You may not believe that artificial intelligence, although it seems to be getting more and more powerful, is actually reaching a dead end. American computer scientist Eric Larsen believes in his book "Making Gods" that almost all mainstream artificial intelligence has fallen into the "Russell's turkey paradox".
This paradox is a tragic story about a turkey:
There was a turkey who was good at inductive reasoning and was brought to the turkey farm. It found that on the first morning, the feeding time was 9 a.m. As a serious inductivist, though, it doesn't jump to conclusions. It waits all the time, observing in all kinds of environments, on Wednesdays and Thursdays, on warm days and cold days, on rainy days and dry days. Every day, it writes new observations into its notebook. In the end, it was not until it had collected a large number of observations that it satisfied its inductive consciousness and came to the conclusion of inductive reasoning – "My feeding time has always been 9 a.m. Unfortunately, this conclusion turned out to be completely wrong, and on Christmas Eve, instead of being fed, it was slaughtered and served to the human table.
Russell's turkey paradox illustrates that an outcome based on inductive reasoning based on a true premise is not only not necessarily correct, but can also be catastrophic. Therefore, this question is what Russell called one of the "central problems of philosophy". Nowadays, almost all mainstream artificial intelligence is based on inductive reasoning, and they need to be constantly "fed" with massive big data before they can finally draw conclusions that are not so reliable, even ChatGPT, which exploded last year.
The knowledge obtained based on inductive reasoning is not only unreliable, but also lacks knowledge innovation because it cannot obtain knowledge that cannot be observed in the data. Russell also gave an obvious example: observing the sunrise every morning does not prove that the sun will rise tomorrow. As Hume put it, our belief that the sun will rise tomorrow is nothing more than a "habit of association."
So, Eric Larsen writes in the book "The Creation of God": "It is foolish to draw out the 'habit of association' without looking deeply at the laws. In his opinion, the knowledge we think we have may be wrong, and the current artificial intelligence is not intelligent. True artificial intelligence must be able to perform three kinds of logical reasoning at the same time, just like humans. The three types of logic are: deductive reasoning, inductive reasoning, and retrospective reasoning.
Among them, both induction and attribution can make mistakes, and although deduction is reliable, the scope of application is limited, so the intelligence of both the three can be called "intelligence". Unfortunately, the current artificial intelligence is basically based on inductive reasoning, and artificial intelligence based on deductive reasoning also has great limitations, while artificial intelligence based on retroactive reasoning has not yet appeared, which has become a big blind spot.
In short, the current artificial intelligence has gone too far on a path of deviation from intelligence, and is doomed to a dead end because it violates the underlying logic of "intelligence". The vision that the era of intelligence will soon be imagined is nothing more than a dream. We underestimate human intelligence and overestimate artificial intelligence.
This is called "techno-kitsch" by Eric Larsen, the author of "The Creation of God." He believes that artificial intelligence is now worshipped as a god, but unfortunately it is a false god. Only by breaking down technological kitsch can we usher in real technological progress. The book "Making Gods" profoundly exposes the research status of artificial intelligence for the first time, destroys the myth of artificial intelligence, and deserves to be a good book to refresh the mind and see the truth of the current artificial intelligence.