Woter AI detection.Hurry - ends Jun 29th

New Year Sales :up to 80% OFF

AI Humanize AI Translator Bypass AI AI Rewriter AI Detector

PRICING

TRY FOR FREE

Missing this framework, 80% of bosses are wasting AI investment

Written by

Silas Grey

Updated on:June-13th-2025

As a business leader, do you have the following concerns?

On the one hand, we deeply feel that AI is a huge opportunity for change, and enterprises cannot fall behind

At the same time, when trying to promote it within the company, you don’t know where to start?

This article will help you sort out the clues, establish a basic framework, and figure out what AI to use in what scenario.

(This article will not explain how to use specific tools, nor will it provide technical training, but only talk about the thinking framework)

How to use AI well is essentially a communication problem

AI is a tool, but it is different from any previous tool. If used properly, AI will demonstrate unique autonomy and intelligence.

The so-called "good use" is essentially a communication and even cognitive problem. Let's draw a quadrant diagram (Note 1):

When using AI, we can get four communication contexts depending on the degree of understanding of the content (data, information, knowledge, experience) involved in the conversation between the two parties:

Common Knowledge Zone: People know and AI also knows
Human Blind Spots: People Don’t Know, But AI Knows
Double Blind Spot: Neither Humans nor AI Know
AI blind spots: People know but AI doesn’t

When you are in different contexts, of course you have to follow different communication principles and methods.

1. Common Knowledge Area

The first quadrant is characterized by the ability for humans and AI to share clear and transparent information.

So in the enterprise scenario, what information can both AI and humans obtain? Nothing more than:

External: general, publicly searchable knowledge information
Internal: Documents and corporate data (finance, operations, marketing, etc.)

Obviously, for content that both parties know, "testing" whether the AI knows it can only be regarded as a curious act and cannot generate additional benefits.

This area is most suitable for "lazy" applications, that is, using AI as an assistant to help you improve efficiency.

It is a certain job that can be done by humans, but it is repetitive, tedious, consumes a lot of time and energy, and has low added value.

What is needed is a hard-working and tireless assistant, and this is exactly what AI with powerful computing power is good at.

Several typical scenarios:

• Batch document review and screening, such as contracts, resumes, etc.

• Fast translation of long documents

• A lot of materials need to be read, sorted and reported

• Database data query (assuming AI can access the database interface) and report generation

There are two key points to making good use of AI in this area:

Data quality is still garbage-in-garbage-out for large AI models, so the data provided to AI must be comprehensive, accurate, and updated in a timely manner;
Data acquisition refers to whether the enterprise has properly "packaged" internal information (for example, whether the database interface is open, whether the internal service has a similar MCP protocol encapsulation). In short, AI should be able to access it easily.

Keywords in the knowledge zone: assistant, efficiency improvement

2. Human Blind Spot

The characteristics of the second quadrant are: the data or capabilities mastered by AI are beyond the scope of human cognition.

When faced with a powerful AI that knows everything, we should of course use it as a teacher, and it should be available at any time and teach tirelessly.

For example, if you are a layman and want to invest in photovoltaic power stations, then let DeepSeek quickly describe the industry for you based on the PEST framework and provide the latest development trends, etc., which will be much faster than asking someone or checking it yourself.

For example, if you don’t know programming, but have a good idea to make an app, you don’t have to spend money to build a team right away. You can first use AI programming tools such as cursor to quickly realize the product prototype and verify user feedback.

This area is also suitable for using AI as a "prosecutor". AI can be meticulous and impartial in finding human omissions or areas that humans are not aware of, avoiding subjective bias, such as:

Use Doubao to conditionally polish the article manuscript and adjust and modify the parts that are not smooth
Send the solution decision to AI, let it review it and raise questions and suggestions for improvement
Build a workflow to identify conflict points and omissions from a large number of work orders

Of course, there is a big problem with using AI in this quadrant: Is the AI’s answer credible?

We all know that AI has illusions and uncertainties, so how to use engineering methods to make AI's answers as correct and verifiable as possible is the key to the practical application of this quadrant.

Keywords in people's blind spots: teacher, prosecutor

3. Double blind area

The third quadrant is characterized by capabilities that neither humans nor AI know and have yet to discover.

Compared with the first two quadrants, this area is more suitable for applications such as creation, prediction, or scientific research.

For entertainment and creative enterprises, they can make full use of AI's "illusion", such as giving Deepseek a title and letting it use its imagination and creativity to write science fiction novels; designing prompt words, pictures or videos (virtual worlds that do not exist)

In addition to creation, you can also use the power of reasoning models to provide massive amounts of data so that large models can find problems in the data and predict potential risks, such as:

Predicting potential failures based on equipment operation data
Predicting potential PR crises based on social media public opinion
Predicting extreme weather

In addition to large language models, many prediction scenarios also require machine/deep learning models.

There will be relatively more applications in scientific research and exploration, such as Alphafold predicting unknown protein structures and NASA using AI to discover new galaxies.

In the double blind area, enterprises can make full use of AI's exploration capabilities to stimulate people's creativity and expand their thinking, such as:

Simulate and verify various combinations to quickly develop new materials and processes
Use AI brainstorming to achieve combined and cross-border innovation. Since AI is not constrained by human experience, it has a broad imagination and can sometimes come up with unexpected innovative ideas.

The key point of applying AI in double-blind spots is: don’t let human limitations limit the performance of AI. In addition, high-quality prompt words are also the key to stimulating model capabilities.

Keywords for Double Blind Zone: creation, prediction, exploration

4. AI Blind Spots

The characteristics of the fourth quadrant are: AI cannot directly obtain the data and information here. For example, many corporate private domain knowledge, industry vertical know-how, and general large models are unknown. If you directly ask the AI model about private domain knowledge, you will likely get a lot of serious nonsense.

In AI blind spots, you cannot communicate with it effectively. In contrast to human blind spots, here we humans are AI’s teachers. We should treat it like a child and find ways to “feed” it with data.

In connection with the common knowledge quadrant mentioned earlier, if the internal data of an enterprise is "unclean" or difficult for AI to access directly, then what you think is common knowledge is actually a blind spot for AI.

At this time, we need to clean the data, use the RAG knowledge base or fine-tune the large model to open private domain knowledge to AI and transform AI blind spots into common knowledge areas.

AI blind spot keywords: knowledge base, fine-tuning ➡️Narrow the blind spot and expand the common knowledge area.

Summarize

The key to making good use of AI in enterprises is communication. Only by using the right AI capabilities in the right scenarios can corresponding value be gained.

Finally, the key points:

1️⃣ Shared Knowledge Area (Assistant Efficiency Improvement): Handle repetitive work and ensure data quality and access.

2️⃣ Human blind spots (teachers check for omissions): AI assists in decision making, beware of the risk of illusions.

3️⃣ Double Blind Zone (Creative Exploration): Stimulate innovative predictions and optimize prompt words.

4️⃣ AI blind spot (knowledge conversion): clean data or fine-tune models to turn private domains into shared ones.

I hope that more and more companies can make good use of AI. If you have any questions about the practical application of large AI models, you are welcome to contact me for communication and discussion.

Note 1: The four quadrants mentioned in the article are quoted from the Johari Window, a theoretical model of communication and self-cognition proposed by American psychologists Joseph Luft and Harry Ingham in the 1950s. It helps people better understand their interactions with others and improve communication effectiveness by dividing interpersonal communication information into four areas.

The Four Areas of the Johari Window

1. Open Area

Definition: Information that you know and that others also know.
Features: This is the most direct and transparent area of communication, such as your name, occupation, some experience, hobbies, etc.
Effect: The larger the open area, the smoother the communication and the easier it is to build trust.
Example: In the workplace, colleagues will gradually get to know each other's basic information through self-introduction or daily communication.

2. Hidden Area

Definition: Information that you know but others do not.
Characteristics: This is the part of an individual that is kept secret or not shared. For example, personal feelings, secrets, privacy, certain thoughts, etc.
Effect: It is normal to keep a private area moderately, but excessive hiding may lead to misunderstandings or alienation from others.
Example: A person may conceal dissatisfaction with a job until the stress builds to a point where it becomes unbearable.

3. Blind Area

Definition: Information that others know but you do not.
Characteristics: These may be blind spots discovered through feedback from others, such as character flaws, habitual behaviors, or how others perceive you.
Effect: The larger the blind spot, the easier it is to fall into "self-perception bias". This area can be narrowed by accepting feedback from others.
Example: A leader may view himself as approachable, but his subordinates may view him as too strict.

4. Unknown Area

Definition: Information that is unknown to oneself and others.
Characteristics: This is untapped potential or undiscovered abilities. For example, a latent talent, an unknown emotional response, or a hidden illness.
Function: Through exploration and experience, the unknown area may be transformed into an open area.
Example: A person may never realize that they are good at public speaking until they unexpectedly perform well at an event.