AI is an abbreviation of “Artificial Intelligence”, and in the dictionary definition, it is described as “a computer system equipped with the functions of human intelligence such as learning, reasoning, and judgment.” (Excerpt from Daijirin 3rd Edition)
However, from an academic point of view, the term “artificial intelligence (AI)” is ambiguous, and different people think of it differently. At present, the definition of artificial intelligence is not clearly defined even among experts.
Secondly, in the definition of artificial intelligence, it is pointed out that “since the definition of intelligence is not clear, artificial intelligence cannot be clearly defined (Osaka University, Minoru Asada)”. Based on the difficulty, the various views can be summarized as follows.
- A system that can imitate humans and simulate brain activity
- Intelligence beyond human level
- Computers that can be “tightened”
Classification of artificial intelligence (AI):
Artificial intelligence (AI) can be broadly divided into several types.
For example, a “computer that thinks like a human” and a “system that replaces some of human abilities” are both defined as artificial intelligence, but they have completely different functions.
Specifically, it can be classified into ” specialized AI ” and ” general purpose AI “, and ” strong AI ” and ” weak AI “.
specialized AI? Specialized AI refers to artificial intelligence that automatically learns and processes tasks in a limited area.
Specifically, it is artificial intelligence with technologies such as image recognition, voice recognition, and natural language processing (detailed in a later chapter). Artificial intelligence, which is currently widely used in the business domain, is specialized AI.
general- purpose AI? General-purpose AI refers to artificial intelligence that can handle various tasks in the same way as humans, rather than responding only to specific tasks.
Even if an unexpected event occurs, human beings can make a comprehensive judgment based on their experience and solve the problem. In this way, artificial intelligence with human-like problem-solving ability is general-purpose AI. At present, the method for realizing general-purpose AI has not been clarified.
In the first place, if artificial intelligence is a computer with human-like intelligence, there is also the idea that general-purpose AI is true artificial intelligence, and specialized AI is nothing more than a machine that automates problem solving.
In addition, there is a classification of strong AI and weak AI advocated by American philosopher John Searle
strong AI? Strong AI refers to artificial intelligence that has human-like self-consciousness and is capable of tasks that require full cognitive ability
refers to artificial intelligence that replaces only part of human intelligence and processes only.
“Strong AI” and “weak AI” are concepts classified from the viewpoint of “whether artificial intelligence has human consciousness and intelligence”. On the other hand, “specialized AI” and “general-purpose AI” are concepts classified from the viewpoint of “problem processing” such as “can we handle a wide range of problems like humans?”
In other words, the relationship between “strong AI” and “weak AI”, and “general-purpose AI” and “specialized AI” is categorized according to the viewpoint from which artificial intelligence is judged . “Strong AI” and “general-purpose AI”, and “weak AI” and “specialized AI” can be said to be similar concepts with different perspectives.
- Applications of Artificial intelligence
AI (artificial intelligence) also used in computer programs for image recognition and board games . Not only does it have a wide range of applications, but it is also steadily being used in business.
Since successful cases of AI are often introduced in the media , some people may think that “with AI , it can be applied to any field of business.” But is that really the case?
Therefore , I will explain what fields AI can be applied to and what kind of business AI can be used for.
AI , which has become popular in recent years, is a machine learning method called “deep learning .” It is this deep learning that supports Go’s computer program ” AlphaGo, ” As a matter of fact, in addition to deep learning, AI methods include search trees, etc., but they are used together with AI without distinction.
Processes that can be executed by deep learning
So what kind of intellectual work can be done by deep learning?
- Image processing:
image processing is one of the typical processing that deep learning is good at . Professor Geoffrey Hinton of the University of Toronto, Canada came up with the idea behind deep learning, and Professor Hinton participated in the image recognition contest ILSVRC to check the performance of deep learning . The research group led by Professor Hinton in this contest drastically updated the conventional error rate and made it excellent, which made the name of deep learning famous.
For example, deep learning requires big data to recognize the face of a particular person. By learning various face data, such as when you are not wearing anything or when you are wearing glasses, you will be able to identify a specific person even when you are wearing a mask,
- Language processing:
Language processing is also one of the tasks that deep learning is good at. A familiar example is a voice response application like Siri . The voice response application converts the sound wave obtained from someone’s utterance into text data. Next, by estimating the text data that has a plausible meaning from the multiple generated text data, it is recognized that, for example, “What is the weather today?” Was spoken.
- Decision support:
Decision-making is the act of choosing the best choice from a given choice. Decision-making issues that have been studied in areas such as economics and business administration are also closely related to game theory, which analyzes the strategies of games involving multiple players. Decision-making problems are the subject of AI research , such as board games such as Go, Shogi, and chess being analyzed from the perspective of game theory .
What is the application of AI to business?
Deep learning is very useful for image processing and language processing, but what kind of business can it be applied to?
- Consideration of business application started mainly for IT companies.
Since 2015 , the use of deep learning has progressed mainly in IT companies.
It was very difficult to develop the infrastructure on our own because a large computer was required to learn big data.
However, because IT companies provide cloud computers that can use AI from terminals such as Amazon ‘s AWS and Microsoft ‘s Azure , it has become relatively easy for any company to use AI.
- Applicable from manufacturing to medical care and agriculture:
There are a wide range of fields in which AI can be used , including many fields such as manufacturing, mobility, and medical care. Big data is important here . In order to operate deep learning, big data for learning is required. It’s not just the data that is sent and received on the Web that is used as big data . AI will be able to handle more data by linking with the IoT (Internet of Things ).
A sensor is mounted on the terminal, and data from the outside world can be acquired from the sensor. One example is autonomous driving. In order to drive autonomously, it is necessary to detect obstacles and pedestrians placed on the side roads and other vehicles moving back and forth and left and right to determine the driving route. Therefore, it is necessary to perform machine learning based on the data acquired by the innumerable sensors attached to the car. With the increase in big data for learning, the range of applications of AI to business is also expanding.
- Future challenges for artificial intelligence
It is said that artificial intelligence, which will continue to evolve in the future, has three major issues: “responsibility”, “judgment criteria for artificial intelligence”, and “increasing unemployment”.
・ It is difficult to judge where responsibility lies
Responsibility is the question of who should be responsible in the unlikely event that a problem occurs using artificial intelligence. For example, let’s say that leaving the judgment to artificial intelligence leads to a big accident. In this case, it is very difficult to judge whether the person who handles artificial intelligence has something wrong with it or whether it is the fault of the person who developed artificial intelligence.
Since the artificial intelligence that actually made the decision cannot take responsibility, it is necessary to establish a criterion for the responsibility and clarify what to do when a problem occurs.
・ I don’t know the criteria for artificial intelligence
Artificial intelligence analyzes and makes decisions as needed from a huge amount of data. However, since all of these operations are performed automatically, it is sometimes difficult to understand the rationale.
If a human makes a decision, we can explain the process and reason for making the decision, but in the case of artificial intelligence, we cannot explain the process, so the issue is how to respond in situations where explanation is required. I am.
・ More people will be unemployed
It is predicted that the work that humans have done so far will be mechanized one after another. If artificial intelligence surpasses humans, the fields of application will increase, eliminating the need for human resources and creating a large number of unemployed people.
In addition, if a certain occupation is eliminated by artificial intelligence, all the people in that occupation will be unemployed, which may further increase the unemployment rate. It is quite possible that human technology and intelligence will deteriorate due to excessive reliance on artificial intelligence.
Many people think that “human beings are dominated by artificial intelligence,” but it can be said that measures must be taken from now on to prevent this from happening.