Friday, June 14, 2024
HomeAISummary of AI Problems | Commentary on Decrease in Employment, Location of...

Summary of AI Problems | Commentary on Decrease in Employment, Location of Negligence Liability, Military Use, etc.

The spread of AI in recent years has been remarkable, and we are surrounded by things that utilize AI.

Although AI enriches our lives, there are also many problems. So, this time I will introduce the problem of AI.

Table of Contents 

  • What is AI
  • Seven pain points in AI
    • ① Take away human jobs
    • (2) Responsibility issues in the event of failure
    • (3) Problems associated with decisions made by AI
    • ④Privacy issues
    • ⑤ Singularity problem
    • ⑥ Risk of military use
    • ⑦ Data security issues
  • 5 examples of problems in AI
    • (1) Decrease in Goldman Sachs traders
    • ②Fatal accident by Uber AI car
    • ③ Automatically classify black people as gorillas with Google Photos
    • ④ Use of AI-equipped drones in the Nagorno-Karabakh conflict
    • ⑤ Tesla car hacking by Keen Security Lab
  • Will AI’s problems be solved in the future?
  • summary

What is AI

AI is an abbreviation for artificial intelligence. AI technology such as machine learning is playing a major role in human intellectual activities such as translation, automatic driving, medical image diagnosis, and Go.

In this way, AI is being used in various fields, but there are also many problems.

Seven pain points in AI

There are seven problems with AI as follows.

  1. take away people’s jobs
  2. Liability issues in case of failure
  3. Problems with decisions made by AI
  4. privacy issues
  5. singularity problem
  6. risk of military use
  7. data security issues

I will explain each.

① Take away human jobs

The first problem with AI is that it will take away human jobs. This problem is called the employment reduction problem.

Machines don’t need to be paid like humans do, and they make fewer mistakes than humans. .

In this way, as AI develops, AI will replace human work, and there is a problem that employment will decrease.

In order to prevent AI from stealing our jobs, we need to think about how to make our jobs unreplaceable by AI.

(2) Responsibility issues in the event of failure

The second is the question of who should take responsibility when AI makes a mistake.

For example, if an AI-equipped self-driving car causes an accident, there is the question of whether the driver or the person who created the self-driving system is responsible.

In this way, the law dealing with negligence due to AI defects has not yet been developed. In the future, it is necessary to draw a clear line by law.

(3) Problems associated with decisions made by AI

The third is the problem associated with decisions made by AI. Specifically, it can be divided into the following two difficulties.

I don’t understand the grounds for the judgment

First of all, when AI makes a decision, there is a problem that the basis for that decision is not known. In this case, the human will actually do what the AI ​​deems to be the best, without knowing why it reached that conclusion.

ethical issues

And there are also ethical issues. AI makes decisions based on rational calculations.

For example, someone might prefer the more expensive watch to the less expensive watch, even though the cheaper watch is more important to someone. There is also the possibility that AI will give priority to people who think that AI is more important in decisions related to human life.

You can see that AI, not humans, making ethical decisions tends to lead to distorted results like this.

④Privacy issues

The fourth problem is the issue of privacy. The question of how much privacy-related information can be collected by AI is one of the hotly debated themes in recent years.

In fact, ads on websites and recommended videos on YouTube are already powered by AI. Appropriate advertisements and recommended videos cannot be displayed without data such as sites visited by users and search history.

To what extent companies should collect personal information from users and how they should manage the collected personal information are issues that need to be considered in the future.

⑤ Singularity problem

The fifth issue is the so-called singularity. It is said that in 2045, AI will reach the singularity, where it will surpass human intelligence.

AI currently has no emotions. However, as technology advances in the future, AI may come to feel and think like humans.

If such robots are to live together with humans, should they be treated as human beings and should be given human rights? It can be said that this is a problem that humans will face in the future.

⑥ Risk of military use

The sixth problem is the danger of AI being used for military purposes.

Weapons such as missiles that can move autonomously even when unmanned are now introduced in most armies. Due to the spread of such independent unmanned combat aircraft, battles will become machines, and war may become easier.

In addition, you will be able to make AI attack humans directly. What if the intelligence of AI surpasses humans at that point? Humans can’t compete with it, and it may cause more damage than ever before.

Furthermore, although it is not for military use, there is a possibility that a full-scale war will start due to an AI malfunction. AI is very dangerous in a military sense.

⑦ Data security issues

The seventh problem is the issue of security.

When using AI, confidential information such as customer information will be handled using networks. In order to enable accurate analysis by AI, a huge data group called big data is required.

The problem is managing this data. If not properly managed, a large amount of confidential data will be stolen at once.

5 examples of problems in AI

Next, we will introduce examples of problems caused by products and services that actually incorporate AI. This time, we will introduce the following five.

  1. Goldman Sachs Declining Traders
  2. Death by Uber AI car
  3. Google Photos automatically classifies black people as gorillas
  4. AI-equipped drones used in Nagorno-Karabakh conflict
  5. Keen Security Lab Hacking a Tesla Car

I will explain each.

(1) Decrease in Goldman Sachs traders

First, let’s take a look at a real-world example of how AI has taken over human jobs.

At its peak in 2000, the U.S. equities trading division at Goldman Sachs headquarters in New York employed 600 traders to buy and sell stocks on orders from large investment bank clients. As of 2017, the department has dwindled to just two people.

Thus, automated stock trading programs have taken traders’ jobs.

②Fatal accident by Uber AI car

Next, we will introduce an example of an actual accident caused by an AI malfunction.

The National Transportation Safety Board (NTSB) has released a report on a pedestrian fatality involving an Uber self-driving vehicle outside of Phoenix, Arizona, in March 2018. It was the first time a pedestrian was killed in a fatal accident involving a self-driving car.

It is said that the problem was that the AI ​​installed in this Uber car did not assume that there were pedestrians on the road in the first place, so it was not possible to recognize the object as a “pedestrian” until just before the collision.

③ Automatically classify black people as gorillas with Google Photos

The third introduces an example of a problem with AI’s discriminating ability.

In 2015, Google apologized for automatically classifying African-American faces as “gorillas” in Google Photos.

Google Photos uses machine learning technology to automatically classify people’s faces, animals, plants, food, etc. in photos posted by users, and tag them to make them easier to search.

The problem was discovered when a man living in New York saved a photo of himself and his friend using the app, and the friend, who was black, was classified as a gorilla.

④ Use of AI-equipped drones in the Nagorno-Karabakh conflict

The fourth is an example of military use.

The Nagorno-Karabakh conflict is a 2020 conflict between Azerbaijan and Armenia over the Republic of Alphatu.

The AI ​​drone, which had a major impact on the battlefield in this conflict, was to find out and attack the presence of Armenian soldiers and tanks.

Until now, soldiers hiding in caves could not be identified from the sky, but this AI drone can detect the presence of soldiers from the presence of electronic devices held by soldiers and attack them.

Weapons, such as gunpowder and nuclear weapons, have greatly changed world history. This AI drone could change the course of military history.

⑤ Tesla car hacking by Keen Security Lab

The fifth is a real-life example of a security issue.

A research team at Keen Security Lab, a subsidiary of Chinese Internet company Tencent, has announced that it has successfully hacked Tesla Motors’ electric car ‘Model S’.

In addition to remotely opening the unmodified Model S doors and trunk, the research team has also succeeded in hijacking the display and remotely activating the brakes of the running Model S.

This attack by Keen Security Lab exploited a vulnerability in the web browser built into Model S, which could be activated by connecting the vehicle to a Wi-Fi hotspot.

In this way, currently distributed AI-equipped products can be easily hacked.

Will AI’s problems be solved in the future?

As we have seen, AI has many problems that need to be solved. However, using AI also has many advantages, such as improving work efficiency and resolving labor shortages. Therefore, the spread of AI will continue to be actively promoted in the future.

Along with the spread of AI-based products and services, human technology to program AI will also progress. By improving the program, the time may come when these problems can be solved.

summary

In this article, we have introduced specific examples of AI problems.

I learned that there are problems associated with the spread of AI and the need to solve them. Not only engineers but also we need to think about this problem for improvement.

RELATED ARTICLES

What is an AI algorithm?

LEAVE A REPLY

Please enter your comment!
Please enter your name here

Recent Posts

Most Popular

Recent Comments