Summary of Key Points
This article reveals the common chaos in companies' adoption of AI through the real experiences of four professionals. Issues include leaders having unrealistic expectations about AI capabilities, using bureaucratic KPIs to evaluate employees' use of AI, abandoning mature tools in pursuit of artificial AI metrics, and frequently changing leadership decisions, which renders AI initiatives ineffective. As a result, employees are overwhelmed with both AI-related tasks and meeting the demands of their bosses, leading to decreased work efficiency and a cycle of exhaustion and inefficiency.
Detailed Explanation
1. Leaders see AI as a “miracle tool” when it’s actually like a “baby that needs feeding”
Many bosses have a simplistic view of AI: they either see it as a tool for laziness (initially) or as something that can do everything (later on). For example, Zhang Ting, a content operator,’s boss initially strictly prohibited the use of AI in content creation but became obsessed with using AI after analyzing competitor materials and demanded that the team rely on it for everything. However, AI cannot work on its own—developers need to provide it with a list of top-tier institutions and reliable information sources (otherwise, it won’t know what “top-tier” means). Similarly, to analyze traffic, the data must be properly organized first. But bosses often ask, “Can’t AI do even that?” without realizing that AI requires data input and training. As a result, the team deviates from their core business goals, leading to poor efficiency and widespread dissatisfaction.
2. Employees are forced to “fabricate” and misuse AI to meet KPIs
Many companies include “AI usage” in their OKR (Objectives and Key Results) metrics, creating absurd evaluation criteria such as weekly AI usage rates and time savings. Product manager Vivian points out that these metrics are unquantifiable: for instance, it’s hard to determine how much time would be saved without using AI to edit documents or analyze user interviews. Employees often have to make up numbers or use AI in unnecessary contexts (like writing weekly reports). Ironically, the end result is a mere formality, with reports about how AI saves time being nothing more than empty promises.
3. Abandoning existing tools for AI that’s not necessary
Some bosses try to appear “AI-driven” by forcing employees to use more complex solutions. For instance, financial journalist Liang Bo used Feedly (an RSS tool) to gather news from external sources but was asked to create an AI-based content selection system. After a week of struggling, he found that AI couldn’t access many websites and exhausted his data quotas; only by copying blogger scripts did he meet the requirements, even though Feedly could have done the same task more efficiently (by directly subscribing to websites and receiving updates). Another example is a company that ignored a built-in “competitor tracking” feature in its project management tool and spent half a month developing a custom AI system, only to realize it was unnecessary.
4. Leadership’s fickleness makes AI initiatives short-lived
AI applications require stable processes, but many bosses’ decisions are based on short-term data fluctuations. For example, the boss of online store operator Xiao Ka praised his AI-driven competitor analysis process last week but immediately dismissed it when sales declined: “What’s the use of this analysis?” As a result, the custom AI system was abandoned, and all the effort was in vain—bosses only care about short-term sales figures. The constant changes in leadership make AI initiatives unsustainable, turning them into one-time uses.
The root of these problems lies in companies’ lack of a rational understanding of AI: it is a tool, not a panacea. It must be integrated with business contexts and employee expertise to truly improve efficiency. Focusing solely on artificial AI metrics leads to internal strife and undermines the purpose of using AI in the first place.