Read the difference between artificial intelligence and human intelligence in one article

Nowadays, it is easy to believe that artificial intelligence has become as smart as human wisdom - if not smarter. For example, some time ago, Google released Duplex AI, which can help users complete outbound calls to book hair salons and restaurants. Because its voice is almost exactly the same as human beings, it is possible to pass through its dialogue partners throughout the process and make it think that it is human.

In addition, Google's subsidiary DeepMind has developed an artificial intelligence that beats the world champion in the most complex board games. Recently, artificial intelligence has proven to be as accurate as a trained doctor to diagnose eye diseases... There are many events that indicate that in the near future, robots may put humans in an unemployed state. .

With the development and breakthrough of technology, artificial intelligence unlocks new fields, new tasks, and new skills at a speed we can see with the naked eye, which were previously considered to be exclusive to human intelligence. But does this mean that artificial intelligence is smarter than human intelligence? In the view of ID: Aiobservation, it is a wrong idea to compare artificial intelligence with humans, because the two are completely different things, even if their functions overlap.

Artificial intelligence is good at processing data, not good at abstract thinking

First, even the most complex artificial intelligence technology, its core is no different from other computer software: running data at ultra-fast rates. AI and its branches, such as machine learning and deep learning, can solve any problem as long as the researcher can convert it to the correct data set.

For example, image recognition. Given the deep neural network, the underlying structure of the deep learning algorithm, and enough markup images, artificial intelligence can compare data in a very complex way and find correlations and patterns that define each type of object. Information marks objects in images that have never been seen before. Of course, the process of speech recognition is also true: if there are enough digital samples of human voice, the neural network can find a common pattern in the human voice and determine if a certain recording belongs to that person.

Recently, the online AI AI Jianhuang voice anti-spam service is based on this theory. In addition to identifying pornographic images, pornographic videos and pornographic texts, the yellow-related voice can also be identified by AI. In order to let the AI ??intelligent machine have the ability to recognize multi-language and multi-language words, it is necessary to have a training and learning process in advance. The product of the Ali Security Department, Nian Xia, said, “You can think of it as a child and need to constantly feed and train. Learning, it has such ability." For example, learning Cantonese, in addition to purchasing training materials from third-party companies, also use the Cantonese TV series on the video platform in the Ali system to train robot learning.

In fact, what we have learned about AI is whether it is a computer vision algorithm for face recognition or cancer diagnosis, or an artificial intelligence network security tool that can evict malicious network traffic, or even a complex AI for playing computer games. Projects have such a same rule. However, technology is constantly changing and improving.

As there is an old saying that "no one is perfect," AI also has its own shortcomings, and what it lacks is abstract thinking, the use of common sense, and the transfer of knowledge. Speaking back to the Google Duplex AI mentioned at the beginning, it may be very good at booking a restaurant or hair salon, but this is two very narrow and very specific tasks. Even this artificial intelligence can use human language tones to complete a natural behavior that mimics human dialogue, but once the conversation deviates from the orbit, Duplex will have difficulty answering in a coherent way. In this case, it either ends the conversation or, with the help of humans, can continue the conversation in a meaningful way.

So far, there have been many examples to prove that once an AI model appears in an event outside its domain of expertise or receives content that is different from the data they have trained, they fail in an illogical way. The wider the scope, the more data the artificial intelligence needs to master, and there are some edge cases that have not been covered by the training data, which will eventually lead to the failure of artificial intelligence. One example is the self-driving car. Although it has traveled tens of millions of kilometers, it is still trying to achieve complete autonomy, far exceeding the need for human beings to become expert drivers.

Humans are not good at processing data, good at making abstract decisions

Start with the data section. Contrary to computers, humans are very poor at storing and processing information. For example, if you want to remember the lyrics of a song, you must listen to it multiple times to remember it; but for a computer, remembering a song is like pressing "save" in the application or copying the file to its hard drive. It's as simple as that. Similarly, for humans, it is very difficult to remember. Even if you do what you can, some not very good memories will still exist in your own mind. For computers, forgetting something is as simple as deleting a file.

To be honest, human beings are far less skilled than artificial intelligence in dealing with data. In all of the examples mentioned above, humans may be able to perform the same tasks as computers, except that in the time it takes for humans to recognize and mark images, the AI ??algorithm can complete the classification of one million images. It is no exaggeration to say that the absolute processing speed of computers allows them to surpass humans in any task involving mathematical calculations and data processing.

However, it should be noted that human beings can make abstract decisions based on instinct and common sense in the case of scarce information. For example, human children learn to organize items at a very young age. But for the AI ??algorithm, it takes years of training to perform the same task. When asked about the difference between smart machines and humans, tech critic Nicholas Carr said, “Computers are not crazy, they can't be contradictory, and they can't be designed to deal with ambiguity. They don't have intuition. ”

For example, when people first come into contact with video games, they can quickly transfer knowledge from everyday life into the game environment, like away from pits, ledges, fires and sharp things (or skip them). They know that they must avoid bullets and avoid being hit by vehicles to survive. But for AI, each video game is a new, unknown world, and it must be learned from scratch. Humans can invent new things, including all the technologies that have led the era of artificial intelligence, while AI can only acquire data, compare it, propose new combinations and presentations, and predict trends based on previous sequences.

Human beings can feel, imagine, dream, be selfless or greedy, love and hate, can lie, and sometimes even confuse the facts. All of these emotions can change their decisions in a rational or irrational way. Man is an imperfect and defective creature made of flesh. Everyone lives uniquely in his own way. Artificial intelligence, at the core, is tiny by dozens of inanimate circuits. Current.

Artificial intelligence is very different from human intelligence, please stop comparing

In general, all of this does not mean that artificial intelligence is superior to human intelligence, and vice versa. Because in essence, the two are completely different things. AI excels at repetitive tasks that have well-defined boundaries that can be represented by data and often perform poorly for tasks that require incomplete decisions to make decisions based on intuition. In contrast, human intelligence is suitable for environments that require common sense and abstract decision making, but not well for tasks that require extensive real-time computing and data processing.

From a different perspective, we should consider AI as enhanced intelligence. Artificial intelligence and human intelligence complement each other and make up for each other's shortcomings. Therefore, they can work together on any task that they cannot do alone. For example, AI is good at using a lot of network traffic to find out anomalies, but it makes mistakes when deciding which ones are real threats to investigate. On the other hand, human analysts are not good at monitoring gigabytes of data transmitted over corporate networks, but they are good at associating exceptions with different events and determining which are real threats. Therefore, AI and artificial analysts can fill each other's gaps.

Admittedly, artificial intelligence can now do more and more things, and their logic becomes more complex, so they have the ability to deal with more complex situations and handle more variable parameters. But artificial intelligence and human strengths are reflected in different fields, which means that we need to explore symbiotic relationships, not competitive ones.

Many people hold the idea that machines will replace human jobs. In the view of Aiobservation, there is no lack of exaggerated propaganda. It turns out that the expansion of artificial intelligence has created more job opportunities than destruction. But, as every technological breakthrough in the past has done, it does eliminate the need for humanity in many missions. But this may be because those jobs are never aimed at humans. Now we spend a lot of valuable human resources and labor on these jobs, perhaps because we have not yet developed a technology that can automate them.

As artificial intelligence becomes more adept at performing more tasks, as human beings, we will have more time to apply our wisdom to creative, social, artistic, sports, literature, poetry, and other valuable applications. By that time, we will use our enhanced smart tools to increase the creativity of these jobs.

Finally, we are not machines, machines are not humans, and the future will be built by artificial intelligence and human intelligence!