Before You Use That Chatbot Response
From the AI IQ Blog
by Paul Boudreau
Technology offers an incredible opportunity to improve project performance. This blog shares the latest research and how organizations are implementing AI into their project methodology. Come with an open mind, increase your knowledge, share your concerns, and become a project manager with new skills to offer an organization.
Recent Posts
Use AI to Improve Stakeholder Communication
AI Makes Data Quality a Project Leadership Issue
AI Changes How We Manage Project Risk
Before You Use That Chatbot Response
Five Ways to Make AI More Reliable on Your Projects
Categories
AI,
Artificial Intelligence,
Ethics,
Machine learning,
Natural language processing,
procurement,
Scope Management
Date
For the many project professionals I contact, using AI does not mean implementing a sophisticated AI system. It means opening a chatbot and asking it to analyze a risk, summarize a document, draft a communication, review a schedule, or recommend an action. The ease of getting an answer creates a new challenge: deciding whether the answer should be used.
PMI’s
Standard for Artificial Intelligence in Portfolio, Program, and Project Management emphasizes the importance of interpreting AI outputs critically and recognizing when generated content does not align with the original intent, factual accuracy, or business policy.
Before using a chatbot response in a project, I recommend a simple check:
Chatbot Response Checklist- Accuracy: Are the facts and calculations correct?
- Context: Does the response reflect the actual project situation?
- Completeness: Is important information missing?
- Assumptions: What assumptions has the chatbot made?
- Evidence: Can important claims be supported by reliable sources?
- Consistency: Does the response align with current project information?
- Confidentiality: Is sensitive project information being protected?
- Impact: What could happen if the response is wrong?
Generating an answer with AI takes seconds. Determining whether that answer is appropriate for the project requires project knowledge, context, and judgment. Using AI effectively is not just about knowing what to ask. It is knowing what to do with the answer.
Posted on: September 14, 2026 08:00 AM |
Permalink
Comments (1)
Please login or join to subscribe to this item
Luis Branco
CEO| Business Insight, Consultores de Gestão, Ldª
Carcavelos, Lisboa, Portugal
Paul, I think this is a very useful shift from asking how well we can prompt AI to asking whether its output is appropriate for a particular use.
I would add one dimension to the checklist: proportionality.
A chatbot response used to improve the wording of an internal message and one used to inform a consequential risk or schedule decision should not require the same level of scrutiny.
The evidence and verification required should reflect not only the likelihood of error, but also the materiality, uncertainty, and reversibility of the consequences if we act on that output.
I would also distinguish verifying the response from assessing whether the inference we draw from it is justified.
A claim may be accurate and supported by a reliable source without necessarily justifying the conclusion or action proposed in a specific project context.
So perhaps, after asking “What could happen if the response is wrong?”, I would add another question: “What does this response actually justify us in doing?”
For me, that is where AI literacy increasingly meets professional judgment: not simply determining whether an answer looks reliable, but deciding what reliance its evidence and context can legitimately support.
Please Login/Register to leave a comment.
|
"Failure is unimportant. It takes courage to make a fool of yourself."
- Charlie Chaplin
|
