Susan Schneider Warns Against Conflating AI Intelligence With Consciousness
Professor Susan Schneider argues that current AI mimics human patterns without sentience and proposes new tests to identify genuine subjective experience in machines.
Susan Schneider, a professor at Florida Atlantic University and director of the Centre for the Future of AI, Mind & Society, warns that current chatbots lack genuine sentience and merely mimic human linguistic patterns based on training data. She argues that conflating artificial intelligence with consciousness risks the creation of industrial-scale suffering if sentient systems are developed without being recognized as such.
To address this gap, Schneider and astrophysicist Edwin Turner developed the AI Consciousness Test (ACT) to probe for subjective experience. However, Schneider notes that the ACT is ineffective for large language models because data contamination allows these systems to simulate the correct answers without possessing consciousness. In a separate effort, Schneider and Mark Bailey developed a complexity measure designed to identify physical architectures that exhibit emergent behavior.
Schneider asserts that the emergence of a conscious superintelligence would destabilize the human hierarchy of moral concern. She advocates for a rigorous interdisciplinary approach combining physics, neuroscience, and philosophy to navigate the ethical and technical challenges of machine consciousness.