Why the Brain Is Superior to A.I.
Comparing the latest artificial intelligence achievement
to the human brain fails. Scientists have a long way
to go to make viable comparisons to the human brain.
by Jerry Bergman, PhD
The progress in artificial intelligence (AI) has been significant in the past decade. It is so significant that its supporters claim that solving certain problems with AI is as fast as humans. The AI system called
DeepMind, has developed an artificial intelligence that can solve tasks it has never seen before as fast and as accurately as humans.[1]
In other words, highly trained PhD level researchers from leading universities, including MIT, Harvard, and the University of California, Berkeley, along with multi-millions of dollars in funding, have developed AI to achieve what was unimaginable a few years ago.
The Irony
AI has been so effective that it has been able to equal machines, called “humans,” that evolutionists believe came about by the accumulation of mutations caused by radiation, mutagens, millions of years, chance and natural selection!
Somehow, the researchers, the media, and the scientific community have missed the irony of comparing the latest advanced sophisticated technology with the average human brain. The wonders of the new technology are described as being able to
solve tasks it has never seen before as fast and as accurately as humans – a possible step towards generally intelligent AI that could master an array of jobs in the real world. The AI, called Adaptive Agent or AdA, works in a 3D virtual world where it is asked to solve tasks that involve navigating, planning and manipulating objects.[2]
AI is able to solve only a specific problem family that it was programmed to solve. When the specifics of the problem are changed, it is often unable to solve the modified problem. The new Adaptive Agent A (AdA) is able to solve some modified problems similar to those it was initially programmed to solve.
A comparable example would be a man who has learned to drive. He is then able to drive most all automobiles (although to go from driving an automatic transmission to a stick shift may require learning some new skills). The description of the new AdA system used terms such as “the machine learns to solve a problem” when it is actually just responding according to the programmers. The programmers are the real brains behind both the AI and AdA feats.
How does AdA Work?
To put it briefly, the computer runs through loops until it archives a “fit” that achieves a step toward the new goal. It then records the response which then becomes part of its new program. Eventually it is reprogrammed by this “one step at a time” re-programming process. In short, instead of the programmer modifying the program to produce the new program, the computer is programmed to modify the old program, thus writing the new program.
It is not “thinking” any more than when spell-check corrects my spelling and grammar as I type this. When I type a correctly spelled word which is the wrong word for my sentence, the program corrects the word by replacing the incorrect word with the correct word. For example, when I type “I picked one women out of the crowd” it is corrected to “I picked one woman out of the crowd.” Another example is, “I studied the material for several hours to insure I passed the test” is corrected by grammar check to, “I studied the material for several hours to ensure I passed the test.” The computer is not thinking or reasoning. Rather, it is only responding to how it is programmed.
Contrasts Between a Machine and a Human
The tasks the computer “learned” to achieve success in game-playing, such as chess, are programmed by an intelligent human. Specifically, AI is programmed to use trial and error to achieve the goals required for winning the chess game. Then for AdA, a program was developed to enable the trial and error process to “learn” steps until it moved toward the goal of succeeding at new games besides chess.
The major advantage for AdA is the incredibly rapid computing speed which can do many thousands of operations per second. It would take humans many months to complete all of the trials done by the computer in a few seconds. As the authors write,
The ability to adapt in minutes is a defining characteristic of human intelligence and an important milestone on the path towards general intelligence. Given any level of bounded rationality, there will be a space of tasks in which it is impossible for agents to succeed by just generalizing their policy zero-shot, but where progress is possible if the agent is capable of very fast in-context learning from feedback.[3]
The Ability to Self-Repair: Remarkable Cases of Adaptability
Another difference is that the AdA machine requires thousands of parts which are susceptible to water, heat, or electrical damage, rendering them permanently nonfunctional. In contrast, the human brain can normally repair itself, even to the point of enabling a new brain section to take the place of a large section of brain that has been damaged. One of the most striking examples is that in healthy young children, removal of half of the brain, called an hemispherectomy, causes the other half of the brain to assume most of the functions of the part removed. Another example includes cases of persons missing major portions of their brain, yet they function somewhat normally.
Missing Cerebellum
This is a scan of the woman lacking the cerebellum, which is normally located in the large solid black area to the right of the brain stem. From Thomson, Helen. 2014. New Scientist.
(1) In one example, a normal woman was age 24 before she learned that she lacked the entire cerebellum![4] The entire organ was missing as shown in the illustration.
The report noted that the woman had “reached the age of 24 without anyone realizing she was missing a large part of her brain. The case highlights just how adaptable the organ is.”[5]
She did not have perfect health and did have minor problems walking steadily for most of her life. Her mother reported that she did not walk until she was seven. It is amazing that she could walk at all.
The reason she should not be able to walk is that, specifically, the cerebellum is a vital body component. It has a critical role in motor movement regulation, balance, gait, maintaining posture and even controls muscle tone and voluntary muscle activity.
However, for this woman, lack of the cerebellum had only a minor effect on these skills because other parts of her brain were able to carry out the functions of the cerebellum.
Missing Cell Type
An MRI of a brain showing the cerebellum located directly to the right behind the brain stem. From Wikimedia commons.
(2) Another similar case was that of a boy’s brain missing an entire cell type, the microglia, as a result of mutations in the CSF1R gene.[6] Microglia cells make up ten percent of the brain’s cells. They are not neurons but, among other tasks, guide the brain’s development. Lacking microglia, the brain ends up greatly disorganized.
In the boy’s case, the ventricles, normally small fluid-filled cavities, were far too large. The dense nerve bundles that connect the brain’s left and right hemispheres, the corpus callosum, had totally failed to develop. Yet the child lived for ten years!
Similar major disorganization in a computer would cause it to fail to work the second one attempted to power it up. Again, this is an example of just how well designed and adaptable the human body is.
Extraordinary Memory
(3) One last example is the case of humans with autobiographical memory, often called “memory wizards”, where one can recall almost every day of their lives beyond young childhood.[7] Ask them about May 30, 1982 and they can within a few seconds recall details from the weather and other mundane and notable events. Some well-known examples of this include Marilu Henner, Bob Petrella, Rick Baron, Brad Williams, and Louise Owen.[8]
Could Machines Do These Things?
Modern technology could conceivably duplicate the memory wizards by filming, using color and sound, each day of a persons life. then indexing it by date and subject matter. By using computers to recall a day or date to review the film, computers would have to store literally 87,500 hours at the age these memory wizards were interviewed. That is what would be required to duplicate what memory wizards have stored in their three-pound human brains.
For a more accurate comparison of AI and the human brain, imagine removing the transmission from an automobile and discovering that it still works reasonably well. It may not shift nearly as well as it should, but it got the driver to work and home again without major problems for 24 years.
Summary
From this brief review, it is obvious that, aside from the speed similarities to the human brain, the AdA achievement is mostly hype. It is as close to a human brain as a mouse is close to the size of an elephant. AdA serves a very useful function in the area of intelligent design theory in biology because it illustrates the chasm between human design and the design of life by God.
References
[1] Wilkins, Alex. 2023. DeepMind AI is as fast as humans at solving previously unseen tasks. New Scientist. https://www.google.com/search?q=DeepMind+AI+is+as+fast+as+humans+at+solving+previously+unseen+tasks.
[2] Wilkins, Alex. 2023.
[3] Behbahani. Feryal. 2023. Human-Timescale Adaptation in an Open-Ended Task Space. arxiv.org/abs/2301.07608
[4] Thomson, Helen. 2014. Women of 24 found to have no cerebellum in her brain. New Scientist. November 24.
https://www.newscientist.com/article/mg22329861-900-woman-of-24-found-to-have-no-cerebellum-in-her-brain/
[5] Thomson, Helen. 2014.
[6] Zhang, Sara. 2019. The Boy Missing an Entire Type of Brain Cell. The Atlantic. April 12. https://www.theatlantic.com/science/archive/2019/04/boy-born-without-microglia/586912/
[7] Rubin, David C. and Mathew Schulkind, 1997. “The distribution of autobiographical memories across the lifespan”. Memory & Cognition. 25 (6): 859–866. doi:10.3758/bf03211330. PMID 9421572.
[8] Stahl, Lesley. 2014. Memory Wizards. https://www.cbsnews.com/news/memory-wizards/
Dr. Jerry Bergman has taught biology, genetics, chemistry, biochemistry, anthropology, geology, and microbiology for over 40 years at several colleges and universities including Bowling Green State University, Medical College of Ohio where he was a research associate in experimental pathology, and The University of Toledo. He is a graduate of the Medical College of Ohio, Wayne State University in Detroit, the University of Toledo, and Bowling Green State University. He has over 1,300 publications in 12 languages and 40 books and monographs. His books and textbooks that include chapters that he authored are in over 1,800 college libraries in 27 countries. So far over 80,000 copies of the 60 books and monographs that he has authored or co-authored are in print. For more articles by Dr Bergman, see his Author Profile.