What is artificial intelligence (AI)?

Today, the word AI has become a familiar word, but how many people understand its essence and mechanism?
It is thought that many people are in a situation where they “somehow understand”.
In this article, we will explain in detail the meaning and definition of the word AI, the advantages and disadvantages of introducing it, and examples.

What is AI

AI is an abbreviation for Artificial Intelligence, where Artificial means “artificial” and Intelligence means “intelligence”.
However, the definition of AI (artificial intelligence) is not clear and precise.
The current situation is that the way researchers perceive things is changing along with the flow of the times.
Overall, the common point is that they have intelligence like humans, and AI is characterized by learning by itself.

Evolution of AI and the Current Situation

We will take a closer look at how AI has evolved and what capabilities the current AI has.

Birth of AI

AI is related to the question “Can machines think?” proposed by British mathematician Alan Turing in 1950.
In 1956, at the Dartmouth Conference held in the United States, mathematician Joan McCarthy proposed that “artificial intelligence is a machine that thinks like a human,” and this is considered to be the origin of AI.

First AI boom (1960s to early 1970s)

At that time, research was conducted on AI that performs “inference” and “exploration” under clear rules.
Inference refers to the symbolic representation and attempts to carry out human thought processes.
Search, on the other hand, refers to finding out how to solve the target condition by dividing it into different situations.
However, AI can only handle things like simple games, and was described as a “toy problem”.

What attracted attention as an AI of this era was a language processing program called “Eliza” developed in 1966, which returned a response (result) as if it were a conversation when instructed in short sentences. was.
It is also the origin of “Siri” used in modern iPhones.

Second AI Boom (1980s)

In the 1980s, a technology known as an “expert system” was developed to train artificial intelligence to use knowledge as rules to solve problems. This triggered the second AI boom.

When people face a problem, they think of a solution based on their experience and knowledge, but the expert system reproduces this with AI and thinks it is correct by reasoning based on theory from various knowledge bases. guide the data.

For example, it is an expert system or similar algorithm that displays recommended products based on the browsing history of a mail-order site, or frequently displays topics that are similar to the theme of the Internet information that is often viewed daily.

Third AI boom (late 2000s to present)

Currently, we are in the midst of the third AI boom caused by the emergence of deep learning, with the practical application of machine learning progressing.
Improvements in the computing power of computers and the ability to utilize huge amounts of data are the factors that have greatly advanced AI technological innovation.
In addition, the practical application of machine learning and the emergence of deep learning are also major factors.

machine learning

Machine learning is the process of finding rules and patterns from input data and automatically constructing algorithms that can identify and predict by applying them to the derived data. is.
Since 2010, as computing power has improved, it has become possible to handle huge amounts of information, and machine learning has been put to practical use.

deep learning

Deep learning is a type of machine learning that is an epoch-making method of deep thinking to discover the rules and patterns behind data.
By connecting multiple layers of algorithms called neural networks, which were developed using the human nervous system as a model, it is now possible to perform processing with high accuracy, which was difficult with machine learning before deep learning, such as image recognition.

Ray Kurzweil, an authority on AI research, points out the prospects for technological innovation as follows.

• AI will have human-level intelligence by 2029
• Technological Singularity Will Come in 2045

Current AI technology

How has AI changed our daily lives? Here are some representative examples.

automatic translation

Language translation technology using AI has evolved tremendously.
With the advent of deep learning, translations are becoming more natural and closer to human speech.
AI can speak in the translated language just by speaking to your smartphone, and you can convert a foreign language website into Japanese with a single button, and it is used in familiar places.

Regtech (Compliant with financial regulations)

“Regtech” is a type of AI technology that has been introduced in financial institutions.
This system monitors risks using AI that has learned patterns of fraud that have occurred in the past, such as checking fraudulent use of credit cards, online identity verification (eKYC), and anti-money laundering (MLC). are widely used in anti-fraud measures.

Benefits of AI

Explain the benefits of introducing AI.

Elimination of labor shortage

AI is better at many tasks than humans. Replacing the work that AI is good at with humans will solve the labor shortage.
In countries such as Japan, where there are concerns about a declining labor force, it is expected that AI will become even more active.

Increased productivity

By using AI, it is possible to produce a large amount of output at a low cost and in a short amount of time.
Especially for monotonous and repetitive tasks, AI is often more productive.

Cost reduction

By replacing the routine work that humans have done so far with AI, it is possible to reduce the number of workers, leading to a reduction in labor costs.

Improving social safety

By using AI instead of humans, it will be possible to perform dangerous tasks and work in places where humans cannot enter.
In addition, using AI to automatically detect infrastructure deterioration and machine failures will help prevent accidents.

Disadvantages of AI

I will explain the disadvantages that occur when AI is introduced.

decrease in employment

There is concern that AI will reduce the number of jobs that have been performed by humans in the past.
However, it is also expected that the introduction of AI will create new occupations that have never existed before, and conversely, employment may increase.

security risk

Since AI possesses a huge amount of useful information, there are concerns about information leaks due to external hacking.
Therefore, countermeasures against security risks are an important issue.

Issues in risk management

As the introduction of AI increases, troubles will also increase if problems arise.
Experts can predict the scope of the problem, but it is difficult for laymen, so a test run with a limited introduction is necessary.
It is also a good idea to use a consulting service to manage AI risks instead of dealing with them in-house.

Temporary increase in cost

A major disadvantage for companies considering the introduction of AI is the temporary increase in costs.
For AI to show its full potential, it is necessary to review the entire business flow, and switching the service used from the core of the system will take a considerable amount of time and cost.

In the long run, it can reduce labor costs and costs, but it is important to note that the initial investment for introduction requires a huge amount of money.

6 AI Use Cases

Here are six examples of how AI is being used.

Case 1 | Marketing

A major credit company analyzed cashing revolving data used by more than 3 million members and implemented a measure to send DMs to those with the highest usage rates.
As a result, the DM response rate doubled and the number of buyers increased.
However, due to the enormous amount of accounting processing, we were able to automate the selection of account items that required specialized knowledge by using AI, and we were able to handle this without increasing the number of staff.

Case 2 | Sales

Insurance companies are implementing insurance proposals to customers using AI.

Based on the information received from the customer using the device and the insurance subscription status, it is possible to propose insurance content that meets the customer’s needs.
Regardless of the number of years of experience, we can propose insurance that meets customer needs, so we can expect a steady increase in orders.

Case 3 | Defective product detection

Major food manufacturers are using AI in their production lines to automate the detection of defective products.
Cameras are used to photograph the materials flowing through the production line, and AI that identifies defective products has made it possible to detect defective products with a high degree of accuracy.
Especially in food, the contamination of defective products poses a great social risk, so an accurate AI-based detection system was extremely important.

Case 4 | Equipment Management

A machine factory introduced a system that uses AI to detect equipment abnormalities based on data such as temperature, humidity, and pressure during manufacturing.
The system learns from normal data and detects equipment abnormalities based on discrepancies with the data, which boasts extremely high accuracy.
Another big advantage is that it can operate 24 hours a day, 365 days a year.

Case 5 | Accounting

A small and medium-sized enterprise has introduced cloud accounting software and is gradually automating accounting operations with AI.
Accounting operations require a certain amount of specialized knowledge, but AI automatically selects and displays account items when deposit/withdrawal data is imported, so even beginners can handle it.
There is also a service that uses OCR technology to directly read handwritten characters to directly read handwritten slips and issue journal entries, and automation is progressing rapidly.

Case 6 | Confirmation work

Genial AI | Tools specialized for AI-OCR and Excel matching

AI is an Excel add-in that makes checking work more efficient by associating Excel detail lines with documents and highlighting where cell data and document data semantically match.

• Utilizing the know-how cultivated to achieve a matching accuracy of 94% in a demonstration experiment with one of the four major audit firms (Big 4)
• Efficient matching and confirmation work between Excel details and documents using AI
• Transcription of characters and tables using AI-OCR
• With a subscription model based on the number of accounts,
  it can be introduced at a reasonable price of 10,000 yen per month

You can improve the efficiency of detailed check work using Excel and PDF, reduce human errors, and improve business efficiency. Please feel free to contact us.

 

summary

AI is a system that has human-like intelligence and can think, act and make decisions.
It is also possible to improve accuracy by learning a huge amount of data by itself.
While the introduction of AI brings significant benefits such as improved operational efficiency and productivity, people tend to think negatively about future employment, saying that AI will take away jobs.
However, AI not only makes operations more efficient, but it is also very useful in preventing human errors.
Wouldn’t it be dozens of times more efficient to let AI do what AI can do, and to dig deeper into areas that only humans can do?
Accurately catching AI information and successfully coexisting with it will be the key to achieving dramatic efficiency improvements.