Summary of the Key Points
This article discusses the phenomenon of AI replacing employees, starting with the popularity of a tool called “ColleagueSkill,” which reveals how companies are using technology to extract employees’ work skills to train AI. Companies like Meta and Alibaba have already begun implementing this practice. The process of AI replacing employees can be divided into three steps: breaking down tasks, centralizing supervision, and reducing wages. Workers are also fighting back through initiatives such as “anti-distillation projects” and legal actions. Finally, the article advises professionals not to simply focus on preventing substitution but to learn how to integrate with AI and become irreplaceable core talents.
I. What is “Colleague Distillation,” and Why Are Workers Panicking?
“Colleague Distillation” essentially involves extracting employees’ skills, habits, and knowledge to train AI systems that can replace them.
- The Popular Tool “ColleagueSkill”: An engineer in Shanghai created a tool that takes 4 hours to organize a colleague’s chat records and document snippets into three small files (totaling just a few dozen KB). These files enable the AI to mimic the colleague’s work, such as creating PPTs or spreadsheets. Although the amount of data is not sufficient for complete substitution, it taps into workers’ fears—that someone who used to slack off yesterday could now become an AI.
- Real-World Applications of Distillation: Meta’s internal project, MCI, goes a step further by installing software on employees’ computers to collect all work-related data (mouse movements, keyboard input, screen shots, etc.) and using it to train AI. The goal is for the AI to handle most of the work, with humans only responsible for directing and reviewing it. Alibaba’s “Second Insight” tool has also helped over 10,000 non-technical employees use AI to complete tasks like creating H5 pages. Technically, it is now feasible to “distill” employees’ skills.
II. Does AI Really Replace Employees in Just Three Steps?
The article argues that this process is not something far-fetched but a concrete set of steps:
1. Task Decomposition: Break down your job into multiple tasks. For example, if Financial Worker Wang’s job consists of 8 tasks, 6 of them can be transformed into AI skills for the AI to handle, leaving only 2 for human review.
2. Centralized Supervision: A small number of employees are then responsible for reviewing the AI’s output, generating more data that further accelerates the replacement process (the supervisors themselves contribute to this process).
3. Wage Reduction: Companies like Meta have eliminated 14,000 positions and invested the saved funds in AI infrastructure, essentially using AI to reduce labor costs.
III. Workers’ Countermeasures: “Anti-Distillation” and Legal Action
Workers are not passive:
- Anti-Distillation Projects: Some people have created tools on GitHub that produce documents that appear professional but lack essential knowledge (making it difficult for AI to learn effectively) or introduce intentional errors into the data (to prevent accurate imitation). There are already more than 50 such projects.
- Legal Action: In a case in Hangzhou, a company used an AI tool to reassign tasks and reduce wages. When the employee refused, they were fired, and the court ruled in favor of the employee, ordering the company to pay compensation of 260,000 yuan. This shows that companies cannot use the introduction of AI as a pretext to exploit workers.
IV. Instead of Preventing Replacement, Focus on Becoming Friends with AI
The article suggests that instead of focusing on preventing AI substitution, people should focus on areas where AI cannot excel:
- Tasks that require professional judgment (e.g., diagnosing complex medical conditions), maintaining customer relationships, innovating in business strategies, and managing complex projects.
- The right approach is to use AI as a tool to improve efficiency (e.g., using it to draft initial documents and then refining them manually) while strengthening your core competencies (e.g., industry insights and cross-team collaboration). In the end, only “tool-based” jobs are at risk of being replaced; those who know how to work with AI will be more valuable.
This article emphasizes that the wave of AI is unstoppable, but professionals can choose to become the ones who steer it, rather than being replaced by it.