The Age of AI

From Open Source Ecology
Jump to: navigation, search

See reviews of the book:

Artificial intelligence is a powerful technology. It must be treated with respect and awareness.

Appears to be another work by The Best and the Brightest, if one reads a critique from an apparently informed reader:

Back to the limitations of AI:

The flaw in this book’s assumptions is very, very fundamental: The meaning of language cannot be reduced to equations. Data can be reduced to 1s and 0s but meaning cannot. This is fact, not theory. Claude Shannon the father of information theory made this very clear in 1948.

Alan Turing, the man who cracked the German code in WWII for the British, invented the modern computer in 1936. Computers by Turing’s own definition cannot make the kind of inferences necessary to understand language without constant supervision from humans.

I’m founder of the world’s largest science research prize, the $10 million Evolution 2.0 Prize. It is a search for the origin of the genetic code. We get multiple submissions every month and no one has come close to solving it. No one knows how to get code without a biological entity writing that code and it’s one of the biggest unsolved problems in the history of science. The problem may be solvable, and if it is, it will lead to a very different kind of computer than the ones we have now.

While language is implicit (inductive), computers require everything to be explicit (deductive). Computers do not do inductive reasoning; they only compute. By definition. They cannot draw inferences that they have not been pre-programmed to draw. Several papers published during the last year mathematically prove this is true. One is my own “Biology Transcends the Limits of Computation” and the other is “Relational Biology and Church's Thesis” by AH Louie. And "The World is not a Theorem" by the rather famous scientist Stuart Kauffman. This is inevitable based on the very definition of a computer.

We talk to Alexa and Siri and they obey simple commands. But they do not understand you. No one is home. This is the problem Facebook is up against with hate speech. Such a challenge is insurmountable within the current definition of a computer. No AI, machine learning or software, no matter how sophisticated, will ever solve this until the computer itself is reinvented. This is why algorithms are so abysmal at making human-level judgments.

Meanwhile your dog and your goldfish are already doing something that is mathematically impossible for any computer to do. They make inferences.

Computers do not think, they cannot make inferences without help - but this book doesn’t deal with this at all. Every AI and machine learning algorithm is merely the extension of a human being and it’s only as good - and conscientious and moral - as the humans that run them.

This book could have been an honest discussion about the assumptions that drive silicon valley. But this book propagates those assumptions as the basis of a utopia that I assure you will never arrive.