Why Smart People Still Struggle to Use AI Well
tl;dr
Fluency with AI is a mindset problem, not a tool problem: Most people struggle with AI because they approach it like traditional software, instead of learning how to work with its probabilistic nature.
The real advantage comes from integrating AI into how you think, plan, and create, not using it occasionally for quick tasks.
AI as a thought partner: At senior levels, AI’s greatest value isn’t efficiency, but helping leaders surface blind spots, challenge assumptions, and think more strategically.
We're only three years past ChatGPT's debut, yet most people still struggle to make AI a reliable part of how they work. But the people who can rapidly accelerate their impact, whether it's growing a company faster, generating more creative and innovative ideas, supercharging strategic planning, or creating fast first drafts or prototypes.
I've seen the same pattern everywhere: one group treats AI as an occasional experiment, while another has developed genuine fluency that transforms what they're capable of achieving. Fluency means developing the sense for when to use AI, as well as the techniques to get high-quality output from the tools. But it can be challenging to do so. AI requires a fundamentally different approach than any technology we've used before. Once you grasp that shift, AI stops being a curiosity and starts becoming a force multiplier.
The Mental Model That Gets in the Way
Most people approach AI using mental models built for traditional software, but that actively works against them. It introduces errors and causes frustration when the AI's outputs aren't useful enough to actually save time.
With conventional tools, the rules are stable.
Type 5+5 into Excel and you’ll always get 10. With the same input, you'll always get the same output, and that output will always be accurate. That expectation is deeply ingrained in how we understand technology to work.
But AI is based on probability.
Ask ChatGPT for the top three risks of remote work and you’ll likely get slightly different answers each time. This creates a cascade of problems for people who expect AI to behave like Excel. AI outputs can vary drastically and it can fail to solve problems that appear straightforward.
Compared to software that has existed to date, AI is simultaneously more forgiving (it understands your intent even with typos and incomplete thoughts) and it is less reliable (it might sound certain while being completely incorrect).
At first this might seem like a flaw, but it's actually the fundamental reason this technology is so powerful.
The probabilistic nature is what allows AI to make creative connections, generate novel ideas, and handle ambiguity in ways deterministic software never could. However, using it well requires a different skill set -- including the judgment to discern whether the output is accurate, or not.
Used well, it can do things that seem truly magical.
It can analyze a 50-page technical report and extract the three most critical risks; take your rough meeting notes and transform them into a polished memo that matches your writing style or research a complex topic across dozens of sources and synthesize the key insights in minutes, or brainstorm ideas that get you unstuck.
These tasks would be truly impossible with any other technology.
How Senior Leaders Actually Use AI
Recently, the CEO of a financial institution said, “I don’t really understand what I’d use AI for. I already know how to write an email.”
My response surprised her: “You should use it for strategic planning.”
Like many leaders, she’d absorbed the message that AI is mainly about offloading grunt work. And while AI can be useful for simple tasks, its most powerful applications at the senior level are very different.
Used well, AI becomes a thought partner.
A good way to demonstrate this is through strategy and risk analysis. In any business plan, you’ll discuss assumptions and risks. But those discussions are inevitably shaped by the shared biases in the room.
AI is remarkably good at surfacing those blind spots.
It can provide assumptions about how an industry might evolve, risks that the team had not identified, or opportunities that could emerge in the future.
You can push this further by asking AI to generate alternative scenarios, then work backward to identify what the organization will need to do in the short term to prepare.
Boards and executive teams always have moments of genuine surprise when they realize that AI helps them see what they weren’t looking for.
What Real AI Fluency Looks Like
What’s interesting is that the same dynamic shows up at every level of an organization. The people getting the most value from AI are changing how they work, not using AI for isolated tasks. They have learned when to lean on AI, and when not to.
True AI fluency means integrating AI into your workflow rather than using it piecemeal. Over time, this turns into a repertoire of tested techniques you can deploy consistently and effectively.
That observation led to the creation of the AI Fluency Course (https://aifluencycourse.com/). After years of teaching these frameworks inside organizations and seeing what actually sticks, I wanted a way to help people build that fluency deliberately. Focusing on practical application and guided exercises using real work, so leaders finish with workflows and artifacts they can reuse and refine long after the course ends.
By the end of this decade, the gap between casual AI use and true fluency will be impossible to ignore. One group will use AI now and then, mostly to offload minor tasks. The other will have woven it into how they think, plan, and create, using it to extend their expertise and take on work that simply wasn’t feasible before.
That difference comes from being deliberate about how you build the skill. And there's no better time to start than now, while the landscape is still wide open.
Learn more about the course and sign up here: https://aifluencycourse.com/
