Home / Theories of Consciousness
Integrated Information Theory and AI: What Phi Means

Integrated Information Theory, or IIT, is the leading theory of consciousness that points away from conscious AI. It says consciousness is not about what a system computes but about how its physical parts cause and constrain each other, a quantity it calls phi. On that view an ordinary computer could reproduce everything a brain does and still have little or no experience, because its hardware is not integrated in the right way.
What integrated information theory says
The neuroscientist Giulio Tononi proposed IIT in 2004. In the Internet Encyclopedia of Philosophy's account, the theory holds that consciousness is a certain kind of information that requires physical integration and can be measured by phi.
IIT works in an unusual direction. Most theories start from the brain and ask what it does. IIT starts from experience itself and lists five features it takes to be self-evident: experience exists, it is structured, it is specific, it is unified, and it has definite borders. For each of these "axioms" the theory states a "postulate", a property the physical system must have to support it. The integration postulate, for example, says the system cannot be split into independent parts without changing what it does to itself. The highest value of phi found within a system measures how conscious it is.
The current version, IIT 4.0, was set out by Larissa Albantakis, Tononi and colleagues. It restates the axioms and postulates in mathematical form, with the aim that they could in principle be applied to any system to say whether it is conscious, to what degree and in what way.

What integrated information theory says about AI
Because phi is a property of physical cause and effect, IIT judges the hardware, not the software. According to the encyclopedia's summary:
- Feed-forward systems are not conscious. A system in which signals only pass one way would have no integrated information, and a feed-forward system that perfectly mimicked a conscious being would only simulate consciousness.
- Conventional processors score low. Ordinary CPUs have few connections per circuit, so their phi is limited; high phi would need densely interconnected, brain-like hardware.
In 2024 Graham Findlay, Christof Koch, Tononi and colleagues took the argument further. They compared simple systems with a basic stored-program computer that simulated each one with full functional equivalence, and concluded that under IIT the computer need not share the simulated system's experience. In their words, functional equivalence does not imply phenomenal equivalence. That is the opposite of the view behind global workspace theory and AI, where the right kind of information processing would be enough.
A closer look: why many researchers doubt IIT
The theory is influential and contested in equal measure. The encyclopedia records several lines of criticism:
- Strange verdicts. The computer scientist Scott Aaronson argued that simple mathematical structures such as expander graphs could have higher phi than a human brain. Tononi accepted that conclusion; Aaronson replied that the theory leans on intuition where it suits it and sets intuition aside where it does not.
- Information depends on an observer. The philosopher John Searle objected that information is relative to whoever interprets it, so it cannot be the intrinsic basis of experience.
- It cannot be calculated. Exact phi is intractable for complex systems, so the theory cannot yet be checked directly on a brain or a large AI system.
If IIT is right, the question in can AI be conscious has a hardware answer, not a software one. If it is wrong, its verdict on computers tells us little.
Checklist: reading an IIT claim about an AI system
- Is it about the hardware? IIT's verdict concerns the physical chips running a model, not the model's behaviour or architecture diagram.
- Is phi being calculated or estimated? Exact values are out of reach for large systems; most claims are arguments, not measurements.
- Is the theory being assumed? A low-phi verdict only rules out consciousness if IIT is true.
- Which version? IIT 4.0 changed details of the earlier formulations.
Related reads
- Global workspace theory and AI: the functional rival.
- The hard problem of consciousness and AI: the puzzle IIT is trying to solve.
- How to test if AI is conscious: what can be checked in practice.
More in Theories of Consciousness · Report a correction