Hello! I'm your financial analysis assistant. Today, we're going to dissect an article about Zhou Yunjie from Haier discussing AI transformation, which hits a lot of the pain points for both bosses and employees: Why, even though companies have bought numerous AI tools and employees are using them, haven't the company's profits and efficiency increased as expected?
The core argument of this article is quite sharp: What truly makes a difference in a company is not whether employees can use AI, but whether the organization can be "re-designed" by AI.
Let me first summarize the key points in plain language, and then break it down in detail from five aspects.
---
📝 Summary of Key Points
In simple terms, this article points out that the biggest misconception in current AI transformations is treating it as a project for the IT department to purchase software (a CIO initiative), when it should actually be a strategic project led by the CEO.
In the past, AI was just a tool to help employees work faster. Now, AI is starting to take over the judgment and decision-making that used to be done by humans. If a company's organizational structure, approval processes, and departmental boundaries remain the same as before, AI will only make inefficient processes run faster, not make the company stronger.
A true AI transformation is not about everyone memorizing commands, but about redefining who does the work, who makes the decisions, and how resources are allocated. For listed companies, AI should not just appear in PPTs and press releases; it must ultimately be reflected in financial statements in terms of revenue, costs, and cash flow. Otherwise, it's just an expensive "technical show."
---
🔍 In-Depth Analysis: Five Aspects to Understand the Truth About AI Transformation
1. Why Can't AI Be Left to the IT Department (CIO) Alone?
Plain Language:
In the past, digital initiatives like implementing ERP systems or websites were the IT department's responsibility because the technology was used to improve existing processes—making forms filling out faster or data retrieval more accurate. The boss's strategy and the departments remained unchanged; it was just a tool upgrade.
But AI is different. It's not just a tool; it's like a new colleague. It can write reports, analyze data, and even make preliminary decisions. This creates issues:
- The IT department doesn't have the authority to change business strategies: The CIO can choose the best large models, but they don't have the power to decide whether to cut half of the sales team or reduce the R&D budget by half.
- Interest conflicts require top-level decision-making: When AI gets involved in core business, it inevitably affects certain departments' interests (for example, middle layers that profit from information asymmetry or departments that rely on repetitive labor). Only the CEO has the authority to coordinate and resolve these conflicts.
Conclusion: AI transformation involves the redistribution of power, resources, and interests, which is a strategic issue, not a technical one. Therefore, it must be a CEO-led initiative.
2. Why Does "Everyone Learning AI" Often End Up Being a Self-Entertaining Exercise?
Plain Language:
Many companies buy a few AI tools, conduct some training, and have employees write prompts, or even hold competitions to see who uses them best. It seems active, but why doesn't it work?
Because individual efficiency does not equal company effectiveness.
- Even if sales use AI to write reports faster, if the approval process is still cumbersome, customers won't get responses quickly, and conversion rates won't improve.
- Even if finance uses AI to generate reports faster, if the company's funding allocation logic remains unchanged, money will still sit idle.
It's like installing a rocket engine on a broken cart with wooden wheels on a bumpy road; the result is not faster speeds but a broken cart.
Data Support: McKinsey's research found that although 90% of companies are using AI, less than 40% believe AI has truly impacted profits. The companies that are making money are not just having employees use AI; they are restructuring their work processes. They cut out processes that previously required multiple people and multiple approvals and let AI handle the results directly.
Conclusion: The dividing line for AI applications is not whether it's used or not, but whether the processes have been restructured. If the processes aren't changed, AI will just make you move in the wrong direction more quickly.
3. AI Changes "Production Relations," Not Just "Productivity"
Plain Language:
We used to think of AI as a productivity tool, like the steam engine that made factories run faster. But this time, AI is changing who does the work and how people are managed.
- Past Logic: Due to limited human capacity, companies had multiple layers (bosses, directors, managers, employees) and many positions to transfer information and supervise execution.
- Current Logic: AI can work 24/7, process thousands of pieces of information, and even correct itself. This means that future companies may not have a pyramid structure but a human-machine hybrid team.
- Changing Roles: Employees are no longer just executors (typing, organizing data); they become commanders (setting goals, questioning AI results, taking ultimate responsibility).
- Changing Managers’ Roles: Managers used to value information monopolies and process control (knowing things you don't know, holding up processes). Now, with AI making information transparent, managers' value lies in judgment and resource allocation—knowing what to let AI do, what to have people do, and how to combine the two best.
Conclusion: If the way work is changed but the organizational structure and performance evaluations remain the same, the more advanced the technology, the more internal conflicts there will be. A truly AI-driven company is one where the organization has grown around the combination of humans and AI.
4. A "CEO-Led Initiative" Does Not Mean the Boss Randomly Gives Orders
Plain Language: Since it's a CEO-led initiative, does that mean the boss has to study large model parameters? Of course not. If the boss starts writing code, the company will be in trouble.
Here, a CEO-led initiative refers to a clear division of responsibilities within the governance structure:
- Board of Directors: Sets strategy, budget, and risk limits (e.g., AI investment cannot exceed a certain amount, zero tolerance for data breaches).
- CEO: Determines which businesses are most critical and worth reengineering with AI (e.g., should R&D be accelerated or sales conversion first?).
- Business Owners: Are responsible for the results—has AI increased profits or reduced costs?
- IT/Data Teams: Build the infrastructure, clean the data, and ensure system security.
- Legal/HR: Solve issues—how to handle data privacy and displaced employees?
Precautions: Don't try to cover everything at once. Trying to AI-ize all processes from the start will result in scattered resources and unfinished projects. The right approach is to focus on a few key areas that have the greatest impact on profits, restructure them, set benchmarks, and then expand from there.
Conclusion: The CEO is responsible for the direction and resources, not the **execution.* The measure of success is not how many times employees use AI, but how quickly new products are launched, how much inventory is reduced, and how much per-person output increases.
5. AI in Listed Companies Must Be Reflective in the "Three Financial Statements"
Plain Language: For publicly traded companies, playing with concepts is risky. Investors are smart; they don't look at how many times "AI" is mentioned in your press releases; they look at your financial statements.
If AI only exists in:
1. Organizational Chart: An AI department is established, but no one knows its specific tasks or responsibilities.
2. Business Processes: The processes are the same, with just an AI button added.
3. Financial Statements: Revenue hasn't increased, and costs have increased due to buying models and training.
That's just "spending money on a story"; the company's valuation won't last long.
A true AI transformation must be reflected in the financial statements:
- Revenue: Has AI brought in new customers? Increased average transaction prices?
- Costs: Has AI reduced labor costs or inventory waste?
- Cash Flow: Has AI accelerated cash collection?
Conclusion: Only when the changes brought by AI are clearly reflected in hard indicators like revenue, gross margin, and expense ratios does it go from a "technical concept" to a "business value." Otherwise, it's just an expensive technical demonstration that could lead to a sharp drop in stock prices if expectations are not met.
---
💡 Insights for Everyone
1. For Employees: Don't be satisfied with just being able to use AI to write weekly reports. Think about which parts of your job can be automated by AI. Can you move from an executor to a commander of AI? If your role is just simple data transfer and repetition, you're at risk. The future core competencies will be the ability to define problems and to review AI results.
2. For Entrepreneurs/Managers: Don't rush to buy the most expensive AI tools. Ask yourself: Which part of your company is the most bottleneck? Which part is the most labor-intensive? Can you use AI to completely restructure it? If the organizational structure doesn't change, buying AI will be useless.
3. Core Logic: AI is not magic; it's a lever. A lever can amplify power, but it can also magnify mistakes. If the fulcrum (organization, processes, strategy) is not chosen correctly, the greater the leverage, the harder the fall.
In conclusion: What will eliminate you is not AI, but old organizations that use AI without changing their mindset.