Hyundai Cuts Crash Test Review Time by 90% With AI

Hyundai is applying the same principle to other operational bottlenecks. A reinforcement learning system now automatically reroutes parts-delivery carts around production-line disruptions. The system replaces a process that once required a worker to manually judge and restore cart sequencing whenever equipment errors tangled the route, cutting unnecessary downtime by 86%, according to Hyundai's release. A separate AI tool lets repair technicians describe symptoms or error codes in plain language and receive a diagnosis pulled from manuals and past service records, cutting maintenance response time by 42%.

The common thread is how Hyundai is using AI to reduce the time employees spend waiting for information and searching for an answer. This category of gain does not require solving the much harder problem of automating physical labor itself.

The industrial deployments sit inside a wider rollout. Hyundai's internal H Chat Pro platform, which gives employees access to ChatGPT and Gemini, as well as Claude, inside a company-controlled environment, surpassed 30,000 active users in July, roughly 80% of general employees at Hyundai Motor and Kia, Hyundai said. That adoption did not happen overnight. The company has been building a unified data infrastructure called the Global One Data Pipeline since 2019, standardizing how information flows across the company years before generative AI existed as a mainstream product. Hyundai called that early investment the actual foundation behind this week's numbers.

Hyundai is separately pushing into what it calls the Physical AI era through its ownership of Boston Dynamics, targeting annual robot production capacity of 30,000 units by 2028.