Technology

Exploring the cause and solution of anxiety by actually developing and running an "anxiety algorithm"


Note

Exploring the cause and solution of anxiety by actually developing and running an "anxiety algorithm"

In order to find out how to control the "anxiety" that you have on a daily basis, the travel search engine “service ended in January 2020''Hipmunk, Co-founder Adam Goldstein, has developed an "anxiety algorithm" and is trying to explain the cause and solution of anxiety by actually simulating.

The Contagion of Concern — Adam Julian Goldstein
https://www.adamjuliangoldstein.com/blog/contagion-of-concern/

In operating Hipmunk, Goldstein said, "I feel good when things go well when things are going well, but I can't consider good things to happen when things go wrong." I noticed that. To solve the question of why real things affect one's cognition, Goldstein takes the biology of "anxiety as a survival strategy"From the relationship between smallpox and immune response"Anxiety algorithm" was created.

For example, in order to develop a robot that moves autonomously, it is necessary to perform programming that allows the robot to survive a crisis. Robots can only learn from past experiences, but if they recognize only the threats they have encountered in the past as "threats", they will miss other threats and dramatically reduce the survival probability.

Rather than ignore things as "This is not a threat", it is better to attack even if the judgment is wrong, because the robot has a higher probability of survival, so it is better to attack even if the probability of "threat" is low Will be. This "boundary that humans and robots consider as a threat" is defined as "paranoia line".

In the figure below, the vertical axis is the “frequency of actual threats” and the horizontal axis is the “probability of threats”, and the paranoia line is located at “91% is safe but 9% is dangerous” It has been pulled. The left side of the paranoia line is "attack," and the right side is "calm."

When the paranoia line is drawn with the "91% safe" line, the probability of dying due to an accidental attack is 4.1% even if an attack is made, and the death rate due to ignoring the threat is 0.4% , The overall survival rate will be 95.4%. It is no wonder that people regard various things as threats and have anxiety in order to have survival if the survival rate increases by considering things as threats.

Robots are less likely to be a “threat” that would normally burn a robot, but the risk increases as stock prices fluctuate and workers are fired. In this case, the robot should raise a higher level of concern than usual.

In the figure below, even if the paranoia line is at the same position, there are similar levels of safety and threat (left), when it is safer (center), and when it is more dangerous (right). ) Shows that the actual attack rate will change.

One needs to set up a "zone of concern" that the robot deserves to be concerned about. What can be dangerous to a robot can only be judged from the experience of the robot, but people "evaluate how dangerous the stimulus is compared to past experience" "attack anything in the concern zone" It is possible to optimize the concern zone by repeating adjustments such as expanding or shrinking the concern zone if there is damage to yourself due to the attack.

When Mr. Goldstein simulated the robot to attack against the stimulus received in the concerned zone, when the attack did not change the robot's survival rate, there was no change in the concerned zone, and a safe one was mistaken. It has been shown that the area of ​​concern shrinks when attacking and expands when the threat is ignored.

Contagion of Concern: Uniform Threat Distribution-YouTube

(embed)https://www.youtube.com/watch?v=2Tft3k9y_jc(/embed)

In addition, a big profit was obtained, and there was a tendency in the concerned zone that "expansion is fast and contraction is slow". This is in common with the tendency of people to “do their initial action quickly and dramatically,” Goldstein said, which is why people are improving their survival rates.

Robots need to pay attention to detect threats, and there are three possible types of inputs: "direct experience of robots," "scenarios that robots envision," and "information obtained from other robots."

Goldstein says that although humans are not robots, some understanding can be gained given that the brain is executing the algorithms described above.

In other words, unlike a safe world, in a world with a lot of threats, "direct experience" is more risky to pay attention, so "scenario assumption" becomes more important. For this reason, people tend to fall into a thought loop when there are many threats.

In addition, "obtaining information from others" tends to be emphasized in situations where direct experience is not available. It is easy to make a judgment that "when a person is running away from a fire, I should also run away", and the causes and effects loop, including myselfRecursionWill be seen with high reflectance.

Also, people in a safe environment have more chances of survival if they are not concerned, while people in a threatened environment are at increased risk of survival. In the figure below, the upper row sets the concern zone at 83% and the lower row sets the concern zone at 99%. The left column shows "safe environment" and the right column shows "dangerous environment". You can see that in a safe environment, the smaller the zone of concern, the higher the survival rate, and in the dangerous environment, the larger the zone of concern, the higher the survival rate.


Trauma of Holocaust survivors is inherited in childrenHowever, this is also thought to be a reaction to increase the survival probability.

And in the case of algorithms, zones of concern change with experience like being "damaged" or "surviving intact". Humans can also learn how to ease their anxiety from here, says Goldstein. Specifically, we propose the following contents.

1: Create a positive environment

If you are already in a stressful environment, it is recommended to create a positive one. When Mr Goldstein felt overwhelmed by his work, he put a 30 minute break into his schedule. He also said he was spending time investing in a non-profit organization that didn't feel good, and spending time learning and exercising.

2: Change how to use the Internet

Social media is especially flooded with negative information. Goldstein recommends staying away from negative social media while continuing to use the Internet to get ideas and learn from certain online communities.

3: Find a supportive person

For Goldstein, who couldn't change his stressful environment while running Hipmunk, the most “positive thing” was to work with people who supported him. It is possible to adjust the zone of concern by finding a collaborative person such as a friend, lover, or therapist.

4: Make more positive inputs

It is thought that "self-compassion" that cares about oneself is an important element that supports success, but in order to realize self-compassion, it is important to understand and encourage oneself on a daily basis. ..

5: Accept anxiety

Since it is natural to feel anxiety with feelings of fear and contempt, Mr. Goldstein seems to be taking a deep breath when he feels anxious and making the area of ​​concern smaller by feeling grateful.

6: Control your imagination

Many of the “negative inputs” that people get are imaginative. But when, for example, a dreadful video begins playing in your head, you can switch to another positive thought. This is called "self-thinking control," which may not work from the start, but training, like meditation, can improve skills, says Goldstein.

Copy the title and URL of this article

Source link

Do you like this article??

Show More

Leave a Reply

Your email address will not be published. Required fields are marked *

Back to top button