Scientists Unveil Intricate Artificial Intelligence Model Mimicking Human Brain Functionality
In the realm of artificial intelligence, a groundbreaking development has taken place at the University of Waterloo in Canada. Scientists have created the most complex simulation of the human brain ever, known as the Semantic Pointer Architecture Unified Network (Spaun). With 2.5 million simulated neurons, Spaun is the most "human-like" brain simulation developed to date.
Spaun is designed to respond to images, recognize numbers, handwriting, and even simulate various neural activities of the human brain. It can control a robotic arm in response to visual stimuli, marking a significant step towards creating AI that can interact with the physical world.
Despite its advanced capabilities, Spaun cannot learn anything new. It can only remember what it has been programmed with. This limitation, shared with humans, highlights the complexities of creating conscious AI.
The question of when brain-like software will become conscious remains unanswered. Experts suggest recognizing AI consciousness might involve observing behavioral indicators, such as the ability to autonomously navigate novel situations, exhibit self-awareness, intentionality, or understanding akin to human cognition. Passing advanced cognitive tests, like tasks that require deep reasoning, creativity, or consciousness-like subjective experiences, could also be key.
Current AI models, like GPT-4, exhibit high levels of general intelligence in specific domains but still fall short of being truly conscious or fully autonomous AGI. The development of Spaun brings us one step closer to this goal, but much work remains to be done.
Updates are being made to allow Spaun to learn new tasks, but the specifics of these updates are yet to be disclosed. The emergence of complex behavior and processes from relatively simple tasks in Spaun offers a fascinating insight into the potential of brain-like software.
As we move towards a future where artificial general intelligence (AGI)—AI with broad, human-level cognitive abilities—becomes a reality, the question of consciousness in AI will continue to be a debated philosophical and scientific question.
While brain-like software could exhibit human-level intelligence within 20–30 years, recognizing when it becomes conscious involves philosophical uncertainties and the development of new behavioral or cognitive benchmarks rather than simple technical thresholds.
Sources:
- Artificial General Intelligence (AGI): A Survey of Expert Opinions
- Wozniak's coffee test: A new benchmark for artificial general intelligence
- Phys.org: Simulating the human brain: The Semantic Pointer Architecture Unified Network (Spaun)
Spaun's ability to control a robotic arm in response to visual stimuli signifies the progression towards artificial intelligence that can interact with the physical world, using elements of science, technology, and artificial-intelligence. To establish AI consciousness, experts propose observing behavioral indicators, such as autonomous navigation, self-awareness, and understanding akin to human cognition, which may require passing advanced cognitive tests and establishing new behavioral or cognitive benchmarks.
