Meta Tells Its Metaverse Workers to Use AI to ‘Go 5X Faster’ (www.wired.com)

Meta’s VP of Metaverse, Vishal Shah, instructed teams to make AI “a habit” and use it to “go 5X faster” under an internal AI-for-Productivity (AI4P) initiative — not mere marginal gains. The directive calls for integrating AI into “every major codebase and workflow,” getting PMs, designers and engineers to prototype, fix bugs and iterate faster, and training to hit a target of 80% Metaverse employee AI adoption by year-end. The push sits amid broader company moves — Zuckerberg has predicted AI agents will write most of Meta’s code in 12–18 months, interviews now permit AI-assisted coding, and the metaverse unit has already consumed tens of billions in investment. For the AI/ML community this is a clear signal that large tech teams will move from experimenting with LLMs to operationalizing them across roles, accelerating demand for developer-facing models, toolchains, observability, and fine-grained governance. But it also amplifies known risks: “vibe coding” and opaque LLM-generated code are producing comprehension and technical debt that make debugging harder and can turn engineers into babysitters for brittle outputs. The implication: productivity gains hinge not just on model quality but on integration patterns — provenance, testing, code review, monitoring, and training — to avoid scaling fragile practices into systemic failures.