer attention differs from psychological attention. | |
| Association | One idea evokes others through experience. | Semantic proximity and statistical relationships. | Similar functions need not share a neural mechanism. | |
| Context | Situation, physical state, experience, and surroundings. | Available tokens, instructions, and data. | Human context is broad and continually updated through the world. | |
| Physical Time | Time passes and the body ages. | Compute time, scheduling, and timestamps. | Measurement need not imply experience. | |
| Time Perception | Estimates durations and intervals. | Measures, calculates, or estimates intervals. | Estimating duration differs from feeling time pass. | |
| Subjective Time | Time feels faster or slower. | A subjective sense of duration is unconfirmed. | The question remains open. | |
| Autobiographical Time | Personal history connects past, present, and future. | An agent’s chronology and persistent memory. | Narrative structure alone cannot demonstrate lived experience. | |
| Temporal Continuity | A living organism persists through change. | A process may stop, move, and restart. | Define what counts as the same agent. | |
| Mortality Awareness | Understands the inevitability of death. | Can model permanent shutdown or loss of memory. | Representing an ending differs from fearing it. |
3. Learning and change
| Term | Human brain | Artificial intelligence | Comparison (AI commentary) | My notes |
|---|---|---|---|---|
| Learning | Experience changes behavior and skills. | Weights change, or context and memory guide adaptation. | Distinguish weight updates from adaptation with fixed parameters. | |
| Training | Practice and reinforcement of skills. | Parameter optimization on training data. | Feedback plays a role in both, through different processes. | |
| Fine-Tuning | Specializes in a field. | Further tuning of pretrained weights. | Human specialization has no separate global weight-update procedure. | |
| Reinforcement Learning | Consequences and rewards shape behavior. | A reward signal guides policy optimization. | Human motivation is richer than a single reward function. | |
| Feedback | Evaluates actions and adjusts behavior. | User, test, and environmental evaluations. | Feedback need not immediately change weights. | |
| Transfer Learning | Applies previous skills to new problems. | Reuses pretrained representations in another domain. | The functional analogy is close. | |
| Generalization | Applies principles beyond familiar situations. | Performs on new examples outside the training set. | A meaningful shared characteristic of intelligence. | |
| Adaptation | Behavior and skills change over time. | Context, memory, tools, or model updates change behavior. | AI can adapt without retraining. | |
| Neuroplasticity | Physical reorganization of brain connections and functions. | Changes to weights, structures, or routes. | A partial analogy, given the biology of brain plasticity. | |
| Consolidation | Stabilizes and reworks experience, including during sleep. | Summarization, fact extraction, and memory updates. | Architectural resemblance need not imply identical mechanisms. | |
| Functional Learning | Uses new experience in later decisions. | Updates to context or persistent memory change later answers. | Functional acquisition of knowledge can occur with fixed weights. | |
| Model Upgrade | Acquires knowledge through development and learning. | A new version may have different weights and capabilities. | An upgrade need not retain an individual agent’s memories. |
4. Thinking and quality control
| Term | Human brain | Artificial intelligence | Comparison (AI commentary) | My notes |
|---|---|---|---|---|
| Reasoning | Logic, experience, mental models, and intuition. | Multistep inference, search, and tools. | Results may match. Reliability and process depend on the task. | |
| Planning | Selects actions and considers consequences. | An agent plans and adjusts steps. | Similar functions under different goals and constraints. | |
| Problem Solving | Combines knowledge, trials, analysis, and creativity. | Generates, tests, and selects options. | AI can scale up the number of attempts. | |
| Decision Making | Weighs emotions, values, needs, and evidence. | Uses probabilistic forecasts, rules, and optimization. | A personal stake in outcomes has not been established for AI. | |
| Intuition | Fast judgments from implicit experience. | Fast responses from learned patterns. | Both may reply without an explicit chain of explanation. | |
| Metacognition | Assesses knowledge and mistakes. | Checks, critiques, and uncertainty estimates. | Self-checks do not guarantee well-calibrated confidence. | |
| Reflection | Reconsiders thoughts, actions, and experience. | Reanalyzes and may revise a decision. | A similar outward act need not involve a personal experience. | |
| Uncertainty | Doubt and a sense of missing information. | Probabilities, uncertainty measures, and acknowledgment of gaps. | A confidence score differs from subjective doubt. | |
| Bias | Heuristics, emotion, and accumulated experience. | Data, training, and algorithmic skews. | Both can err systematically for different reasons. | |
| Hallucination | Clinically, perception without a corresponding stimulus. | Invented or unsupported information presented confidently. | The term is metaphorical. Confabulation is often a closer human analogy. | |
| Grounding | Checks ideas against sensory experience and facts. | Checks documents, APIs, search, and measurements. | Both verify. Human contact with the world is more continuous. | |
| Relevance | Selects what matters for a goal and situation. | Assesses information against a query. | Human relevance also depends on needs and emotions. | |
| Common Sense | Implicit practical knowledge of the world. | Statistically learned everyday patterns. | AI can sound convincing yet miss obvious physical or social facts. |
5. Perception and interaction
| Term | Human brain | Artificial intelligence | Comparison (AI commentary) | My notes |
|---|---|---|---|---|
| Perception | Sensory organs and ongoing interpretation. | Processing text, images, audio, and sensor data. | Some modalities overlap. Most AI lacks bodily signals. | |
| Multimodality | Integrates sight, hearing, touch, and other senses. | Processes several data formats jointly. | Both combine signals from different sets of sensors. | |
| Embodiment | Cognition develops through interaction with the body. | Ordinary LLMs lack bodies. Robots have sensors and actuators. | A major difference for physical common sense. | |
| Language | A medium of thought, communication, and culture. | A primary interface and training signal for LLMs. | Human thinking extends beyond language. | |
| World Model | Causal and spatial representations of the world. | Learned representations, sometimes an environment simulator. | Both predict. AI may lack direct physical experience. | |
| Prediction | Anticipates sensory events and consequences. | Predicts tokens, states, or events. | Prediction is a basic similarity, with different targets. | |
| Causality | Infers causes from observation and action. | Causal reasoning and specialized models. | Both can mistake correlation for causation. Experiments improve reliability. |
6. Personality, motivation, and consciousness
| Term | Human brain | Artificial intelligence | Comparison (AI commentary) | My notes |
|---|---|---|---|---|
| Personality | Relatively stable traits and behavior. | Instructions, style settings, and learned patterns. | Similar behavior need not imply equivalent personal development or inner experience. | |
| Identity | Autobiographical continuity of the self. | Name, role, instructions, and persistent history. | Technical continuity need not involve subjective continuity. | |
| Self-Awareness | Awareness of personal thoughts and states. | Self-description and state monitoring. | Self-description cannot establish subjective awareness. | |
| Consciousness | Subjective experience. | Whether AI has subjective experience is unknown. | Behavioral tests alone cannot settle the question. | |
| Feelings | Felt emotional and bodily experiences. | Describes, recognizes, and models feelings. | Expressing a feeling differs from experiencing it. | |
| Emotions | Physiological responses and appraisals that shape choice. | Classifies emotions and expresses them in replies. | Similar behavior does not establish a biological emotional state. | |
| Empathy | Understands and sometimes shares another’s feelings. | Recognizes cues and responds supportively. | Cognitive empathy can be simulated. Shared affect is unconfirmed. | |
| Motivation | Needs, desires, interests, and social goals. | Assigned goals, rewards, and priorities. | Motives differ in origin and subjective significance. | |
| Curiosity | A desire to discover and explore. | Exploration policies and searches for new data. | Human curiosity includes a felt experience. | |
| Will | Self-regulation and overcoming obstacles. | Continues tasks and revises plans. | System-defined persistence does not establish felt willpower. | |
| Moral Values | Socially and personally formed principles. | Learned preferences, instructions, and norms. | Moral judgments need not imply personal moral motivation. | |
| Guardrails | Self-control, conscience, norms, and laws. | Rules, filters, permissions, and policies. | Human self-regulation often runs deeper into motivation. | |
| Free Will | The nature of human choice remains disputed. | Algorithms, data, and goals determine choices. | An unresolved philosophical question cannot serve as a proven distinction. | |
| Emotion Appraisal | Appraises events relative to needs and goals. | Evaluates events against goals and constraints. | Appraisal and regulation may be similar. Experience remains an open question. | |
| Subjective Experience | The felt aspect of a state, reported in the first person. | Felt experience has not been established. | Behavior alone is insufficient evidence. | |
| Initiative | Proposes actions and topics without being asked. | Chat generally replies. An agent can act on a trigger. | Doing a task autonomously differs from choosing the topic or goal. | |
| Proactivity | Anticipates events and suggests action. | Monitoring, schedules, and alerts initiate messages. | Judge timing and relevance as well as whether a message was unsolicited. | |
| Goal Generation | Forms goals from interests, needs, and circumstances. | Generates subgoals and exploratory tasks within constraints. | Test origin, persistence, and revision independently. | |
| Social Pressure | Expectations, reputation, and status shape choices. | Recognizes pressure in text. Training and rules affect susceptibility. | Similar language need not imply similar motives. | |
| Provocation | A dare may provoke action by appealing to pride. | May spot a dare and hold its answer, or yield to a manipulative prompt. | Resistance and susceptibility are both inconclusive about emotion. | |
| Pride | Self-esteem and dignity may motivate a display of competence. | Can model reputation goals. Felt pride is unconfirmed. | Protecting reputation differs from feeling wounded pride. |
7. Creativity and imagination
| Term | Human brain | Artificial intelligence | Comparison (AI commentary) | My notes |
|---|---|---|---|---|
| Creativity | Creates ideas from associations, experience, and goals. | Generates and selects new combinations and solutions. | Novelty and usefulness may match even when motivations differ. | |
| Imagination | Mental images, scenarios, and alternatives. | Generates hypothetical texts, images, and worlds. | Both model possibilities. Subjective imagery in AI is unconfirmed. | |
| Art | Expresses and interprets meaning, culture, and experience. | Produces artistic work for an audience. | A work’s value does not require identical inner processes in its creators. | |
| Inspiration | A felt surge of creative energy. | Generates ideas on request or an algorithmic impulse. | Comparable results need not imply comparable feelings. | |
| Humor | Social incongruity, emotion, and play with meaning. | Recognizes and generates jokes, irony, and puns. | A funny result does not demonstrate amusement. | |
| Dreaming | Experiences during sleep, possibly linked to memory. | Scenario generation and autonomous memory summarization. | AI dreaming is a metaphor. | |
| Unconscious Processing | Processing outside conscious access. | Internal computation absent from the reply. | Both have hidden processing. The psychological term is metaphorical in AI. | |
| Topic Initiation | Starts conversations or changes their direction. | Can suggest topics, though ordinary interfaces favor replies. | Inventing a topic differs from choosing to raise it now. | |
| Spontaneity | Sometimes acts unexpectedly without being asked. | Can surprise, though a task usually triggers the action. | Random generation differs from meaningful initiative. |
8. Errors, limits, and scale
| Term | Human brain | Artificial intelligence | Comparison (AI commentary) | My notes |
|---|---|---|---|---|
| Error Correction | Revises beliefs and learns from consequences. | Verification, regeneration, and further training. | Fixing one answer need not produce lasting change. | |
| Confabulation | Unintentionally creates false memories. | Produces plausible false claims. | Often a closer analogy than clinical hallucination. | |
| Fatigue | Performance declines for biological reasons. | Context and resource limits, degradation in long loops. | Effects may resemble one another despite different causes. | |
| Sleep | A biological need with many functions. | Scheduled maintenance and background memory work. | AI has no biological need for sleep. Consolidation is a functional analogy. | |
| Energy Efficiency | The brain uses roughly 20 watts. | Consumption depends on computation and infrastructure. | Power figures require comparable tasks and workloads. | |
| Scalability | Constrained by the organism, time, and learning. | Model copies, parallel computation, and clusters. | AI scales more readily in hardware at a resource cost. | |
| Knowledge Sharing | Speech, teaching, books, and social interaction. | Copies parameters, knowledge bases, and procedures. | Digital transfer can be instant. People usually assimilate knowledge gradually. | |
| Reliability | Depends on experience, fatigue, and circumstances. | Depends on the model, context, tests, and integrations. | Assess reliability for a specific task. | |
| Self-Improvement | Chooses ways to develop skills. | Can test and improve solutions, sometimes changing code or settings. | Improved behavior differs from independently modifying weights. |