The WITH AI framework provides a practical, human-centered method for using generative AI without surrendering responsibility. Users begin by asking what they are trying to accomplish, invite AI only where it adds value, test the result, and ensure human judgment owns the outcome. The framework positions AI as a working partner while preserving human purpose, critical thinking, ethical responsibility, and accountability.
There is an important difference between using AI to support the work and asking AI to do the work for us.
The first approach can extend our ability to explore, organize, compare, practice, and create. The second can produce an answer without producing understanding.
That distinction is central to the WITH AI intervention framework developed through my doctoral research. The framework is designed to help people engage with generative AI as active participants in the work, not passive recipients of an output.
W: What are you trying to accomplish?
Begin with the task, problem, or desired outcome rather than with a particular AI tool.
Clarifying the purpose helps prevent technology from shaping the work simply because it is available. A vague request such as “make this better” gives the tool responsibility for deciding what better means. A stronger starting point names the audience, desired outcome, constraints, source material, and standard for quality.
In some situations, ordinary software, a conversation with another person, or no technology at all may be the better choice. AI can help refine a question, but it should not determine the purpose.
I: Invite AI where it can add value
Identify the part of the task where AI could provide meaningful assistance.
AI might help generate possibilities, organize information, ask clarifying questions, explain unfamiliar material, recognize patterns, translate text, simulate a conversation, or automate a routine step. Inviting AI selectively keeps the person's experience, context, and priorities at the center of the work.
Effective use is rarely one prompt followed by one finished answer. It is an interaction in which the person provides context, examines the response, challenges assumptions, requests alternatives, and improves the direction. The user is not merely editing the final wording. The user is shaping the work.
T: Test the result
AI-generated responses can be incomplete, inaccurate, biased, overly agreeable, or presented with more confidence than the evidence warrants.
Testing means checking important facts, examining assumptions, comparing the response with original sources, considering who may have been overlooked, and deciding whether the result actually fits the situation. The amount of checking should increase as the possible consequences become more significant.
This is especially important in work involving people, policy, safety, privacy, strategy, or consequences that cannot be easily reversed. Plausibility is not the same as truth, judgment, or fit. Verification is not a final administrative step. It is part of the work.
H: Human judgment owns the outcome
A person remains responsible for the final decision, how the result is used, and how it affects others.
This includes protecting private information, following applicable rules, seeking qualified human expertise when necessary, and recognizing when AI should not be used. An answer can be accurate and still fail because it does not reflect the organization, the people involved, or the reality of the situation.
Ownership should remain clear. AI can make recommendations, propose alternatives, and help us think through a decision. It cannot accept accountability for the choice. If your name is on the work, you are responsible for it.
Use reflection to improve the partnership
After using AI, ask what the tool contributed and what you contributed. Did it save time, improve quality, introduce a new perspective, or make expertise more accessible? Did it also create new review work, reduce understanding, or tempt you to accept a convenient answer too quickly?
Reflection turns isolated use into learning. It helps people decide when AI adds value, when a different approach would be better, and how the workflow itself should evolve.
Working WITH AI is not about protecting every old way of working or rejecting automation. It is about preserving the human capabilities that make the work meaningful and accountable while using technology to extend what people can accomplish.
