Part I. Artificial Intelligence and the Nature of Intelligence
What is intelligence, and how does artificial intelligence compare with human intelligence?
When Machines Begin to Challenge Our Understanding of Intelligence
For much of human history, the ability to reason, learn, solve problems and communicate complex ideas has been regarded as a defining characteristic of human intelligence.
Artificial intelligence is challenging that assumption.
Today, AI systems can solve mathematical problems, write computer programmes, analyse scientific information and perform increasingly complex reasoning tasks. Some achieve results that equal or exceed human performance in particular domains.
Yet these achievements raise a question that is surprisingly difficult to answer:
When a machine demonstrates intelligent behaviour, does that mean it possesses intelligence in the same sense that we do?
Consider a simple example.
A calculator can multiply enormous numbers faster and more accurately than most human beings. We recognise its computational superiority, but rarely describe it as intelligent in the broader human sense.
Now consider an AI system capable of discussing philosophy, analysing scientific evidence or solving an unfamiliar problem.
At what point does increasingly sophisticated information processing become something we should recognise as genuine intelligence?
And is there a meaningful distinction between a machine that produces an intelligent answer and one that understands what that answer means?
These questions form the foundation of Part I.
Intelligence Is More Difficult to Define Than It Appears
Human intelligence encompasses numerous abilities, including learning, reasoning, memory, adaptation, creativity and problem-solving.
But intelligence is not a single, universally agreed-upon property.
A scientist developing a new theory, a musician improvising a composition and a child learning to speak all demonstrate different aspects of intelligence.
Artificial intelligence complicates the picture further.
Modern AI systems can acquire useful patterns from large quantities of data and apply those patterns to tasks they were not explicitly programmed to perform.
This is a significant technological achievement. However, the relationship between successful performance and genuine understanding remains a subject of scientific and philosophical debate.
In 1950, the mathematician Alan Turing approached the question of whether machines could think by proposing a test based on their conversational behaviour.
His work helped establish a fundamental question that remains relevant today:
Is intelligent behaviour sufficient evidence of intelligence, or does genuine thinking require something more?
Three Investigations into Artificial Intelligence
01. What Is Intelligence, and Can Machines Truly Think?
If an AI system can reason, learn and solve problems, what distinguishes its intelligence from our own?
This investigation examines the nature of human and artificial intelligence, including machine learning, reasoning and the distinction between producing correct answers and understanding their meaning.
It considers whether intelligence should be judged primarily by observable capabilities or whether deeper properties are necessary.
At its centre lies a deceptively simple question: What does it actually mean to think?
02. What Would Count as Artificial General Intelligence?
Could artificial intelligence develop the broad intellectual capabilities associated with human intelligence?
Artificial general intelligence, commonly known as AGI, is generally understood as AI capable of performing competently across a wide range of intellectual tasks rather than being limited to a narrow speciality.
But there is no universally accepted threshold at which an AI system becomes AGI.
This investigation examines competing definitions, the capabilities of contemporary AI and the challenges involved in measuring general intelligence.
It also considers whether impressive performance across many tasks necessarily demonstrates the flexibility, adaptability and understanding associated with human intelligence.
03. Could Artificial Intelligence Surpass Human Intelligence?
What happens if artificial intelligence becomes more capable than human beings across most intellectual domains?
AI already exceeds human performance in certain specialised tasks. The more consequential question is whether future systems could achieve a much broader intellectual superiority.
This investigation examines artificial superintelligence, the possibility of continued capability improvements and the technological, computational and practical limitations that may constrain progress.
It also considers what increasingly powerful AI could mean for scientific discovery, human decision-making and our ability to understand and govern the systems we create.
What Does Intelligence Tell Us About Consciousness?
There is another question running beneath these investigations.
Does intelligence require consciousness?
Human beings experience the world subjectively. We do not merely process information; we experience thoughts, perceptions and emotions.
Artificial intelligence can process information and produce sophisticated responses. But whether an artificial system could possess subjective experience is a separate and unresolved question.
An AI might demonstrate extraordinary intellectual capabilities without being conscious. Equally, consciousness may not require the advanced reasoning abilities associated with human intelligence.
Understanding this distinction is essential before asking whether machines could become self-aware or sentient.
It also provides the connection to Part II of this research programme, where we investigate artificial consciousness, self-awareness and the possibility of machine sentience.
Research Perspective
The development of artificial intelligence challenges more than our assumptions about technology. It challenges our understanding of intelligence itself.
We should neither dismiss artificial intelligence as mere imitation nor assume that impressive performance demonstrates human-like understanding.
The evidence establishes that machines can perform an expanding range of intellectually demanding tasks. What remains disputed is how these capabilities should be interpreted, how far they might develop and whether they require the same underlying processes as human intelligence.
The central question is no longer simply whether machines can perform intelligent tasks. It is what those achievements reveal about the nature of intelligence, and what they do not.
That is where this investigation begins.
Selected References and Further Reading
- Turing, A. M. (1950). ‘Computing Machinery and Intelligence’. Mind, 59(236), 433–460. https://doi.org/10.1093/mind/LIX.236.433
- Morris, M. R. et al. (2024). ‘Levels of AGI for Operationalizing Progress on the Path to AGI’. Proceedings of the 41st International Conference on Machine Learning. https://deepmind.google/research/publications/66938/
- Stanford Institute for Human-Centered Artificial Intelligence (2026). The 2026 AI Index Report. https://hai.stanford.edu/ai-index/2026-ai-index-report
This article forms Part I of The Nature of Intelligence, Consciousness and the Future of Humanity, an independent research programme by Michael Cloke. The investigations distinguish established evidence from contested interpretations and questions that remain unresolved.
