Programmer: How We Know Computers Won’t Surpass the Human Mind – Walter Bradley Center for Natural and Artificial Intelligence

Heres a brief excerpt from Chapter 22 of Minding the Brain (Discovery Institute Press, 2023), The Human Minds Sophisticated Algorithm and Its Implications, by programmer Winston Ewert. His discussion is based on the halting problem: No computer knows when another computer will halt, though humans do.

Is the human mind a computer? If not, what is it? Before we can answer, we must first clarify, what exactly is a computer? Historically, the term computer actually referred not to machines but to humans. Typically, these were teams of people working together to perform long and tedious calculations. They helped with such tasks as computing the positions of planets, producing mathematical tables, and simulating fluid dynamics. What made them computers was that they were following a procedure. They were not expected or allowed to engage in creative thinking or problem-solving; instead, every action they took was guided by the procedure given to them. All that our modern computers do is automate this procedure- following activity. Human computers and machine computers are similar in that they operate strictly by following a procedure.

What exactly constitutes a procedure? A procedure provides a step-by-step method for solving a particular class of problems. The procedure defines how to proceed at every step of the task, leaving no decision up to the judgment of the person or machine following the procedure. In the context of computers, these procedures are typically called algorithms.

However, not every task can be reduced to a procedure. Researchers working in theoretical computer science have proven that a number of tasks cannot be reduced to a procedure. There is no procedure that can be written that will reliably perform these tasks. For example, there is no procedure that determines whether a logical statement, in first-order or higher logic, follows from a given set of premises.

Is everything that the human mind can do reducible to a procedure or program, even if we are not consciously aware of the procedure? Could we, in principle, duplicate the abilities of the human using a computer program? Or are there at least some tasks that the human mind can accomplish which cannot be reduced to a procedure? Are there things that the human mind can do which could not be duplicated by any procedure or program?

We have sketched an argument that generating cognitive abilities requires greater cognitive ability. This has a number of interesting consequences: First, human cognitive ability will never be matched by artificial intelligence. We have argued that the only way to obtain an accurate partial halting detector is using a more powerful halting detector. When humans devise artificially intelligent systems, they use their internal powerful halting detection abilities to verify and/or construct the implicit halting detection present in the artificially intelligent system. However, they are only capable of devising a halting detector less powerful than the one they have. As such, we would expect that while humans will get better at building artificial intelligence systems, they will never be able to match themselves.

Second, the singularity will not happen. The idea of the singularity is that an artificially intelligent system will be able to build a slightly more intelligent artificial intelligence (AI) system. That system will, in turn, devise an even more intelligent system. This process, repeated over and over, will culminate in artificially intelligent systems which will leave humans far behind. However, the only way to obtain a partial halting detector is using a more powerful partial halting detector. An AI system cannot build a slightly more intelligent partial halting detector. Thus, the singularity will not occur.

Third, the human mind has a transcendent origin. Standard evolutionary theory claims that the human mind was produced by natural selection operating on random mutations. However, this would be a case of a very computationally simple process constructing an accurate, highly powerful halting detector. This cannot happen if the only way to obtain a partial halting detector is by using a more powerful halting detector. Instead, the human mind must have derived from something with more powerful halting detection abilities. Yet we cannot explain the human mind by an infinite regress of increasingly powerful partial halting detectors. Rather, the human mind must eventually be explained by a non-computational form of intelligence for whom the halting problem is no obstacle.

You may also wish to read: Programmers: Why materialism cant explain human creativity Eric Holloway and Robert Marks explain why its unlikely that the mind that enables human creativity is merely the product of animal evolution. The total space-time information capacity of the universe falls significantly short of the ability to generate meaningful text of only a few hundred letters.

Read the rest here:

Programmer: How We Know Computers Won't Surpass the Human Mind - Walter Bradley Center for Natural and Artificial Intelligence

Related Posts

Comments are closed.