Opening

There are twenty questions you could ask about an AI initiative. There are five that matter. This lesson teaches you which ones and why. These aren't technical questions. They're governance questions. And getting clarity on these five determines whether your initiative succeeds or fails.

An engineer presents an AI system for real estate valuation. They show you impressive performance metrics: 92% accuracy on test data. You're about to approve a 12-month project. But then you ask: 'How was your test data selected?' Long pause. 'We used 10 years of historical transactions.' You follow up: 'What percentage were properties in neighborhoods with significant demographic change?' Longer pause. 'We... haven't specifically analyzed that.' That question just saved you from deploying a model that would likely discriminate against certain neighborhoods. Most leaders don't ask it. You will after this lesson. The questions you ask in 30 minutes can prevent a crisis.

This problem appears everywhere. In boardrooms, vendors pitch AI systems that promise dramatic outcomes. In email, executives debate whether an AI initiative is worth funding. In planning meetings, teams argue about which AI projects are real opportunities versus hype. The language is unfamiliar. The claims are large. The stakes are real. And most leaders don't have a framework for cutting through to what's actually true.

Sarah, the Chief Risk Officer at a $1.2B insurance company, recently sat through a pitch for an AI system that would improve underwriting accuracy. The vendor claimed 22% better accuracy and lower claims loss. Impressive. But Sarah didn't know what questions to ask. Is 22% real? How was it measured? On what data? Against what baseline? The vendor's answer was well-rehearsed but didn't actually address what Sarah needed to know. She left the meeting uncertain, which is worse than skeptical. At least skepticism has a clear direction. Uncertainty leads to inaction or to defaulting to whoever speaks with the most confidence.

You're about to change that. You're going to learn to ask the right questions and understand the difference between real AI capability and vendor aspiration.

Why This Matters

Most leaders ask too many questions or the wrong ones. Like asking a startup about profitability (premature) instead of user adoption (what matters now). Same with AI: certain questions matter more than others, depending on the stage. Knowing which five to ask separates leaders who make good decisions from leaders who get overwhelmed by detail.

A healthcare company deployed an AI system to triage patients. The model was trained on 15 years of historical data from a hospital system where women waited longer in the ER than men. The AI learned this pattern and replicated it. The model wasn't explicitly programmed to discriminate. It simply learned from the historical data. One tough question during proposal review—'How was training data selected? What biases might it encode?'—would have caught this. 67% of failed AI initiatives in organizations cite 'discovered problems during or after deployment' that would have been caught by asking the right questions upfront. You don't need to be an ML engineer to ask smart questions. You need discipline and clarity about what matters.

Let's put numbers to the cost of getting this wrong. Gartner reports that 68% of AI initiatives fail to deliver business value in the first 18 months. The reasons? Mostly not technical. Mostly organizational. But it starts with misunderstanding what's real. Leaders allocate $2.1M to an AI initiative expecting a 30% efficiency gain. They get a 7% gain because the vendor's 30% was based on perfect implementation with dedicated change management, and the company deployed it in a business-as-usual environment. That's a $1.4M gap between expectation and reality.

McKinsey research shows that only 8% of firms scale AI successfully from pilots to enterprise value. The other 92% get stuck. And a primary reason is that the initial business case was built on inflated projections. Stakeholders funded the pilot based on a 'conservative' 20% uplift claim. The pilot delivered 8% uplift. Stakeholders feel betrayed. Funding for the next AI initiative becomes political. This is a organizational cost: eroded trust in AI initiatives, risk-averse decision-making, and competitive disadvantage against firms that can fund and execute AI effectively.

The other cost is opportunity. If you're too skeptical of AI because you've been burned by hype, you'll miss real opportunities. The companies winning at AI right now aren't the ones throwing money at every vendor. They're the ones who can tell the difference between solid technical work and vendor BS. They say 'yes' to the real opportunities. They say 'not now' to the premature ones. They allocate capital effectively. That's a competitive advantage that starts with understanding hype versus reality.

The Core Idea

The five questions: (1) What business problem does this solve? (Not 'what capability do we want to use') (2) How will we know if it's working? (Specific metrics, not vague goals) (3) What's our data situation? (Do we have it? Is it good quality?) (4) What happens if it fails? (Mitigation plan? Escalation?) (5) Who owns the outcome? (Clear ownership, not committee)

Here are the seven questions every leader should ask about any AI project:

1. SUCCESS: How will we measure success? ('Improved accuracy' is vague. '85%+ accuracy on held-out test set covering all customer segments' is clear.)
2. FAILURE: What would it look like if this failed? And what would that cost? (Be specific about business impact.)
3. DATA: Where does the training data come from? What could be wrong with it? (Biases, gaps, obsolescence.)
4. EDGE CASES: What situations is this model NOT designed to handle? (Be explicit about limitations.)
5. MONITORING: How will we know if the model drifts or degrades post-launch? What's the detection process? (And who owns it?)
6. ROLLBACK: If we need to disable this model on an hour's notice, can we? (Yes means you have a plan. Silence means you don't.)
7. ETHICS: Who benefits from this system? Who could be harmed? Have we tested for those harms? Organizations that ask these seven questions as part of their review process catch 78% of potential issues before deployment.

To understand this more deeply, let's build a framework. Mature AI technology (worked on real business problems for 5+ years):

• Classification: Is this email spam? Is this image a cat? Is this transaction fraudulent? This works well.
• Regression: Given these inputs, predict this number. Will this customer spend $X in the next quarter? This works well.
• Anomaly detection: Is this data point unusual relative to the pattern? Has network behavior changed? This works well.
• Recommendation: Given what users like, what should we recommend next? This works well in specific domains with good data.

  These technologies have real track records. They save money. They improve processes. They've been in production for years. When a vendor claims these capabilities, you can be reasonably confident the technology itself is solid. The question becomes: Will it work on your data? Will adoption succeed? Are the economics real?

  Emerging AI technology (2-5 years in production, rapidly improving):

  • Generative language models: Writing, coding, reasoning across domains, explaining, summarizing. Real capability. Real limitations. Hallucination is a real problem. These tools are genuinely useful but require human oversight.
  • Vision models: Specialized to specific domains. Very good at specific tasks. Don't generalize well to new domains. The headline accuracy is often deceptive.
  • Time-series forecasting with deep learning: Better than traditional methods in some cases. Not in others. Requires careful validation.

    When vendors claim these capabilities, you should probe more. The technology is newer. The failure modes are less well understood. Implementation requires more experimentation.

    Vaporware AI (claimed but not production-ready):

    • 'Our AI will replace your whole customer service team.' Nope. It's a tool that handles 35-45% of routine inquiries, requiring human review on complex cases.
    • 'This AI system doesn't need maintenance.' Nope. All systems drift. All systems require monitoring and retraining.
    • 'Our AI understands your business problems after reading your documentation.' Nope. Understanding comes from experimentation with your actual data and processes.

      Here's the key distinction: Real AI advantages in production come from bounded, well-defined problems with good data. Real disadvantages come from oversized change management, data quality issues, and integration complexity. The hype focuses on the capability. Reality includes the integration.

Think of It Like This

Most leaders ask too many questions. The signal-to-noise ratio is terrible. Focus on the five that matter and you'll make better decisions faster.

These seven questions are like the pre-surgery checklist doctors use before operating. They ask: What are we doing? What could go wrong? Have we tested the surgical site? What's our backup plan? These disciplines have cut surgical error rates by 40% in the past 20 years. The same rigor applies to AI deployment. Asking hard questions isn't being difficult. It's being prepared.

Let's extend this analogy further. When you're evaluating a new manufacturing process, you'd ask:

• Where was this tested? In a lab? In a pilot facility? In production for two years?
• On what products? The ones we make? Similar products? Very different products?
• What assumptions does it rely on? Specific labor skills? Specific equipment? Specific material quality?
• How sensitive is the gain to those assumptions? If labor quality drops 10%, does the gain drop 5% or 50%?

  AI evaluation follows the same logic, but translated into data and model language. A vendor claims their system improves loan approval accuracy by 18%. Ask:

  • Tested on what data? Data from your bank? Data from similar banks? General lending data?
  • What types of loans? Mortgages? Personal loans? Small business loans? All types?
  • What's the baseline accuracy? Compared to what? Manual review? An older system?
  • How does accuracy vary by applicant demographic? (This is legally important.)
  • How often will the system recommend 'escalate to human'? (This is operationally important.)
  • What's the worst-case scenario? If the system is wrong, what happens? Is it reversible?

    A 18% improvement that's tested on your data, across your loan types, with demographic parity and clear escalation paths is different from an 18% improvement that's based on academic datasets and hasn't been tested on your applicants. Same accuracy number. Different reality.

What This Looks Like in Real Life

When someone pitches an AI initiative, ask the five questions in order. Watch how the answers improve after they've been asked a few times in your organization. People start preparing better answers. They think more carefully. Governance happens through the questions you ask.

A financial services company was about to deploy an AI system to approve small business loans. Default rates had been 8% historically. The AI model showed 5% default rates in backtesting. Excited leadership wanted immediate deployment. One director asked Question 1: 'How are you measuring success?' The team answered: 'We'll compare actual default rates to the 8% baseline.' The director pushed: 'But what if the model learns to avoid underserved communities? How would we detect that?' That question triggered deeper analysis. They discovered the model had indeed learned geographic biases. Fixing took three months. But deploying would have caused far worse damage—to customers, to the company's reputation, and to the company's legal exposure. You now know that asking uncomfortable questions in advance prevents catastrophic outcomes.

Let's walk through a fourth example in detail. A logistics company with 800 employees and $400M annual revenue evaluated an AI system to optimize their delivery routes. The vendor showed a case study where a similar company reduced delivery costs by 22%. Impressive claim. Before committing $3.2M to the implementation, the company did a detailed pilot.

The pilot revealed several reality gaps:

First, the 22% in the case study was for the vendor's 'standard' delivery environment: urban delivery, predictable traffic patterns, stable fleet size. The logistics company operated in three environments: urban (30% of volume), suburban (40%), and rural (30%). The vendor's system was highly optimized for urban. On suburban and rural routes, the system's recommendations often created longer drive times because they didn't account for the sparse pickup/delivery pattern. The 22% gain compressed to 6% across all routes.

Second, the case study assumed the system would run on historical data. But the company wanted the system to optimize routes in real-time. Real-time optimization requires the system to know traffic conditions, driver availability, and customer timing constraints as they evolve. The vendor's system was good at 'given these constraints, here's the best route.' It was poor at 'these constraints are changing; adjust now.' Retraining and redevelopment would cost another $800K and take 6 months.

Third, the case study didn't account for driver adoption. Drivers who had been optimizing routes themselves for years didn't trust an AI system's recommendations, especially when those recommendations contradicted their experience. The company needed 4 months of change management, driver training, and iterative adjustments before drivers actually followed the AI's recommendations.

The result: A 6% delivery cost reduction (instead of 22%) took 9 months to implement (instead of the projected 4 months) and required $4M in total investment (instead of $3.2M). The system is valuable. It's working. But the gap between vendor claim and delivered value was substantial. The company now has a realistic view of what the system does. And they know that next time they evaluate AI, they'll pilot on their actual data and conditions, not just trust the case study.

Where People Get This Wrong

Mistake 1: Asking all twenty questions. Mistake 2: Not asking the top five. Mistake 3: Asking questions but not listening to answers that don't match your hopes.

73% of organizations ask these questions but don't enforce the discipline of acting on the answers. Mistake 1: Asking questions but not waiting for clear answers. 'How will we monitor for drift?' 'We'll figure it out in production.' Silence in that conversation is red. Don't move forward. Mistake 2: Accepting technical jargon as an answer. 'How will we detect bias?' 'We'll run a fairness audit.' What does that mean exactly? Who runs it? When? What triggers action if issues are found? Get specifics. Mistake 3: Asking the questions but not acting on the answers. You learn the model can't handle edge cases in production. And then... you deploy anyway. That's governance theater. Real governance means using the answers to make decisions.

Let's add three more mistakes that leaders often make:

Mistake six: 'If we implement this AI system, it will fix our underlying data quality problems.' Wrong direction. AI amplifies bad data. If your data quality is poor, an AI system trained on poor data will make poor decisions confidently. You fix data quality first, then add AI. A customer analytics AI system trained on messy customer data will confidently categorize customers incorrectly. It won't suddenly become insightful. Fix the data. Then add AI.

Mistake seven: 'This AI system is a one-time investment. Build it and we're done.' No. AI systems require ongoing maintenance. Models drift over time. New data patterns emerge. New regulations require new constraints. The model you build in month six won't perform the same in month eighteen. Budget for continuous monitoring, retraining, and optimization. Most failed AI initiatives failed because the organization budgeted for implementation but not for operation.

Mistake eight: 'The vendor handles all the risk. If the AI doesn't work, it's their problem.' Legally and operationally, it becomes your problem. Your brand suffers if the AI makes bad recommendations in your name. Your risk exists. You need governance, monitoring, and the ability to turn the system off. Vendors can't take that responsibility away. They can share it. But they can't eliminate it.

Practical Takeaways

Memorize the five questions. Make them your standard. Every AI initiative gets asked the same five. No exceptions. As your team gets better at answering them, add a sixth question. But not before.

The investment in asking questions is trivial compared to the cost of deploying a broken system. Create a simple one-page checklist for your organization using the seven questions. When anyone proposes an AI project, they complete the checklist. Their answers become the basis for approval. If they can't answer a question clearly, the project gets sent back for more work. This takes 30 minutes. It prevents millions in failure.

Sixth, establish an AI evaluation checklist for your organization. What information do you need before you fund an AI initiative? (Testing on your data? Reference customers? Failure mode analysis? Pilot costs? Change management plan?) Standardize the questions. Everyone uses the same framework. This prevents the situation where one leader asks tough questions and another leader approves the initiative without those answers.

Seventh, after an AI system launches, publish a 'reality report.' Compare vendor claims to actual results. 'Vendor claimed 40% efficiency gain. We achieved 12%. Here's why: [data quality, adoption friction, implementation scope].' This builds organizational learning. It teaches your team to hear vendor claims with appropriate skepticism. And it focuses attention on the real levers that determine success: adoption, data quality, and integration, not just the AI algorithm.

Key Insight

Most leaders ask too many questions or the wrong ones. Focus on these five and decision-making becomes clear.

The discipline of asking seven specific questions prevents the majority of AI project failures. You don't need to be a technician. You need to know what to ask.

Before You Move On

Memorize the five questions. When someone pitches an AI initiative, ask them. Watch how the answers improve after they've been asked a few times.

Choose one AI initiative in your organization (or industry). Go through the seven questions. What would the clear answers be? What gaps exist in how the project is being described?

Reflect on a recent AI initiative in your organization (or your industry). What were the original projections? What has the actual impact been? What accounts for the gap, if any? Is the gap because of hype, or because of valid reasons like implementation complexity or change management friction?

Now do this: Find one claim you're tempted to believe about AI. It might be 'AI will replace 40% of white-collar jobs by 2027' or 'Our AI system will improve accuracy by 30% with no organizational change needed.' Write down why you believe it. What's your evidence? What could prove you wrong? Run it against this lesson's framework. Is it a bounded claim about a specific technology on specific data? Or is it an oversize claim that sounds good but lacks specifics? This is the habit that separates decision-makers from people who get burned by hype.