When someone says, “We should use AI,” they’re not giving you a requirement; they’re giving you a signal. From a PMI perspective, your role is to translate that into value by first asking what problem we’re actually trying to solve.. If the outcome isn’t clear, the solution shouldn’t be either. From there, identify the real need (automation, augmentation, insights, or user interaction), validate whether the necessary data actually exists and is usable, and define success in measurable terms. Only after assessing feasibility, technical, organizational, and governance constraints, should scope be defined. And in some cases, the right answer is not to use AI at all.
Absolutely. “We should use AI” is a directional impulse, not a requirement. It’s no different from someone saying “We need automation” or “We need a dashboard.” It signals ambition, not clarity. The real work begins when we translate that impulse into something operationally meaningful:
What outcome are we trying to improve—speed, accuracy, cost, experience, or decision‑making
What process is actually breaking or underperforming
What data exists, what condition it’s in, and whether it can support the ambition
What constraints—technical, organizational, ethical, or governance—shape the solution space
Only after that discovery can we define whether AI is the right tool, one of the tools, or not needed at all. In practice, the most responsible AI decisions come from teams that are willing to say “AI isn’t the answer here” just as confidently as they say “AI can help.”
...
1 reply by Paul Waggoner
May 13, 2026 4:22 PM
Paul Waggoner
...
Excellent points, it might be best not to define the project as AI related until it becomes obvious that AI can really help accomplish business goals.
“We want something that interacts with users.” That’s experience design—assistants, copilots, or conversational layers.
“We want to enhance what people already do.” That’s augmentation—reducing cognitive load, not replacing roles.
What tends to go wrong is when all of these get thrown into one bucket called “AI,” and suddenly the conversation jumps straight to models, vendors, or tools. That’s usually the moment I know we’re misaligned.
The real red flags for me:
The “solution” is being discussed before the problem
Data is assumed to exist, be clean, and be accessible
Success is described emotionally (“better,” “faster,” “smarter”) instead of measurably
Governance, risk, and operational impact are treated as afterthoughts
Everyone is using the same word—AI—but meaning completely different things
I’ve been in plenty of conversations where “AI” meant automation to one person, analytics to another, and a chatbot to a third. The giveaway is when the discussion feels circular or when people agree… but for different reasons.
That’s why the first job is always translation: What outcome are we trying to improve, and what kind of intelligence—or none at all—does that actually require?
Curious to hear others’ stories navigating these mixed signals. Saving Changes...
Ryan RothPM I| Fellowes IncGlendale Heights, Il, United States
When someone says, “We should use AI,” our first question should be What is the problem? From there you can gather your requirements and thus adding value to the project and request. Saving Changes...
Usually when people talk about AI they are thinking of CHATGPT and Agents, they think there is a universal database where we can all access and find the information they are looking for, for an example they think we can collect all the information from students that are playing soccer and that are about to graduate. Saving Changes...
when a someone ask they want to use ai, i ask them what you trying to solve and guide them through a set of questions to understand their context without judging them. Clarifying their thoughts will be helping them to think about the Why and instead of how. the underlying problem is still the same only the approach changes. understanding the problem will help us develop a better solution . specifically the tendency to apply AI to problems customers don't actually have or to procedures built for a human-only era.
Whenever someone suggests, “we should use AI,” it’s usually a sign that there’s an underlying challenge or pain point they’re hoping to address—rather than an actual solution in itself. To make these conversations productive, it’s helpful to start by clarifying exactly which business outcome they’re trying to improve. Are they aiming to cut costs, speed up a process, boost quality, or gain new insights? Once we know the real goal, we can dig a little deeper to understand what role they envision AI playing. Is it about automating routine tasks, assisting people in their work, providing recommendations, or even making decisions outright? From there, it’s essential to pause and evaluate whether the situation truly warrants an AI solution. This means taking a closer look at whether we have the right data available, assessing our risk tolerance, and being honest about our organization’s technological maturity. Sometimes, the best solution might not involve AI at all. By unpacking these suggestions thoughtfully, we’re more likely to land on approaches that genuinely fit our needs and capabilities. Saving Changes...
Whenever someone suggests, “we should use AI,” it’s usually a sign that there’s an underlying challenge or pain point they’re hoping to address—rather than an actual solution in itself. To make these conversations productive, it’s helpful to start by clarifying exactly which business outcome they’re trying to improve. Are they aiming to cut costs, speed up a process, boost quality, or gain new insights? Once we know the real goal, we can dig a little deeper to understand what role they envision AI playing. Is it about automating routine tasks, assisting people in their work, providing recommendations, or even making decisions outright? From there, it’s essential to pause and evaluate whether the situation truly warrants an AI solution. This means taking a closer look at whether we have the right data available, assessing our risk tolerance, and being honest about our organization’s technological maturity. Sometimes, the best solution might not involve AI at all. By unpacking these suggestions thoughtfully, we’re more likely to land on approaches that genuinely fit our needs and capabilities. Saving Changes...
Avinash KharePM II| MAP-IT Consultant Project ManagementAmbernath (East), Maharashtra, India
Generally at the start when AI gained traction , AI was seen as only to automate the work or being more productive or efficient. The real value lies in which sectors like health , education AI can come and solve problems providing outcomes that really matter. . Saving Changes...
I still think of AI as yet to evolve to substantially support Architecture and Engineering. There are a lot of questions on legality, liability, client support that seem to Architects and Engineers to constantly require human presence while trying to embrace AI tools that are important to stay contemporary. Saving Changes...
PMI Team Member
