Why Are Companies That Say "Challenge and Reform" Not Using AI?

There is a common phrase heard in companies.

"Let's challenge ourselves."
"Let's reform."
"Let's take initiative."
"Let's improve."
"We need to change."

On the surface, these words sound very positive.

However, when you work on the front lines, you might feel this way.

If they really want to change, why don't they use the tools available?

Especially for those who are using generative AI or LLMs, the word "reform" from the company can seem quite hollow.

If You Say "Reform," You First Need to Update Your Tools

Reform is not just about motivation.

Reform means changing the way work is done, the tools used, the flow of information, the decision-making process, and the scope of responsibility.

For example, with LLMs today, you can quickly create drafts of the following tasks:

• Meeting minutes
• Work standard documents
• Training materials
• FAQs
• Checklists
• Email texts
• PowerPoint structures
• Excel table designs
• Business processes
• Questions new employees should ask
• Potential omissions in explanations
• Interview scripts
• First drafts of reports

Creating these from scratch would normally take several hours to days.

However, by giving voice notes, memos, or existing documents to an LLM, you can create a first draft in a very short time.

Of course, a human needs to do the final check.

However, what humans should do is fact-checking, decision-making, approval, and adjustments according to the site.

There is no need for humans to spend time on rewriting phrases or formatting documents.

"Information Security is a Concern" Is Understandable, But Stopping There Is Hasty

A common reason for opposing AI implementation is information security.

"What if confidential information leaks?"
"What if personal information is entered?"
"Who will take responsibility?"
"What if it gives the wrong answer?"

These concerns are not something that can be ignored.

In fact, they should be taken seriously.

However, stopping here is wrong.

Concerns about information security are not reasons to not use AI, but reasons to design a safe way to use it.

For example, there are the following ways to use it:

• Do not enter confidential information into cloud AI
• Use a local LLM
• Use it only within the company's LAN
• Create an environment where it is not sent externally
• Separate information that can be input and information that is prohibited
• Anonymize personal or customer information
• Start with low-risk tasks
• Always have a human check the output
• Define usage logs and areas of responsibility

There is room for design in this way.

"Security is scary, so ban everything" is too simplistic.

The option of placing an LLM in a shared local network

If you are concerned about sending information to the cloud, another option is to set up an LLM environment within your company's shared local network.

For example, by preparing an LLM that runs within the internal LAN and ensuring that no data is sent externally, you can significantly reduce the concern of "being afraid to throw confidential information into the cloud."

Of course, there are also challenges with a local LLM.

• Model performance
• Operation and management
• Access control from internal terminals
• Handling of input logs
• Who will perform maintenance
• Scope of responsibility in case of incorrect answers
• Cost
• Internal training

Nevertheless, it is more constructive to think about "what range is usable" rather than stopping at "it's impossible."

At the very least, you can start with low-risk areas such as formatting meeting minutes, drafting initial versions of standard documents, draft FAQs, initial versions of educational materials, consultations on Excel functions, and draft emails.

"Trying to make do with manpower" is not reform

The problem is not just that the company does not use AI.

A bigger problem is telling the front line to "change" without updating the tools.

For example, the following situation may occur:

• The amount of work increases
• You are told to create standard documents
• You are told to conduct training
• You are told to eliminate individualization
• You are told to reduce overtime
• But you are not allowed to use AI or automation
• No rules are created
• No authority is given
• In the end, you are told to "try to make do with manpower"

This is not reform.

It's just a matter of willpower.

Even if the words are "challenge," "reform," and "improvement," if the content still relies on human power, the reality remains old-fashioned.

Spending too much human time on detailed wording of documents

In many companies, a lot of time is spent on the detailed wording and formatting of documents.

"This expression is slightly different" "More understandable" "Fix this" "Overall different" "The flow of the document is a bit off" "The language is too stiff" "More polite"

Of course, the quality of the document is important.

However, many of these revisions can be significantly shortened by asking the LLM to provide multiple options.

For example, the same content can be presented in the following ways:

• For executives
• For the front line
• For new employees
• Concise version
• Polite version
• Bullet point version
• For PowerPoint
• For email
• For standard documents
• For meeting explanations

There is no need for humans to write from scratch each time.

Humans just need to look at the options provided by the LLM and determine whether they match the on-site facts and whether they are content that the company can release.

True Reform Is Not About "Trying Harder" But About "Removing"

When we think of reform, we often imagine adding new things.

However, what is truly needed by the front line is not adding, but removing.

• Remove manual work
• Remove time spent explaining the same thing repeatedly
• Remove reliance on memory after meetings
• Remove time spent rewriting documents
• Remove time when new employees don't know what to ask
• Remove personalized explanations
• Remove back-and-forth due to ambiguous instructions
• Remove tasks that humans don't need to do

For this reason, AI is a very suitable tool.

In particular, text creation, summarization, structuring, checklist creation, document creation, and FAQ creation are areas where LLMs excel.

In other words, the essence of introducing AI is not "using it because it's popular."

It is to use it to reduce working hours.

How to View Companies That Don't Use AI as a Job Option

You don't need to avoid all companies that don't use AI.

Depending on the industry, size, and security requirements, it may be delayed in adoption.

The problem is not that they don't use it, but that they haven't analyzed the reasons for not using it.

When looking at potential employers, the following perspectives can be helpful.

Positive Responses

• Not yet fully implemented company-wide, but trying it for meeting minutes and document creation
• Currently establishing security rules
• Considering how to use it by department
• Starting with low-risk tasks
• Welcoming people who can use it for business improvement
• Viewing generative AI as a managed item, not a banned one
• Interested in reducing the workload of the front line

Dangerous Responses

• Don't know much about AI, so not using it
• It's dangerous, so banning everything
• Doing it manually shows more sincerity
• We have always done it this way, so we won't change
• But they say to reform and to challenge
• Increasing workload but not increasing tools
• Believing that improvement is the responsibility of individuals
• Viewing people who can increase efficiency as handy people

Companies in the latter category may be more likely to shift the cost of change onto the front line, more than just not using AI.

"Challenge Slogans" and "Actual Change" Are Different

A company's slogan of "challenge" is easy.

It can be written on a poster.

It can be said in morning meetings.

It can be said in policy announcements.

However, real challenge is changing the way work is done at the front line.

And this involves some pain.

• Need to create rules
• Need to define responsibility areas
• Need to review previous ways of working
• Managers also need to learn
• Need to accept failure
• Need to give authority to the front line
• Need to invest in tools

Saying "let's challenge" while avoiding these is not reform.

It is a push of idealism onto the front line.

Summary

It is natural to feel uncomfortable with companies that talk about "challenges," "reforms," and "improvements" but do not use AI.

Not using AI is not necessarily bad.

However, companies that do not even design how to use it, citing reasons such as information security and responsibility issues, are dangerous.

If you really want to reform, the following are necessary:

• Acquire new information
• Try tools that can be used
• Break down risks
• Start with low-risk tasks
• Consider local LLMs and internal environments
• Divide tasks between what humans do and what AI does
• Reduce the working hours of the front line
• Mechanize rather than become a convenience service

Reform is not about telling the front line "try harder."

It is about creating a system that works even without human effort.

In that sense, companies that say "reform with manpower without using AI" are quite off track.

What is really needed is not just a challenge slogan, but the renewal of tools, rules, authority, and responsibility.

---

Related Article Topics

• Do companies that do not use AI have a future?
• Characteristics of companies that talk about "reform" but impose a "grind mentality" on the front line
• List of administrative tasks that can be reduced with generative AI
• Can local LLMs be a cybersecurity measure for small and medium-sized enterprises?
• Reasons why AI introduction can be a basis for choosing a new job
• Why the company's "Let's challenge" sounds empty
• The limits of manpower improvement and the reality of AI utilization
• How much can standard documents, meeting minutes, and training materials be created with LLMs?