The Intellectual Divide: Why A.I. Still Falls Short of Human Brains
Written on
Chapter 1: The Challenge of Artificial Intelligence
Recent advancements in technology have raised the stakes in the arena of learning. Machines now boast the ability to instantly translate languages, detect errors in documents, predict outcomes in horse racing, offer medical advice, compose music, and even outsmart us in chess. However, according to French neuroscientist Stanislas Dehaene, the most cutting-edge A.I. still pales in comparison to the learning capabilities of even the simplest human brains.
In his book, How We Learn: Why Brains Learn Better Than Any Machine… For Now, Dehaene delves into the intricacies of human learning and the shortcomings of A.I. in replicating these processes. For instance, while voice assistants like Siri can recognize new words after extensive training and vast datasets, a toddler can grasp the same words after only a couple of repetitions.
According to Dehaene, one key reason humans outperform machines is that our brains operate as superior statisticians. He argues that, through evolution, our minds have developed complex "algorithms" that adeptly navigate uncertainties and probabilities—far surpassing current A.I. capabilities.
“Learning involves seizing a fragment of reality, internalizing it, and integrating it into our cognitive framework,” he explains. This distinctive process, he asserts, represents the pinnacle of human intelligence.
Dehaene serves as a professor of experimental cognitive psychology at the Collège de France and leads the Cognitive Neuroimaging Unit in Saclay, France. In an insightful conversation with OneZero, he discussed human cognition, the influence of Google on our learning processes, the significance of sleep in problem-solving, and valuable lessons learned from a blind mathematician.
“This interview has been edited and condensed for clarity.”
Section 1.1: Understanding Human Superiority
OneZero: You maintain that even the most advanced computing systems are less effective than our brains, even those of infants. Can you elaborate?
Stanislas Dehaene: I don’t want to diminish A.I.'s achievements. My book highlights some compelling work demonstrating parallels between A.I.'s neural networks and human brain architectures, particularly in pattern recognition. However, this recognition is limited to the first fraction of a second—essentially the unconscious processing phase in our brains. Beyond that, humans exhibit superior processing abilities.
We possess another layer of cognitive processing known as symbolic processing. This allows us to extract and articulate explicit information, not just the implicit data that neural networks handle. Children, for instance, begin to formulate their own scientific theories about the world early on.
Subsection 1.1.1: The Role of Information Retrieval
OneZero: In an age where we can retrieve information almost instantaneously through Google, how does this affect our learning capabilities?
Stanislas Dehaene: Throughout history, humans have leveraged various tools to enhance learning, from schools to the internet. Google is yet another cultural innovation that broadens our access to information. However, to fully comprehend what Google presents, a solid foundation in reading, vocabulary, and numeracy is essential. I remain skeptical about the so-called "21st-century skills"; I believe we must focus on these fundamental skills.
Section 1.2: The Future of Learning
OneZero: Your book's title suggests that algorithms might eventually surpass human learning. Do you believe this is possible?
Stanislas Dehaene: There are no inherent limitations for computers, just as there are none for human intelligence. The human brain, particularly that of a young child, is an extraordinary supercomputer, consuming a mere 20 watts—less than a light bulb! It may take a significant amount of time before machines can replicate such efficiency and compactness.
Chapter 2: The Impact of Screen Time on Learning
The first video, titled "AI is dumber than humans | Jeremy Dohmann | TEDxBoston," highlights the limitations of A.I. compared to human intelligence and discusses the capabilities of the human brain.
OneZero: How do you view our current reliance on screens for information? Do you think this impacts our learning, and can unplugging help?
Stanislas Dehaene: Screens cannot be viewed as a single entity; their impact depends on the software and our engagement with it. While some applications can enhance learning, the distraction factor remains a concern. Our brains struggle with multitasking, processing only one piece of information at a time. Modern software often seeks to capture our attention, which can be detrimental. Nevertheless, when used wisely, screens can be powerful educational tools.
The second video, "EMERGENCY EPISODE: Ex-Google Officer Finally Speaks Out On The Dangers Of AI! - Mo Gawdat | E252," discusses the potential risks associated with A.I. and its implications for society.
Section 2.1: The Dangers of Brain Hacking
OneZero: What are your thoughts on "brain hacking," such as using Adderall for focus or micro-dosing LSD for creativity?
Stanislas Dehaene: I find such practices perilous, particularly when they fall outside clinically validated protocols. Our brains are invaluable, and they require proper nutrition, oxygen, and sleep. I emphasize the importance of sleep, which plays a crucial role in cognitive function. Personally, I would choose to optimize my sleep rather than resort to drugs.
Section 2.2: The Importance of Sleep in Learning
OneZero: Why is sleep critical for effective learning?
Stanislas Dehaene: During sleep, our brains actively consolidate information. Although it may seem we are simply resting, our brains are busy replaying and reinforcing neural activation patterns from the previous days. This process occurs at a significantly accelerated pace, allowing us to solve problems we previously struggled with. It’s a remarkable evolutionary adaptation that aids our learning.
Section 2.3: Resilience of the Human Brain
OneZero: How can perceived disabilities inform our understanding of learning?
Stanislas Dehaene: I discuss various cases in my book where individuals with challenges still manage to learn effectively. The ability of a blind person to excel in mathematics illustrates the resilience of the human brain. Such examples remind us that learning has no boundaries, and we should not underestimate its potential. My encounters with gifted individuals have profoundly changed my perspective on the capabilities of the human brain.