Question for the AI age: do machines and humans learn the same way?

• AI has evolved from separating data into different groups to excelling at numerous tasks.

• In 2024, smartphones began to feature AI models, and five of the seven scientists who won the 2024 science Nobel Prizes were for AI work.

• The age of AI also promises to provide insights into the human brain, as existing AI models are inspired by animal brains.

• Neuroscientists argue that AI struggles with tasks that humans find easy, like understanding motives.

• The paradox of AI stems from the human brain’s evolutionary, biological origin, while AI does not.

• Learning in humans and animals is driven by a dual focus on meeting immediate biological needs and adapting to a constantly changing environment.

• Learning is an ongoing, dynamic interaction between an organism and its environment, requiring adaptation to both predictable stimuli and the unpredictability of environmental fluctuations.

Humans’ Advantages Over Machines in Learning

Human Curiosity and Data

• Biologists at the Heidelberg Laureate Forum argue that humans are naturally curious, exploring the world and learning within a social and cultural context.

• Children’s experiences are sensory, multi-dimensional, and richer than data fed into AI systems.

• Human learning requires less data to solve a problem with the same level of proficiency, as humans have been trained on such data over millennia.

Speed and Efficiency of Learning

• Humans can absorb new information rapidly, building on past experiences and knowledge in a flexible, adaptive way.

• This ability to continuously improve on prior lessons without extensive reprogramming gives humans a significant edge in dynamic environments where new information and challenges emerge constantly.

Transfer Learning

• Humans are remarkably good at “transfer learning,” applying knowledge and skills from one context to entirely different, unfamiliar scenarios with relative ease.

• This ability to generalize is still a significant challenge for machines and artificial networks, which are typically confined to narrow domains and struggle to adapt to new or unforeseen contexts without retraining.

Motor-Skill Learning

• Humans excel at motor-skill learning, which is not as efficient as neural networks.

• Machines can operate very fast but their energy consumption is much higher, especially when they process large datasets.

Reliability of Machines

• Machines are more reliable than humans, but they are built for repeatability and can perform the same task again with consistent precision.

• In contrast to the brain, neural network models are often trained to search exhaustively for solutions to complex tasks, making them more likely to discover new, better solutions to problems than humans can.

Artificial to Human Learning: Differences and Challenges

• The neural network of each brain, artificial or biological, differs from human learning.

• The reductionist approach to neurons treats them as point processes communicating via electrical impulses.

• Feedback loops drive learning, a principle used in reinforcement learning.

• Artificial neural networks have expanded understanding of memory storage and access in the brain.

• AI models have prompted a revisit of how the human brain learns.

• Symbolic operations were assumed to represent information about the world, achieving perception, planning, reasoning, etc.

• Early AI models used expert systems for complex reasoning.

• Contemporary neural networks use connectionist models, starting with a blank slate and using pattern recognition techniques.

• The question remains: which type of AI is better for human learning?