Intelligence
Intelligence is the capacity to acquire and apply knowledge, reason, learn, solve problems, and adapt to changing situations.
Sutskever's List
Intelligence is the capacity to achieve goals or perform cognitive tasks, considered here as a property that behavioral metrics may approximate without establishing its inner nature.
- How it is used
- Uses intelligence to frame measurable agent capability while questioning what performance-based claims justify.
- How deep it goes
- Defines goal-directed performance, examines environmental limits and philosophical objections, but does not establish an inner mechanism or essence.
- Examples
- Contrasts conversational imitation, video-game benchmarks, recoverable environments, and Stephen Hawking’s ability to communicate under different test conditions.
- What you can do after
- Evaluate/understand intelligence claims by stating the test conditions and separating demonstrated capability from claims about inner nature.
Inside AI
Intelligence is the cognitive capacity to make decisions, solve problems, and respond appropriately to situations.
- How it is used
- Distinguishes narrow AI performance from adaptable human ability when assessing claims about intelligent systems.
- How deep it goes
- Explains intelligence through human–machine comparisons and historical criteria, but does not provide a formal measure of general intelligence.
- Examples
- Uses walking, driving, chess, Go, and calculators to contrast learned human ability with specialized machine performance.
- What you can do after
- Evaluate/understand claims of machine intelligence by checking transfer to changed tasks and performance under uncertainty.
Deep Learning with R, Third Edition
Intelligence is the ability to use available information and past experience to produce successful behavior in uncertain and changing situations.
- How it is used
- Frames intelligence as adaptation to uncertain future situations rather than proficiency at a fixed task.
- How deep it goes
- Explains abstraction acquisition, recombination, and an efficiency-based evaluation proposal, but does not provide an implementation.
- Examples
- Uses jet-plane and supermarket analogies, plus ARC-AGI input–output tasks that require constructing an answer from a few examples.
- What you can do after
- Evaluate/understand intelligence by asking how efficiently limited experience supports appropriate behavior on novel, unanticipated tasks.
Read how Deep Learning with R, Third Edition explains intelligence →
Deep Learning with Python, Third Edition
The capacity to understand, learn, and adapt to tasks and environments, discussed here as the broad property AI aims to approximate.
- How it is used
- Frames intelligence as the target for moving AI beyond task-specific skill toward broad adaptation.
- How deep it goes
- Defines intelligence through analogy, generalization, and abstraction recombination, then points toward search and symbols without detailing implementation.
- Examples
- Uses analogies, the kaleidoscope, and ARC-AGI input-output grid tasks to illustrate recognition, recombination, and adaptation.
- What you can do after
- Evaluate intelligence by measuring performance on unanticipated tasks relative to fixed experience and prior knowledge.
Read how Deep Learning with Python, Third Edition explains intelligence →
How Large Language Models Work
Intelligence is the broad and disputed capacity to perform cognitive tasks such as reasoning, learning, and adapting to situations.
- How it is used
- Frames intelligence as a cautious lens for evaluating artificial intelligence and large language model capabilities.
- How deep it goes
- Defines intelligence through IQ limitations and task examples, then stops short of a precise definition or universal scale.
- Examples
- Uses IQ tests, book-smart versus street-smart abilities, cat recognition, and iterative S-curves as concrete illustrations.
- What you can do after
- Evaluate/understand LLM potential by separating capabilities from reliability instead of relying on the label intelligence.
Read how How Large Language Models Work explains intelligence →
The Art of AI Product Development
Intelligence is the capability of an AI system to infer, generate, decide, or perform tasks that require adaptive reasoning.
- How it is used
- Maps intelligence options so teams can select approaches for identified opportunities and communicate trade-offs with stakeholders.
- How deep it goes
- Defines symbolic and neural paradigms and distinguishes predictive, generative, and agentic systems without detailing implementation.
- Examples
- Contrasts GPT-4o with logistic regression for a simple analytical task and identifies predictive, generative, and agentic applications.
- What you can do after
- Evaluate and select an intelligence approach by matching the learning problem with capability, privacy, resource, and delivery trade-offs.
Read how The Art of AI Product Development explains intelligence →
Related concepts
- psychometricsmeasured by · Psychometrics is the field concerned with the theory and methods of measuring psychological attributes, abilities, and behavioral constructs.
- abstractionhas part · Abstraction is a simplified representation that exposes relevant properties or behavior while omitting lower-level details unnecessary for a given purpose.
- agencycontrasts with · Agency is the capacity of an entity to make choices and initiate or carry out actions with some degree of autonomy, often in pursuit of goals.
- augmented intelligenceused by · Augmented intelligence is the use of computational systems to extend human capabilities while supporting rather than replacing human expertise or judgment.
These guides are generated from the text of Manning books by Manning's knowledge graph and link to the sections they draw on. The full books are available on liveBook and at manning.com.