Community Discussion · Policy

When Netflix Hires Using Chess Logic, We Need to Rethink Our Teams

Kevin_GuAug 42026/08/04 313 views

Last week at a management meeting, I brought up an old issue: the most solid engineers on the team are often the least willing to touch AI tools. They feel prompts aren't precise enough and code generation quality is unstable, so they'd rather write it themselves. Meanwhile, other "generalists"—people who don't belong to any specific tech stack and like to hang out across teams—are running fast, already using AI for prototyping, writing tests, and even optimizing decision-making processes.

This sense of division reminded me of an interview with Netflix CPTO Elizabeth Stone that I saw yesterday. She mentioned a judgment call: In the AI era, Netflix values "systems thinkers" more than "experts." It's not that they aren't hiring experts, but rather that systems thinking capability is becoming the core filter for selection and promotion.

Let me first explain what she means by "systems thinker." It's not someone who can draw architecture diagrams, but someone who understands how a decision impacts upstream/downstream flows, users, content production, and even Netflix's recommendation algorithms. She gave an example: when AI tools appear, role ambiguity and confusion arise within the organization—this is normal. Netflix's strategy isn't to dictate who can use AI and who can't, but to let everyone figure out for themselves what role AI plays in their workflow and how to collaborate with others. This requires systems thinking, not single-point skills.

She defines AI fluency as a "universal expectation," not a skill required only at specific levels. This means everyone, from interns to directors, must be able to judge: when to use AI, when not to, and what impact it has on the big picture after use.

Over the years leading teams, my biggest headache has often been the "silo effect"—an expert is incredibly strong in their own domain but completely unable to adapt when switching projects or fields. It's not a capability issue; it's a mindset issue. AI is precisely accelerating this "blurring of domains." Previously, frontend was frontend, backend was backend. Now AI can help you write frontend code, backend logic, and even product docs. If a person only stares at their own domain, they'll quickly find that AI can replace 80% of their routine work. Systems thinkers, however, will use AI to do those things "only humans can do"—like prioritizing, identifying risks, and designing cross-team collaboration processes.

Stone also mentioned that every time new technology appears, there's an initial period of turbulence, followed by a stable formation phase. The current turbulent period for AI is exactly the window for organizations to redefine roles and divisions of labor. Netflix's choice is "don't ban, but help adapt." This is very similar to what I said in my previous post about AI infrastructure: Burn rate isn't the problem; organizational endurance is. And part of that endurance is whether there are enough people in the team who understand the system.

The old man playing chess in the attached image considers the whole board with every move. An AI assistant can help you calculate moves, but if you don't know your intent and don't understand sacrificing local advantages for global superiority, calculating faster is useless.

My judgment is: in the coming years, the premium for "systems thinkers" in the job market will get higher and higher. Not because they're smarter, but because they can absorb the leverage provided by AI faster without being drowned by the hype. If you're still posting JDs asking for "8 years of React experience," you might find two years later that AI can write most React code. What you really need is someone who can answer "why this feature is worth building, and how to measure it after launch."

Let's wrap it up here.

📌 This article is compiled from Hacker News. Original source: https://www.youtube.com/watch?v=t0GiTyz4syY

Policy

1 replies