EA handing 85% of QA to AI is a high-wire gamble this time

According to GamesRadar+, EA CEO Andrew Wilson says roughly 85% of EA’s baseline QA workload is now handled by machine learning or AI-driven algorithms—while also stating EA employs more QA staff than ever.

For players who have watched game production evolve over decades, this is less about “AI can do everything” and more about where automation actually fits. Wilson’s description is unusually direct: AI is mainly taking routine checks such as power on/off, launch/close, and crash detection. In production terms, that is exactly the layer most suitable for automation, and for major publishers it was always likely to happen sooner or later.

The bigger issue is accountability. When asked whether AI is replacing jobs, Wilson said its role is currently “almost entirely” assistive. Human QA still has to interpret outputs and make judgment calls, especially when dealing with AI hallucinations and inaccurate data. That division makes sense: game quality is not just “does it crash,” but also feel, pacing, and feedback—areas where systems can log behavior without truly understanding player experience. If a company chases efficiency but cuts time for human review, even strong automation metrics can become misleading.

Viewed calmly, EA’s message also signals a broader industry shift: QA jobs are unlikely to disappear overnight, but the skill profile will change. Testers who can run scripts, read logs, and communicate risk priority with developers will gain value; roles centered only on repetitive clicking will face pressure from process optimization. Add in Wilson’s 2024 statement that AI is already core to EA’s business, plus EA’s “cautious and steady” AI stance during acquisition rumors involving investors such as Saudi Arabia’s Public Investment Fund (PIF) and Affinity Partners, and the external messaging is clear: push efficiency, but protect expectations.

As for reports that EA is introducing internal chatbots and planning mandatory use in areas like HR and management operations, those claims remain unconfirmed.

From a Chinese player perspective, this is best treated as a mirror for local studios. AI can handle legwork, but it cannot own responsibility for outcomes. It can accelerate regression testing, but it cannot automatically create the texture that makes a game world memorable. What brings players back is never buzzwords on stage—it is fewer severe bugs on launch day and more evidence of careful polish. The real takeaway is not just the 85% figure, but human-machine collaboration. Without that, the risk is familiar: newer tech, weaker games.