How our understanding of AI is changing - from chess to prediction machines
In previous articles, we have traced the history of artificial intelligence and explored the very nature of AI and the Moravec paradox. Today, I'd like to continue our discussion by looking at how our perceptions of intelligence have evolved over the years in the context of AI development - from the days when playing chess was considered the pinnacle of intellectual ability, to contemporary concepts that view AI as advanced "prediction machines."
Chess as a determinant of intelligence
The year 1997 and the victory of the IBM Deep Blue computer over chess grandmaster Garri Kasparov was a moment that shocked the world and was considered a breakthrough in the development of AI. As I mentioned in my article on the history of AI, chess has long been seen as an indicator of high intelligence, and chess champions were regarded as individuals with exceptional mental abilities. The victory of a computer over a human in this game seemed to confirm that AI was capable of reaching human levels of intelligence, and even surpassing them.
However, in retrospect, our view of chess and its relationship to intelligence has changed. As I wrote in an article on the essence of artificial intelligence, we have come to understand that chess is not the pinnacle of human intellect, but rather a mathematical problem with clearly defined rules and a finite set of possible moves. In this sense, a chess program is not significantly different from a scientific calculator - it can perform tasks difficult for a human, but this does not mean that it thinks in the same way.
From "intelligence" to "prediction machines"
As technology has advanced, we have begun to question whether the term "intelligence" is appropriate when referring to AI. Some researchers suggest that the use of the word "intelligence" can be misleading, as it falsely implies that machines can do the same things as humans. Instead, AI is increasingly being referred to as "prediction machines." This term better captures AI's ability to analyze data and predict future trends, events or outcomes.
Modern AI technologies are not about "real" intelligence, but about the ability to make predictions based on the analysis of huge data sets. In this context, AI is comparable to oracles, which make predictions but have no human personality, conscience or emotions. As we noted in the Moravec paradox article, AI does what humans do, but in a completely different way.
Moravec's paradox in the context of AI evolution
Moravec's paradox, perfectly illustrates this difference - what is easy for humans, such as face recognition, is very difficult for computers, and what is difficult for humans, such as playing chess, turns out to be easy for AI. This is because human skills related to perception and motor skills have evolved over millions of years, while the ability to think abstractly is relatively new.
Summary
Our journey from chess to "prediction machines" shows how much our perception of intelligence has changed in the context of AI. From the enthusiasm of chess victory, we have moved to a more realistic and critical view of AI's capabilities and limitations. We have come to understand that AI is not an "artificial human," but a new kind of entity with different abilities and mechanisms of action. The concept of "prediction machines" better reflects the characteristics of modern AI systems, which, through data analysis, are able to predict and optimize various processes in our world.
I invite you to share your thoughts in the comments!
Note!
This article was prepared with the support of Claude 3.5 Sonnet, an advanced AI language model that helped organize and present content while maintaining an authentic message and educational value.This article has also been automatically translated from Polish using DeepL. If you find any errors, please let me know in the comments.
