Generative AI covers large language models, image generators and related systems that output new material in response to a prompt. In IoT it is rarely the thing running on the device; instead it supports the people and systems around the devices. Examples include writing a maintenance summary from a week of alarms, generating configuration or firmware boilerplate, or creating synthetic training data for rare fault conditions that are hard to capture in the field.
Generative models also power AI agents that plan and act on device data. The main cautions are reliability and provenance: outputs can be fluent and wrong, so anything that reaches an actuator or a compliance report needs checks. Synthetic data must also be validated against real measurements before a model trained on it is trusted. See What is AIoT? for the broader context.
Part of the IoT Glossary. Spotted an error or a missing term? Tell us.