Mogic considers

What comes next after the AI interview is

As I was talking to an intern on my way home from a hangout, she said, "These days, all the interviews are AI interviews.

"Oh, all of them?"
Yes, AI makes me lose my temper.
Yes, it's hard to keep in time with it.
And they give me feedback at the end.
I'm getting that kind of advice in an interview.
"You need to dig a little deeper, or be more logical."
Wow, that's tough.

Large companies may receive hundreds or even thousands of applications, so they may want to narrow down the number of applicants who proceed to the interview process, given the number of people in the human resources department.

This change makes it difficult to comment on the spot when a person is interviewing, but with AI, you can even give quick advice to the applicant.

Of course, the applicants are sophisticated and probably have practiced mock AI interviews with AI counterparts in advance.

Now, I wonder how I would handle this AI interview thing if I were you.

It's a bit sneaky, but eventually they will use AI to plan their strategy.

The AI interview service is then asked to list the AI interview services that large companies are likely to use, and at the same time, compare their common specifications and differences, identify the range of settings that HR can make, determine which items are common to all companies and which items determine whether the AI interview is a match for a particular company, and determine what the AI itself looks for in the way it asks questions and what it looks for in terms of word structure, sentence structure, volume, tone, and facial expression (3 minutes so far). The AI itself will ask the questions in what form, and check what it is looking for in terms of word and sentence structure, voice volume and tone, and facial expressions (3 minutes have elapsed so far).

If the number of applicants is 300 to 100, and the ratio is expected to be 3 times, the average score may cause the applicant to be dropped, so the applicant should take the average as the base score and earn points only for the items in the top 1/3, and slip through (5 minutes have passed so far).

Only then do some AI mock interviews and score them. After that, we adjust the initial hypothesis and service specifications for any deviations, and let the AI mock interview become a part of our body of work. We will not do this for more than an hour, because if we do too much, we will over-optimize and be counterproductive. (Spend 60 minutes here).

And after thinking about it briefly, I have a sense that it's kind of a cycle.

For AI interviews, use AI to understand how the AI interview service works, fine-tune it with an AI mock interviewer, and get advice during the real AI interview.

Not only does the process of interviewing change, but it seems to imply something else.

If it is not just a "human to AI" interview, what does that mean?

Maybe it's a preview of the same operations in large companies by five years from now.

In other words, they work with AI against the goals set by AI, are evaluated by AI, and are trained by AI where they fall short.

If so, you may want to start now to anticipate the all-encompassing AI environment and how it should be there.

The combination should not explode.

Graph problems found in information engineering are often applied to organizations.

An organization is a group of people, and 10 people are more informative than 2.

As the number of people increases, the number of connections automatically increases, and as connections are made, information is exchanged, so the overall volume of information increases rapidly.

This is probably one of the reasons why some people seem to have a different temperature when there are many people in the room.

So, what kind of increase in the amount of information is there depending on the number of people?

The number is called the combinatorial explosion of the network, so let me quote a few words from that.

oooooooooooooooooooooooooooooooooooooooooooooo
Combination Explosion
h ttps:// w.wiki/HSNh

In a computer network, the number of communication paths required increases rapidly as nodes are added. This is called a combinatorial explosion. In reality, however, the increase is not exponential, but strictly polynomial at best.

When two nodes need to communicate with each other, they can be directly connected by one appropriate one-to-one communication path. However, when a third node is added, three communication paths are needed: six for four nodes, ten for five nodes, fifteen for six nodes, and so on. Generalizing this, the number of communication paths l at n nodes increases on the order of the square of n, as follows

l = n(n-1)/2
l = n(n-1)/2

If we calculate the amount of information in terms of the number of people in the organization, l

l = 1 for 2 people
5 people, l = 10
10 people, l = 45
25 people, l = 300
50 people l = 1,225
100 for l = 4,950

The curve starts sharply at roughly the 15th person.

Therefore, when the number of employees exceeds 10, a method called "organization" is introduced.

We have organized the workforce by position, divided it into departments, defined job descriptions, created standardized flows, developed work rules, held company-wide presentations, and so on.

Well, when there is too much information to process, there is more uncertainty, so it is easier to create problems, speed slows down, profit margins drop, retirements increase, and so naturally you have to do something about it, which equals tinkering with the organization.

Perhaps the creation of KPI goals for each department is one mechanism to control this information explosion.

Hmmm...now that I've sorted all this out, I'm wondering what an "information explosion" really is.

Can't we make the information explosion operable instead of treating it as a bad thing like a higher order infinite set?

Just because there is a lot of information, is it okay to keep only the important information and cut out the rest?

Is it sufficient to simply introduce customary "organizational methods" just because the number of people has increased?

And now, going back to the premise of the combinatorial explosion, it is interesting to note that the nodes (people) and nodes are all equivalently connected.

If people were not equally connected and could be graded (concentrated) well, it would slow down the information explosion, and so on.

Once again, I bring you the rest of the quote from earlier.
ーーーーー
One way to reduce this is to put a generic means of mediating information at the center, in which case the number of communication channels need only be equal to the number of nodes n. The drawback in this case, however, is,

1: One-to-one communication may be possible without any specific procedures, such as telephone or teletype, but to support intermediary means, it is necessary to introduce some communication protocols, such as TCP/IP or SMTP, because it is necessary to control the route based on the destination attached to the communication contents.

2: If the central means of mediation fails or underperforms, all communications are affected.

Because of the disadvantages of 2:, actual designs do not always need to reduce the number of communication paths, but often build complex systems with a certain amount of redundancy.
oooooooooooooooooooooooooooooooooooooooooooooo

Yes, you can come up with many ways of doing things with just one way of thinking.

This year is the year of the Noon, so I would like to raise my gaze and think about the nature of activities in which people come together.

We look forward to your continued patronage as we begin the new year with a lighthearted spirit.