Eps 37: The computer is smarter than the brain but the brain is still like a computer
| Host image: | StyleGAN neural net |
|---|---|
| Content creation: | GPT-3.5, |
Host
Dianne Douglas
Podcast Content
The brain has greater flexibility, generalizability, and learning abilities than the current-generation computers. Computers may emulate multiple aspects of human brains over the coming decades, but a computer that can possess all of human brains abilities, and that is similarly effective at using resources, is still within science fiction territory.
Because of these advantages, computers will be able to generate far deeper heuristics and statistical information about decisions than human brains. A computer can process information exponentially faster than a human brain.
An important distinction between computers and the brain is the way in which information is processed in each system. The computer and brain also share similarities and differences in signal modes from their basic units. Both the brain and computer contain large numbers of elementary units--neurons and transistors, respectively--that are connected together into complex circuits for processing the information communicated through electrical signals.
Of course, the brain is as much of a physical object as computers; the brain works partly by using electrical signals, solving difficult problems, and processing information one way or the other. Beyond that perspective, one assumes the brain works like a computer, and that its processing can be worked out in an analogy to computing processes.
This type of thought was taken to its final expression in a brief book, The Computer and the Brain , where the mathematician John von Neumann stated explicitly that the functions of the human nervous system are, prima facie, digital. It is easy to see how human minds and computers are frequently talked about in an analogous way. The comparison of computers to brains has been illuminating for computer engineers as well as neuroscientists.
As neuroscientists unlock even more secrets about the brain--aided, more and more, by using computers--engineers may draw even greater inspiration from how the brain works in order to continue improving computer architectures and performance. Just as importantly, however, such studies will make the strong case that rather than worrying about being replaced by artificial intelligence, humans should instead focus on increasing the intrinsic capacity of the brain to do things digital computers cannot, like creating new things and ideas. In fact, our robot future may prove to be more beneficial than detrimental, as computers will help us learn more about the workings of the brain, while brain-machine technologies such as deep-brain stimulation offer the promise of new treatments for diseases and disorders. We are living in a world in which computers are beating humans at chess, Go, and even Jeopardy.
One of the big mysteries about artificial intelligence is that - while they have gotten exponentially more powerful= in the last 50 years - computers have become just slightly better, and completely failed, at performing many tasks performed by human brains instantaneously. Humans are still vastly superior to computers in a number of ways. Yet, there are scientists who are certain that AI will not just catch up with humans, but will quickly outperform us once computers achieve our level of intelligence. Experts agree that humans still surpass computers for overall intelligence, creativity, and common-sense knowledge, or understanding, of the world.
Computers may be programmed to duplicate some of these tasks, but they will not have an inherent capacity for creation in the same way as humans. Computers can be programmed with huge libraries of information, but they cannot experience life in the same way humans can. The problem is, the brains are not only doing computations, but also being able to provide interpretations and meanings for its own higher-level processing of information. To realize a computer at the highest level, i.e., one similar to the brain, we need to pay attention to conscious human capabilities, and the way that these affect system-wide information processing.
This is the reason; a position can argue that one does not need to posit computers-like-brains, or conscious machines, that are superior to human capabilities. In other words, human brains are limited by biology, while computing systems in the future are not, in theory, bound by those limits. Some cognitive scientists--notably Anthony Chemero at the University of Cincinnati, author of the book "Radical Embodied Cognitive Science" --now completely reject the idea that the human brain works like a computer. For over a half-century, psychologists, linguists, neuroscientists, and other experts in human behavior have maintained that the human brain works like a computer.
In this post, I want to address the misperception that ones brain is like a computer, explaining why the analogy of artificial intelligence=human brain is, at best, misleading, and, at worst, dangerous. The human brain, of course, remains far more complicated than either of these models; it is bigger, denser, more efficient, more interconnected, has more sophisticated neurons - and juggles multiple algorithms simultaneously.
One notable difference between a human brain and computer flash storage is the capacity for neurons to join together with each other to help create and store memories. The human brain has approximately 100 billion neurons, and approximately one quadrillion--one million billion--connections known as synapses connect these cells together. Yes, humans learn from mistakes too, but we are much more fallible when it comes to solving the kinds of puzzles computers are good at solving.
Computers have to be programmed, or told, on how to learn, and computers then programmed to only operate in these situations. Moreover, successes in those areas will not result in superiority for computers. For instance, although self-driving cars will learn to drive, they will still require detailed 3-D maps in order to operate effectively, in contrast with humans. Abstractions and broad concepts are indeed helpful from theoretical point of view; concepts such as calculation, information, and brain processing, however, have no obvious explanation.