This functionality is experimental and may be changed or removed completely in a future release. Elastic will take a best effort approach to fix any issues, but experimental features are not subject to the support SLA of official GA features.
The autoscaling feature enables an operator to configure tiers of nodes that self-monitor whether or not they need to scale based on an operator-defined policy. Then, via the autoscaling API, an Elasticsearch cluster can report whether or not it needs additional resources to meet the policy. For example, an operator could define a policy that a warm tier should scale on available disk space. Elasticsearch would monitor and forecast the available disk space in the warm tier, and if the forecast is such that the cluster will soon not be able to allocate existing and future shard copies due to disk space, then the autoscaling API would report that the cluster needs to scale due to disk space. It remains the responsibility of the operator to add the additional resources that the cluster signals it requires.