Artificial intelligence systems such as large language models often give users the impression that they are interacting with something genuinely intelligent. Yet many experts argue this impression is misleading, tells The Conversation. Philosopher Daniel Dennett famously described AI systems as demonstrating competence without comprehension, suggesting that machines can perform impressive tasks without truly understanding them. The debate, therefore, hinges on a deeper question: what exactly counts as intelligence?
Philosopher and AI researcher Blaise Agüera y Arcas offers a different perspective. He argues that intelligence may fundamentally be the ability to predict patterns in the world. From this standpoint, systems trained on large datasets that learn to anticipate likely outcomes exhibit a key property of intelligence. Modern AI models excel at identifying patterns in language, images, and other forms of data, which allows them to produce convincing responses and make useful predictions. If prediction lies at the core of cognition, then these systems may legitimately qualify as intelligent, even if their intelligence differs from that of humans.
The article also notes that human beings often treat human cognition as the benchmark for intelligence. However, nature provides many examples of intelligence that operate very differently from human thought. Animals such as birds, insects, and octopuses solve complex problems despite having vastly different brain structures. Artificial systems may represent another distinct form of intelligence rather than a flawed imitation of human reasoning.
Skepticism persists partly because intelligence is frequently associated with consciousness, emotion, or subjective experience. AI systems do not possess these qualities, which leads many critics to conclude that they cannot truly think. Yet others argue that intelligence does not necessarily require consciousness. Instead, it may emerge from any system capable of learning patterns, adapting to new information, and making predictions.
Ultimately, the discussion reveals that disagreements about AI intelligence reflect deeper philosophical disagreements about the nature of the mind. As AI systems continue to improve, they may force society to reconsider long-standing assumptions about what intelligence really means.