Summary of Key Points
The main argument of this article is that the integration of AI into product managers' workflows is inevitable. The question is no longer whether to adopt it, but how to adapt to it. In the AI era, the role of product managers shifts from creating fixed rules to managing probabilistic risks. The pace of AI adoption varies across different fields (fast in customer service, slow in core transactions). AI will also change the way organizations collaborate, requiring cross-departmental data sharing. Individuals can start by practicing in non-core areas, building project expertise, and understanding the associated risks, with the ultimate goal of clearly assessing the underlying uncertainties.
Product Managers in the AI Era: From Following Recipes to Choosing Routes
Traditional product managers are like chefs who prepare meals according to fixed recipes, focusing on execution details to ensure quick results. AI-powered product managers, on the other hand, are like drivers who need to consider road conditions (data/model feedback) and choose the best route to achieve their goals while minimizing costs (efficiency). The biggest difference is that decisions are now based on probabilities rather than binary choices. For example, instead of simply directing users to a payment page when they click a button, product managers must weigh the risks associated with different options.
The essence remains the same: the value of decision-making. While in the past, product managers discussed feature implementation with engineers, they now collaborate with algorithm engineers to understand the probabilities behind these decisions, all with the aim of making the right choices.
Speed of AI Adoption Across Fields
AI is not being implemented uniformly across all areas. The key factor is the balance between risk and reward:
- Fast adoption: Areas with clear rules, low risks, and significant benefits from automation (e.g., customer service, where AI can easily learn from predefined knowledge bases without causing significant errors) are more likely to adopt AI quickly.
- Slow adoption: Core transaction processes (e.g., payment, ordering) face greater challenges because mistakes by AI could lead to significant losses. In these cases, caution outweighs innovation, resulting in a lower demand for advanced AI technologies.
The Need for Cross-Departmental Collaboration
AI relies on large datasets and models, necessitating data sharing across departments. For example, in the author's project, while one department had the data but lacked business understanding, another department had complaints but lacked the resources to address them. Only by collaborating could the project succeed. However, traditional organizations often resist such collaboration, leading to delays.
Personal Adaptation to AI
To prepare for AI, follow these steps:
1. Start as a “passenger”: If you’re hesitant to use AI in core processes, start with non-core tasks, such as developing AI-powered customer service assistants or operational tools.
2. Build your expertise: Even if projects don’t launch immediately, document how problems are defined, models are evaluated, and monitoring indicators are set up. These skills will be valuable when opportunities arise.
3. Clarify the risks: When presenting your ideas to management, focus on the benefits (e.g., 95% efficiency improvement) and potential risks (e.g., 5% error rate with X-level losses), and show how you can mitigate these risks.
In conclusion, AI doesn’t require product managers to change careers; it provides new tools. Algorithms handle the calculation of probabilities, while product managers are responsible for assessing risks. Those who can clearly understand and manage these risks will be better positioned for success. Product value ultimately comes from how technology balances benefits and risks.