Supercomputer blundered

“The IBM Watson supercomputer can now participate in the debate, IBM has created artificial intelligence,” read all the technology news feeds . They refer to the results of the Milken Institute Global Conference at which IBM introduced the new Debater feature of its supercomputer, without five minutes of artificial intelligence, Watson, which is that the computer supposedly can freely and reasonably discuss any given topic like an ordinary person.
The news was received with a bang - quite a while we had not heard of progress in the development of artificial intelligence.
Unfortunately, in the wake of joy, few people bothered to check whether Watson really has the analytical skills or its creators decided to deceive the audience and imitate the activity.
I examined Watson’s opinions in detail and came to very disappointing conclusions - the debate failed.

Let's see together what is wrong with the smartest computer of the decade.
The task was set as follows: to speak out for and against restricting access to violent computer games for children ( 45th minute of the video )

It was assumed that Watson, after analyzing the entire Wikipedia, will draw up its “opinion” based on it .
The first argument, the computer quoted the Wikipedia archive:

The impact of violent video games leads to an increase in physiological arousal, aggressive thoughts and feelings, and also worsens social behavior.
(Exposure to violent video games results in increased physiological arousal, aggression-related thoughts and feelings, as well as decreased pro-social behavior.)

The phrase uttered is completely copied from the Wikipedia article Disputes about video games (Video game controversies) , but it’s just bad luck - at the beginning of March it was this phrase that was edited, it is no longer on Wikipedia.
It turns out that Watson takes information not directly from the Internet, from current sources, but from a copy of Wikipedia, which was downloaded to it at least two months ago - this is a huge time for a computer that claims to be omniscient. Or are the developers deliberately slowing down the development of artificial intelligence?

Next phrase

In addition, these violent games, or the lyrics are the cause of manifestation of teenage aggression in real life
(In addition, these violent games or lyrics actually cause adolescents to commit acts of real-life aggression.)

torn out of context. Why did IBM talk about lyrics? He was not asked about this. The computer just found a phrase with similar words and uttered it, but just didn’t realize that the source article was about cruel songs.

Finally, the violent video games can increase children's aggression
(Finally, violent videogames can increase children 's aggression.)

finished the computer his arguments “for” another copy of similar words from Wikipedia . Moreover, this “argument” repeats the previous two.

We pass to the arguments against. Watson began with an unfounded statement that can hardly be called an argument, and, of course, copied from Wikipedia :

C the other hand, violence in video games is not associated with aggression
(Violence in videogames is not causally linked with aggressive tendencies)

Most of the children who play violent video games, do not have problems
(Most children who play violent video games do not have problems.)

this ( copied ) phrase is a classic argument in sophistry - any opponent will immediately notice that what matters is not the fact that most players are not aggressive, but that gamers are more aggressive than ordinary people.

The last argument against is also not logical:

Video games
play is part of an adolescent boy's normal social setting

If something is normal in one environment, this does not mean at all that it will be normal in society as a whole. So, society will only lose from the fact that all adolescents will be cruel and aggressive and will consider it normal.

So to summarize:
Watson has access only to its internal obsolete databases
Watson just copies parts of sentences similar in composition to Wikipedia
Watson rips phrases out of context, if only there are keywords
Watson doesn't think at all about the logic and quality of arguments
Watson is relatively good related proposals
Watson makes the sentences almost pure.

Thus, Watson and the entire IBM Research team miserably failed their first public debate.

To some extent, this is not surprising - let's try to find some information on the IBM website . Oh what is that? Does search results work poorly? Could at least put a search from Google, or from Mail.ru at least.


Unfortunately, in order to claim the title of Artificial Intelligence in the 21st century, it is already not enough to have a semantic search and a database of connecting words, you really need to reason, think, at least. So far, IBM has not been able to realize this, although they have tried very hard to make a sensation - because otherwise, why demonstrate a publicly poorly working technology?
Yes, a supercomputer can find the answer to the quiz question on Wikipedia faster than anyone, yes, it can calculate all the moves in a chess game - these are all the tasks of a dumb machine. But he still has not learned how to reason independently, at least to some extent, to create as a person.
We do not need false sensations.
Maybe another time, IBM.