Before tackling whether we can build conscious robots, we must grapple with defining consciousness itself—a slippery concept that even experts struggle to pin down today. Perspectives vary widely across scientific fields, shaping debates on whether machines could ever possess it amid accelerating AI advances.
This raises another key challenge: If we create a conscious machine, how do we confirm it's genuine rather than a sophisticated simulation? How can we tell real sentience from a clever imitation?
Some neuroscientists view subjective consciousness as a perceptual 'illusion' born from intricate brain processes. Replicate those in an AI, they argue, and consciousness follows—but verification remains elusive.
Daniel Dennett from Tufts University suggests a rigorous Turing test, where a machine convinces interrogators of its awareness through sharp, persistent dialogue. Michael Graziano at Princeton proposes a bolder path with his 'attention schema' theory: consciousness as the brain's simplified model of its own attention mechanisms.
Graziano believes we could engineer machines with comparable self-models. 'If built transparently, we could peer inside and verify its rich self-representation,' he notes. 'It would think and believe it has consciousness—elements we could confirm by tracing its information processing.' For him, consciousness could emerge in software, biology, or hybrids.
Anil Seth at the University of Sussex remains skeptical. 'It's unclear if consciousness transcends its biological substrate,' he says. Detection might hinge on brain-like structures and materials, such as those in organoids.
Integrated Information Theory (IIT), pioneered by Giulio Tononi, offers a metric: 'phi,' measuring information integration. A phi above zero signals consciousness—but calculating it for complex systems like AI is practically impossible today.
Phil Maguire at Maynooth University, Ireland, argues machines are inherently disintegrated: 'We analyze their parts independently without needing consciousness as an explanation.' True integration defies part-by-part dissection, dooming machines to unconsciousness.
Selmer Bringsjord at Rensselaer Polytechnic Institute echoes this, positing consciousness stems from a non-material essence essential for human-like intelligence. Machines, lacking it, can never match us.
