I think I’d start with, “What are we actually trying to accomplish?” Saying we should “use AI” can mean so many different things. I also think we need to understand what AI policies or guardrails are in place so we’re protecting the work, the data, and the people involved. Saving Changes...
It depends on who is saying it. Most people use AI like Google on Steroids. There are just a few who take the time to understand the complexities or push capabilities. Most of the time, when I hear the request, it just means the client is looking for ways to streamline work or make pain-points more bearable. It's my job to figure out where AI will help, and where AI will hurt. Saving Changes...
When someone says “we should use AI,” I first try to understand what problem we are actually trying to solve. I look at the purpose, the type of work, the data involved, and how much human judgment is still needed. The problem starts when everything is simply called “AI,” because automation, analysis, decision support, and generative AI are very different things. I’ve been in conversations where people were using the same word but talking about completely different things. Usually, the questions they ask and the result they expect make the difference clear. Saving Changes...
I agree with you, many times people are just pressured to use AI, but it is necessary to get the requirements clear first.
When someone says" we should use Ai", there is definitely more to the story that needs to be unpacked. By understanding why someone suggested that, we can explore how and when to implement if needed. Different people perceive ai differently, therefore it's always best to do a depth analysis, so that when ai is run, it doesn't face any resistance from users. Saving Changes...
Carlos Andrade ErasoAcademic Faculty Member| Universidad del CaucaPopayan, CAU, Colombia
The signals I look for are the business problem, AI capability, data requirements, level of uncertainty, and delivery approach. For example, predictive analytics, generative AI, computer vision, and traditional automation have very different success criteria, risks, and technical dependencies. When everything is simply labeled “AI,” teams can underestimate data readiness, model validation, integration, cybersecurity, and change-management requirements. It can also lead to unclear scope, unrealistic expectations, and difficulty defining meaningful KPIs or acceptance criteria. Saving Changes...
Carlos Andrade ErasoAcademic Faculty Member| Universidad del CaucaPopayan, CAU, Colombia
I have seen conversations where “AI” meant everything from a rules-based automation to a machine-learning model or a generative AI solution. The first signal is usually when stakeholders describe the expected outcome differently—for example, one person talks about automating a process while another expects the system to learn, predict, or generate content. I usually clarify the problem, decision or task being augmented, data involved, expected output, and how success will be measured. That quickly reveals whether we are discussing the same solution or simply using “AI” as a broad label. Saving Changes...
Anonymous
In the contact center world, QA is a space where AI has grown and expanded beautifully. With capabilities such as call scan and score automation, getting the “sentiment” from a call, bots that can role play and score interactions, etc. Most recently, I proposed creating a role play bot for new hire learning; my leadership team loved the idea, but were understanding it as a QA bot that would scan existing interactions and score them. It wasn’t until they shared a new QA form that they mentioned how excited they were to be able to resolve their poor scan counts that are currently manual. I had to set up time for me to walk them through what the bot can and cannot do, as well as gathered details of what they wanted to solve for so that I can build a different solution along with the role play bot. Sometimes things don’t translate well from description to understanding until one sees it in action, so it’s always great to follow up with a demo or some visual process flows allowing full understanding of what the AI tool used will do. Saving Changes...
One thing that stood out to me from this module is how often we use the term “AI” without first clarifying what we actually mean or what problem we are trying to solve. I’ve noticed that conversations about AI can quickly become too broad. Someone may say, “We should use AI to improve efficiency,” but that doesn’t tell us what outcome we are trying to achieve. As project professionals, I think it is important to step back and ask: What are we trying to accomplish? Is the goal to automate a task, predict an outcome, generate content, support decision-making, or create a more autonomous process? The discussion also made me realize that choosing the right AI approach should come after understanding the business need—not the other way around. Otherwise, we may end up selecting a tool simply because it is popular rather than because it actually solves the problem. For me, the biggest takeaway so far is that good AI project work starts with asking the right questions and clearly defining the desired outcome before deciding which AI solution to use. Saving Changes...
Stephen KabantiokTechnical Auditor| NGMCPort Harcourt, Choose A State Or Province, Nigeria
Mar 25, 2026 11:29 AM
Replying to anonymous
...
I agree with you, many times people are just pressured to use AI, but it is necessary to get the requirements clear first.
That's true we need to manage our expectations around how far can AI go and how close can we trust it to get things done. I use AI and am curious to ask what makes it think in certain ways after it has spilled out some outcome. AI is here and will continue to improve our work but before we get the AI we all have been waitng for lets be more specific when prompting. Saving Changes...
"You can make more friends in two months by becoming interested in other people than you can in two years by trying to get other people interested in you."
PMI Team Member
