Humans, Computers, and Learning.

The following passage is from Michael French's book "Invention and Evolution: Design in Nature and Engineering" published in 1994.

"Humans can still beat computers at chess, even though the computer can carry out 'mental' operations thousands of times faster than the human, because the man's thinking is of altogether a higher type. But the same human might be able to play the piano, and tennis, and chess, and, moreover, speak several languages (while no computer can speak one), drive a car and ride a bicycle, do real mathematics and a thousand other things which are beyond the computer: we are still at an early stage of learning. With learning, however, we touch the sort of area where the gap between our efforts and natural design is widest. We can make computers 'learn', but only in very limited ways which have been built into the software. We can build in the conditioned reflex, the response taught by association, like Pavlov's dogs salivating at the sound of the bell they had learned to associate with food. We can make adaptive control systems that 'observe' their own performance and improve it, rather like a human learning a skill of coordination, such as catching a ball. What we cannot yet do is to make a computer that can perceive a problem and teach itself how to cope with it. For all the talk of 'artificial intelligence', that objective still seems very remote."

The observations made by Michael French are still relevant today, almost 30 years later. While computers have made significant advances in their ability to process information and perform certain tasks, humans still possess unique cognitive abilities that allow them to perform a wide range of complex tasks. As French notes, the gap between our efforts and natural design is widest in the area of learning. While computers can be programmed to learn in limited ways, they still cannot perceive a problem and teach themselves how to solve it in the same way that humans can. This remains a significant challenge in the field of artificial intelligence, despite advances in machine learning and other related fields. French's observations underscore the ongoing need for continued research and development in the field of AI, as we seek to bridge the gap between human and machine intelligence.

Stuart Russell, a Professor of Electrical Engineering and Computer Science at the University of California, Berkeley. In his 2019 book "Human Compatible: Artificial Intelligence and the Problem of Control," Russell writes that "current AI systems are nowhere near as flexible or general-purpose as human intelligence". He notes that while machines are highly efficient at performing specific tasks, they lack the ability to generalize and learn new concepts in the same way that humans can.

This sentiment is echoed by Pedro Domingos, a Professor of Computer Science and Engineering at the University of Washington. In his 2015 book "The Master Algorithm: How the Quest for the Ultimate Learning Machine Will Remake Our World," Domingos writes that "today's machine learning algorithms are narrow and brittle, incapable of learning outside the domains they were designed for".

While significant progress has been made in machine learning and related fields, the ability to create a machine that can perceive a problem and teach itself how to solve it remains a significant challenge. As Domingos notes, "the ultimate goal is to create an algorithm that can learn anything a human can learn, and perhaps even more".

As someone born in late 1980s, I recognize that Michael French's observation about the limitations of machines and the unique cognitive abilities of humans was made at a time when computers and technology were rapidly advancing, but had not yet reached the levels of sophistication and ubiquity that we see today. Since then, there have been significant advancements in artificial intelligence and machine learning, and computers can perform specific tasks with incredible efficiency.

However, I agree that there is still a fundamental difference between human intelligence and machine intelligence. Humans possess a unique set of cognitive abilities, including creativity, flexibility, and general problem-solving ability that cannot be fully replicated by machines. While machines can perform tasks quickly and efficiently, they lack the human capacity to learn, adapt, and innovate.

As someone who grew up during a time when the internet and digital technologies were becoming increasingly integrated into everyday life, I have seen firsthand the rise of artificial intelligence and machine learning in various areas. I understand that these technologies have the potential to improve our lives in many ways, but also recognize the ethical and societal implications that come with their development.

References:

French, M. (1994). Invention and evolution: Design in nature and engineering. Cambridge University Press.

Russell, S. (2019). Human Compatible: Artificial Intelligence and the Problem of Control. Penguin.

Domingos, P. (2015). The Master Algorithm: How the Quest for the Ultimate Learning Machine Will Remake Our World. Basic Books.