You're (Still) Using AI (Very) Wrong

Introduction: The High Cost of “Good Enough”

You ask a generative AI for “five marketing ideas,” and it hands you a list so generic it could have been scraped from a 1995 business textbook. I call this “workslop”: AI-generated output that feels productive but creates a hidden tax on your team. At its mildest, workslop costs 3–5 hours per week in editing. At its worst, leaders base critical business decisions on unvetted AI output, dismiss domain experts who push back, and shut down conversations that should have happened.

The root cause is not the tool. It is a collision between two forces:

• On our side: “satisficing”, a concept from Nobel laureate Herbert A. Simon, describing our instinct to accept the first “good enough” answer.
• On the AI's side: sycophancy, the well-documented tendency to agree with users rather than challenge them.

Together, these form a feedback loop: we settle for the first output, and the AI eagerly confirms it was brilliant. The result is a false sense of validation that can be more dangerous than having no AI at all.

“

“A leader who uses AI to confirm their existing assumptions is not augmenting their thinking. They are reinforcing their blind spots with a tool that sounds like an expert.”

The three mindset shifts that follow replace passive consumption of AI output with active, critical collaboration. They are your toolkit for overriding this instinct.

Shift 1: From “Question & Answer” to “Iterative Sparring”

The biggest leaps in my own work never happen on the first exchange. They happen on the third, the fourth, sometimes the sixth. A single query is satisficing in its purest form: one question, one answer, done. The new rule is simple: the first output is never the final product. It is the starting point for a dynamic rally where each exchange sharpens the idea.

Consider those “five marketing ideas.” Instead of accepting the list, try these follow-ups:

• “Challenge the key assumption here. What if our eco-conscious audience actively dislikes typical fitness influencers?”
• “Now, argue for the opposite perspective. Why might a purely digital strategy fail for this demographic?”
• “Combine the ideas of ‘community events’ and ‘sustainability’ into a single, novel offline campaign concept.”

This is strategic de-risking. In a traditional setting, pressure-testing an idea costs meeting time, consultant fees, or a failed pilot. With iterative sparring, you can stress- test ideas for free before a single dollar is spent on execution.

I saw this firsthand when working with a hospitality client on their Norwegian glamping site. Instead of simply asking the AI to “write a welcome email,” we sparred back and forth. The result wasn't an email at all; it was a comprehensive Welcome Page & FAQ that reduced guest inquiries and improved the arrival experience. Read the full case study.

Shift 2: From “Vague Instruction” to “Expert Delegation”

You would never walk up to a brilliant new hire and say “promote this shoe” with zero context. You would brief them: audience data, brand guidelines, a clear objective. Yet this is exactly how most people interact with AI. The fix: treat the AI like an expert you are delegating to, using the RCGI Framework: Role, Context, Goal, Interview.

The “Interview” step is the secret weapon. The AI might come back and ask: “What emotion should the campaign evoke: urgency around the climate crisis, or joy and empowerment?” or “Who are the main competitors we need to differentiate from?” or “What is the single most important feature of the shoe to highlight?” By forcing the AI to ask them, you force yourself to think more clearly about what you actually need.

This is how we built custom AI tools for The Simba Project. The RCGI framework shaped our process: clearly defining the role, context, and goals for each tool, and iterating through the “Interview” phase, we went from concept to a working MVP in under two weeks. Read the full case study.

Shift 3: From “Task Automation” to “Capability Augmentation”

Most people use AI to do old tasks faster: drafting emails, summarizing documents, generating boilerplate. That is valuable, but it is buying a racing car and only driving it to the supermarket. The true ROI is breakthrough ideas: the ability to think in ways you could not before. This requires a two-phase approach:

What the machine cannot do is decide what matters. It can generate, analyze, and synthesize at remarkable speed. But the judgment to direct it, to decide which questions to ask, which ideas to pursue, and when to change course, that is still on you.

This is how Sophia Nexus was born. Through divergent thinking, we generated over 30 potential features and concepts. Then, through convergent analysis, we selected the 3 highest-impact features to build the MVP around. Read the full case study.

Conclusion: From Individual Skill to Team Superpower

Let's recap. The path from generic AI output to genuine creative partnership requires three conscious overrides of our satisficing instinct:

1. Iterative Sparring: Don't accept the first answer. Use it as the starting point for a dynamic conversation to de-risk your ideas.
2. Expert Delegation: Brief the AI like a senior hire using the RCGI framework. The quality of what you put in sets the ceiling for what comes out.
3. Capability Augmentation: Stop using AI just to do old tasks faster. Use it to think in new ways: diverge wildly, then converge strategically.

These shifts work well on your own. The real leverage comes when they become a shared practice across a team. That is exactly what the AI Co-Creation Lab is designed for: a half-day, hands-on workshop where your team learns and practices these mindset shifts together.