Mind Matters Natural and Artificial Intelligence News and Analysis
the-first-model-of-the-computational-mechanism-is-an-arithmometer-stockpack-adobe-stock.jpg
The first model of the computational mechanism is an arithmometer.
The first model of the computational mechanism is an arithmometer.

Computers Are Getting Faster But Are They Getting Smarter? No.

Computers are Turing machines, limited to operations that can be completely understood in relation to their programming

Won’t quantum computers be smarter than regular ones? No. Still No.

What about optical computing, computing with DNA, or some other exotic form of computation? Always No.

A skeptical reader might ask, Why such a definitive answer? How do you deal with the spectacular performance of deep learning? What about AlphaGo Zero? What about Watson? What about the infamous Deep Blue? What about quantum supremacy? Don’t these examples all disprove your point?

No.

All forms of computation past, present, and future will be physical. And all physical phenomena can be modeled by a Turing machine (pictured). No matter how fast the computer runs, the computer will never be more powerful than a Turing machine. A Turing machine consists of five simple operations, a tape, and a set of symbols to write on the tape. All of these components and their interactions are very straightforward, and consequently the operation of any Turing machine can always be completely understood.

This fact crushes any claim that a new computational system will move the dial an iota towards making computers smarter. All computers will forever and always be Turing machines. and Turing machines are stupid. The possibility of smart machines is crushed by the “Turing
vise,” so to speak.

Why do I say Turing machines are stupid?

First I should define “smart.” To define “smart,” I refer to the Lovelace test. The Lovelace test considers a system smart if it can produce unexpected output even after every component and interaction of its system is fully known.

By this definition, neither a Turing machine, nor anything that a Turing machine can model, can be considered smart. All of the components and their interactions are very straightforward. Consequently, the operation of any Turing machine can always be completely understood. Accordingly, the Lovelace test shows that a Turing machine is not smart.

If all physical processes can be modeled by a Turing machine, and all computers past, present and future, will be made of some physical material, then the Lovelace test states no computer will ever be smart.

So, to return to our original question: are computers getting smarter, as opposed to bigger and faster?

As it is plain to see, the answer is No.


You may also wish to read these articles by Eric Holloway:

What’s hard for computers is easy for humans Some of the surprising things computers have a hard time doing and why.

and

Human ingenuity vs. the computer’s halting problem. In a dialogue with a friendly skeptic, I suggested an explanation he found astounding but it’s the only possible one.


Eric Holloway

Senior Fellow, Walter Bradley Center for Natural & Artificial Intelligence
Eric Holloway has a Ph.D. in Electrical & Computer Engineering from Baylor University. He is a current Captain in the United States Air Force where he served in the US and Afghanistan He is the co-editor of the book Naturalism and Its Alternatives in Scientific Methodologies. Dr. Holloway is an Associate Fellow of the Walter Bradley Center for Natural and Artificial Intelligence.

Computers Are Getting Faster But Are They Getting Smarter? No.