AI-900 Practice Question: Describe Artificial Intelligence workloads and considerations

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

• ✗

  Using AI to analyse data collected near the geographic borders of a country

  Why it's wrong here

  A geographic border is a political/country boundary and has nothing to do with computing topology; data collected near a country border is still remote from Azure's datacentres and would require network egress, latency, and potential data movement unless processed on site. The 'edge' in Azure IoT is defined by the last mile of the network, such as on-premises devices and gateways, not by physical proximity to national borders. Analysing such data near the border is only edge AI if compute occurs at the data source.

• ✓

  Running AI inference locally on devices for low latency, offline capability, and data privacy

  Why this is correct

  Running inference locally on an edge device means a trained model executes on the device itself, so input data never leaves the device; this avoids cloud round-trips, giving deterministic low latency for real-time decisions, retains functionality during connectivity loss, and keeps sensitive data on-device. This aligns with Azure's edge AI scenarios, such as Azure IoT Edge running containerised models at the source. No cloud dependency means data privacy and bandwidth savings.

• ✗

  Using AI to detect adversarial attacks at the network perimeter

  Why it's wrong here

  In AI, the 'edge' is the distributed computing tier where data is generated and processed (CPUs/NPUs on devices), not the network security perimeter; adversarial AI attacks are input manipulations that trick a model after deployment, while network perimeter security is an IT security domain using firewalls and intrusion-prevention systems. Detecting adversarial attacks on the network boundary neither runs inference locally nor addresses the computing edge, so this conflates two unrelated uses of the term.

• ✗

  Deploying AI to the most remote Azure region for disaster recovery

  Why it's wrong here

  Azure disaster recovery relies on geo-paired regions, which are still centralised cloud datacentres that require network connectivity and introduce latency; deploying to a distant region provides failover redundancy but does not move compute next to the data source. Edge AI instead runs inference on cameras, sensors, or gateways at the point of collection, so a 'remote region' is not equivalent to the edge and misses the low-latency, offline, and data-privacy goals.

Go deeper

Related to this question