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.

• ✗

  Including all team members in the AI development process regardless of technical skill

  Why it's wrong here

  This option conflates workplace diversity with the AI principle of inclusiveness. Inclusiveness in AI-900 refers to the design and intended outcome of the AI system itself — ensuring it serves people with disabilities, varied cultural and linguistic backgrounds, and diverse demographic groups equitably. A heterogeneous development team can reduce blind spots, but it does not guarantee that the end product is accessible or empowering to all users. The principle's focus is on who the AI benefits, not who builds it.

• ✓

  Ensuring AI systems empower and benefit all people including those with disabilities and diverse demographics

  Why this is correct

  This correctly defines inclusiveness in Microsoft's responsible AI framework: AI systems should empower and benefit everyone, including people with disabilities and people across diverse demographic categories. It requires accessible design, such as support for screen readers, sign-language translation, and alternative text, as well as language support that reflects the variety of users' backgrounds. The principle also demands equitable performance across different age groups, genders, cultures, and abilities, so that no segment of the population is underserved. This is the standard AI-900 definition of the inclusiveness principle.

• ✗

  Making AI models available to all organisations regardless of their budget

  Why it's wrong here

  Providing AI models to organizations regardless of budget addresses affordability and access through licensing or pricing, which is a business or economic consideration. The inclusiveness principle, by contrast, focuses on the AI system's ability to empower and engage everyone by accommodating human diversity — such as people with visual, hearing, or cognitive impairments and speakers of non-dominant dialects or languages. A free or low-cost AI can still be non-inclusive if it does not work equitably across demographic groups or fails to support accessibility requirements. Hence this option misidentifies a commercial distribution model as the core principle.

• ✗

  Including diverse training data sources to improve model accuracy

  Why it's wrong here

  Diverse training data is a practical technique for mitigating bias and improving model generalization, but it is a means, not the definition of inclusiveness. Inclusiveness is a governing principle that asks whether the AI system's interface, outputs, and value are accessible and beneficial to all people, including those with disabilities and underserved language groups. A model can be trained on diverse data yet still fail inclusiveness tests if it lacks accessibility features or does not support assistive technologies. Therefore this option describes a data quality tactic, not the inclusiveness principle itself.

Go deeper

Related to this question