Why AI is smarter than humans?
The field of artificial intelligence has made rapid progress in the last ten years. Will the computer catch up with humans in terms of intelligence and creativity?...
Why AI is smarter than humans? Will computers become as smart as humans?
And then what? Artistic creativity is considered by many to be a unique selling point of man, which can never be achieved by seemingly cool calculating machines.
But the computers are catching up:
For example, how long does it take a furniture designer to sketch a dozen design proposals for a new armchair that looks like an avocado?
If you assign this task to Dall-E, an artificial intelligence (AI) developed by the US company OpenAI over the past two years, then within a few minutes you will receive completely different and usually meaningful suggestions.
Regularity experts:
Dall-Ai is just one of several highlights of AI research that have amazed experts and the public with amazing regularity over the past ten years. Based on so-called deep neural networks, algorithms achieved new top performances (PDF) in areas such as image recognition, text comprehension or motion control and defeated human champions in games such as Go, Starcraft and poker.
AI and in particular the neural networks:
In view of this rapid progress, the question arises as to where this development will lead. AI and in particular the neural networks are inspired by human intelligence and the biological structure of the brain. Will you achieve and maybe even surpass this role model? And what impact would this have on human society? While this question is seriously discussed in academic circles, especially in the Anglo-Saxon area, it seems that in continental Europe the focus is on issues of data protection or the prevention of discrimination by algorithms, which are more relevant in the short term. This article pleads for taking the possibility of a future AI on a (super)human level seriously, and points out that the social consequences would go far beyond the mentioned and already practically relevant problems today.
What can AI do today?
It is a hallmark of today's AI that it is successful in very limited fields of tasks and sometimes surpasses human experts. However, it is generally a big problem for the known algorithms to transfer what they have learned to new situations or contexts. For example, an appropriately trained network can recognize different dog breeds in photos. But it has no understanding of what dogs actually are, and does not know anything about their needs or behaviors.
The goal is to obtain a mathematically well-defined cost function:
This insular talent of today's algorithms is related to the fact that they are usually trained by so-called supervised learning. The goal is to minimize a mathematically well-defined cost function - for example, an error rate in image recognition or the number of lost chess games. Such a clear success criterion is also often used in computer science as a yardstick by which different working groups can compare their results. But it also leads to very one-sided capabilities on the part of the finished algorithms.
Goal:
The human brain, on the other hand, learns largely unsupervised and thus not towards an externally predetermined goal. Although the survival of the species can be understood in the evolutionary sense as an optimization problem posed from the outside, the learning mechanism of natural selection can only affect the genetically determined architecture of the brain, but not the special knowledge and skills of the individual individual. Nevertheless, people can easily solve a very wide range of very different tasks - be it recognizing another person, maintaining balance on a bicycle, learning a musical instrument or planning a vacation trip.
Given this enormous variety of cognitive abilities, it is difficult to grasp what exactly is meant by intelligence.
Why AI is smarter than humans?
Rokid asks: Can artificial intelligence really be smarter than humans?
Sun Zhengyi said: By 2040, the number of robots in the world will reach 10 billion, thus exceeding the human population, and the singularity will come. Einstein said that artificial intelligence will eventually destroy people.
So do you think it was Einstein or the wheelchair itself?
Einstein said that artificial intelligence will eventually destroy people.
So do you think it was Einstein or the wheelchair itself?
Does this mean that artificial intelligence:
Ke Shao has been fighting Alphago for two days. There is no tension. Alphago will definitely win. Does this mean that artificial intelligence is smarter than humans? which is trained by reinforcement learning. In addition, artificial intelligence is much inferior to humans in many aspects, such as attention, thinking and imagination, which are inferior even to 6-year-old children, not to mention creativity and emotions.
Go-Play:
Then take a look at this Go game, the performance of AlphaGo is not like the previous ones, but more and more human, but more and more incredible.After the game, Ke Shao said that the point is not that AlphaGo is getting worse, but that this human ability is not good enough, so people need With the help of artificial intelligence we can check ourselves, cancel the prejudices that we thought we had to do or could not do and re-understand Go.
Look, this is the prudence of people. People know how to reflect on their own cognitive deficits, how to use external forces to improve themselves, to show the human mind. He lost in the Go, but at least Ke Shao won.
Not only to surpass human intelligence in terms of computing power:
Of course, artificial intelligence not only seems to surpass the computing power of man, but also gradually replaces man in factories, hospitals and law firms, but man has a general intelligence, which is a combination of different intelligences, and intelligence is far from achieving this. In the future, when data and computing power are getting stronger and stronger, whether a strong artificial intelligence should be achieved or not? Let's get to the singularity, I don't know.
Primary:
In short, I personally think that the most important thing is that we need to learn how to use artificial intelligence, how to use artificial intelligence to improve all spheres of life and how to improve the continuous development of people fördern.In in the future, people will have to use artificial intelligence to collect and process data in the cloud, which will make our life more and more interesting.
