liveBook · Manning

Human intelligence

As explained in Machines that Think

The human capacity for language, reasoning, adaptation, creativity, and related cognitive abilities.

Human intelligence can be considered through observable abilities such as language, reasoning, adaptability, and creativity, or through the cognitive mechanisms that produce behavior. Similar performance alone does not establish similar mechanisms. section 1.3.4 section 5.4.6

The Turing test makes behavior a practical basis for judging intelligence: a machine imitates a human in written dialogue, and a judge tries to tell them apart. This evaluates interaction, not what happens inside the mind; resemblance therefore does not settle whether imitation amounts to understanding. section 1.3.4 section 5.4.6

General intelligence need not imply:

  • Perfect performance, even in areas of competence. section 9.5.1
  • The ability to perform every cognitive task. section 9.5.1
  • A human-like architecture or superiority to humans across almost every domain. section 9.5.1

Judgments of broad capability are stronger when they accumulate evidence across tasks—from conversation and school-level reasoning to programming, creative work, planning, and research assistance—rather than treating one test as decisive. Comparisons with artificial neural networks should also distinguish outcomes from the mechanisms behind them, including differences in goals, learning, memory, and representation. section 9.5.1 section 5.4.6

Figure 1.5 The Turing Test. A human judge engages in written dialogue with two unseen participants. One is human and the other a machine, but their identities are hidden from the judge. If the judge cannot reliably distinguish the machine from the human based on their responses, the machine is said to pass the test.

The diagram shows the written-dialogue setup used to judge whether a machine can be distinguished from a human. section 1.3.4

Table 5.1 Key structural differences between human cognition and artificial neural networks. Similar behaviors can emerge in both systems, but they arise from fundamentally different mechanisms governing goals, learning, memory, time, and representation.
Dimension
Human cognition
Artificial neural networks
Goal formation
Arises from internal processes shaped by needs, intentions, and external context
Externally specified as optimization targets during training
Learning signal
Consequences of actions and interactions with the environment
Numerical loss minimized during training
Memory
Continuously reshaped through use and recall
Parameters remain fixed between training runs
Temporal organization
Learning and action unfold together over time
Training and inference generally occur in separate phases
Representation
Concepts refer to objects and events in the world
Representations capture statistical structure in data
Adaptation
Development accumulates along an irreversible history
Models can be retrained, reset, or copied

The comparison outlines ways similar behavior can arise from different cognitive and computational structures. section 5.4.6

How Machines that Think covers it

How deep it goes
Contrasts behavioral evidence with underlying mechanisms but does not detail human cognitive processes.
Examples
Uses written-dialogue imitation and a cascade of tasks from conversation to research assistance.
What you can do after
Evaluate claims of human-level intelligence by weighing varied performance and separating behavioral resemblance from mechanism.
How it connects
Connects human intelligence to the Turing test, general intelligence, and artificial neural networks.

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Machines that Think. The section links point to the parts of the book it draws on.