Why Performance Consulting Is Even More Important in the Era of AI Slop

Am I Crazy to Zag When Everyone Else Is Zigging?

There’s a lot of noise in Learning & Development about the impact AI is having.

Courses in hours instead of weeks. Videos without production crews. AI-generated scenarios, assessments, graphics, scripts, avatars, coaches, simulations and the list seems to grow every week. And I’m starting a consulting company focused on performance consulting and solution architecture.

Am I crazy?

Maybe. But I think the ability to create learning faster actually makes the work we do before we create anything more important, not less. AI has made it a lot easier to produce something, but it hasn't made it any easier to know whether that something is worth producing.

The rise of the term “AI slop”, shorthand for the flood of low-quality, mass-produced digital content created with generative AI, should serve as a warning to us all. AI didn't invent bad content, but It has dramatically increased the velocity at which it can be produced.

The first question when asked to design a learning solution should still be: What performance are we trying to create? This matters even more when our production capacity becomes nearly unlimited.

Fast and Bad Is Still Bad

There is an enormous opportunity for AI to enable L&D to produce solutions that are data-informed, more quickly deployed, more personalized and more dynamic. We should embrace that.

AI can help us analyze information, brainstorm solutions, generate prototypes, create scenarios, produce media, develop assessments, personalize experiences and provide a conversational interface. Work that once took weeks can sometimes happen in days—or hours. That's good.

But fast and bad is still bad.

If anything, fast and bad may be worse. Before AI, the cost and effort involved in developing a learning program created a natural constraint. We couldn't build everything, so at least some decisions had to be made about what deserved investment.

AI removes much of that friction. 

• Give an AI tool a 100-page policy document and ask it to create a course, and it can.

• Give it a PowerPoint deck and ask for 20 micro-learning modules, and it can.

• Give it a competency model and ask for 50 roleplay scenarios, and it probably can.

But should it?

That's not a production question. It's a performance consulting and solution architecture question. And that is why I think these practices become more, not less, important in the era of AI.

Don’t Use AI to Skip the Process. Use AI to Accelerate and Improve It.

The EDUCATE Solution Architect process moves through three phases: Performance Gap Analysis, Solutioning, and Project Definition. Within each, there are activities that traditionally require significant amounts of interviewing, analysis, synthesis, comparison, writing, production, documentation, and planning.

AI can dramatically accelerate that work without eliminating the judgment that makes it valuable.

Phase 1: Performance Gap Analysis

Step 1: Define the Desired State

AI can help turn large amounts of business information into a clearer picture of success. Feed it strategic plans, performance measures, customer research, job descriptions, stakeholder interviews, and other available inputs, and use it to:

• Synthesize stakeholder perspectives.

• Identify recurring business priorities and performance expectations.

• Surface contradictions between stakeholders.

• Translate broad goals into hypotheses about observable employee behaviors.

• Suggest potential measures of success.

• Generate questions for follow-up stakeholder interviews.

• Identify gaps in the information collected.

This approach helps you push beyond generalities to define what employees would do differently, which business metrics would improve, how managers would recognize success, and what customers would experience differently.

Step 2: Create the Desired-State Persona

Once successful performance has been defined, AI can rapidly synthesize those requirements into draft personas.

Ask it to combine role information, audience research, performance expectations, and stakeholder input to describe the desired performer in terms of: Behaviors and actions. Mindset and confidence. Interactions and impact. AI can also create multiple personas when different audience segments contribute to the same outcome differently.

Instead of spending hours drafting and redrafting personas, the Solution Architect can spend that time challenging, validating, and improving them with the people who understand the work.

Step 3: Define the Present State

This may be one of AI's most valuable applications.

The inputs describing current performance are often messy and distributed: survey comments, interviews, performance reports, customer feedback, assessment results, manager observations, support tickets, operational data, recorded calls and other sources.

AI can help:

• Analyze large volumes of qualitative input.

• Cluster recurring performance problems.

• Identify patterns across different sources.

• Compare perceptions among employees, managers, and stakeholders.

• Surface potential anomalies worth investigating.

• Identify where additional evidence is needed.

AI can help us see patterns more quickly, but we shouldn't allow it to turn correlations into unsupported diagnoses about why those patterns exist. 

Step 4: Create Current-State Personas

AI can then turn those findings into personas representing the reality of today's workforce.

It can synthesize recurring behaviors and misconceptions, motivations and blockers, and decision patterns into recognizable employee profiles. It can also identify meaningful variations within the audience that might justify multiple personas rather than designing for an imaginary "average learner."

This is particularly valuable because the goal isn't simply faster persona development.

It's the ability to base personas on far more evidence than a design team could practically synthesize manually.

Step 5: Analyze the Performance Gap With EDUCATE

Now AI becomes a thinking partner.

Give it the desired-state personas, current-state personas, supporting evidence, and performance outcomes and ask it to compare them through each EDUCATE lens.

It can help surface questions such as:

• Engage: Where might motivation be misaligned with the desired behavior?

• Diagnose: Where might employees lack awareness of their own opportunity to improve?

• Upskill: What behaviors or capabilities differ between current and desired performers? What enabling knowledge supports them?

• Consolidate: Which capabilities need to be integrated and rehearsed in realistic situations?

• Assess: Where do we need evidence that employees are actually ready to perform?

• Transfer: What support might employees need at the point of performance?

• Evolve: What foreseeable changes could invalidate assumptions underlying the solution?

AI can generate hypotheses quickly. The Solution Architect determines which are credible, investigates them, and decides which functional needs belong in the architecture.

Analyze Constraints and Preferences

Before moving to deliverables, AI can also help make sense of the practical realities surrounding the solution.

Provide the functional needs along with budget, timeline, geography, technology, SME availability, regulatory requirements, existing systems, learner preferences, and stakeholder expectations.

AI can identify:

• Potential conflicts between requirements and constraints.

• Constraints likely to have the greatest design impact.

• Assumptions requiring validation.

• Opportunities to leverage existing technology or resources.

• Questions that still need to be answered.

• Trade-offs stakeholders will ultimately need to make.

This is where AI starts helping the Solution Architect move from “What would ideally work?” to “What could realistically work here?”

Phase 2: Solutioning

This is where generative AI's ability to rapidly explore alternatives becomes particularly powerful.

For each EDUCATE functional need, AI can generate multiple ways of addressing it.

Instead of immediately deciding that a skill gap requires a workshop, for example, ask AI to generate alternatives across live, virtual, self-directed, manager-led, peer-based, workflow, and AI-enabled experiences.

Then ask it to compare those alternatives against the factors that matter to the project.

The EDUCATE playbook identifies considerations including scalability, upfront development cost, delivery cost, speed to market, maintenance, ability to update, accessibility, practice opportunities, measurement capability, and expected business impact.

This opens up some interesting possibilities.

Give AI the same functional requirements and ask:

Create one architecture optimized for lowest total cost.

Then:

Create one optimized for fastest implementation.

Then:

Create one optimized for maximum opportunity to practice.

Then:

Create one optimized for a global workforce with limited facilitator capacity.

Finally:

Compare the four approaches and identify the tradeoffs we would be accepting with each.

AI can allow a Solution Architect to explore dozens of potential architectures in the time it once took to develop one. And going through this process doesn’t negate leveraging AI to more rapidly develop the solution. It does raise the probability that you’ve landed on the right ones. 

The Solution Architect isn't outsourcing the decision. AI is expanding the decision space. And because each proposed deliverable can be traced back to an EDUCATE need, there should be a clear answer to: Why is this part of the solution?

The end result is more likely to be the integrated ecosystem envisioned by the playbook rather than a single default modality.

Phase 3: Project Definition

Once the architecture has been selected, AI can help turn it into an executable plan. This is less glamorous than generating an AI avatar, but potentially just as valuable.

AI can help develop and pressure-test:

• Final deliverables — Turn the architecture into an inventory of components and assets.

• Development budgets — Develop initial effort and cost assumptions for each deliverable.

• Delivery costs — Model ongoing costs associated with different implementation approaches.

• Resource requirements — Identify the roles, expertise, technology, and materials required.

• Dependencies — Map relationships between deliverables, decisions, SMEs, technology, and other workstreams.

• Milestones and project phases — Organize the work into a logical development and implementation sequence.

• Timelines — Generate initial schedules based on effort, dependencies, resources, and target dates.

• Risks and assumptions — Identify potential failure points and, perhaps more importantly, assumptions that should be validated before development begins.

These are the core elements needed to convert the solution blueprint into an executable project.

AI can also make scenario planning much easier.

• What happens if the budget is reduced by 20%?

• What if the launch date moves forward six weeks?

• What if SMEs aren't available until next quarter?

• What if live delivery isn't feasible in one region?

• What if we need to scale from 500 employees to 10,000?

Instead of rebuilding the plan manually each time something changes, AI can help model the implications and identify where tradeoffs need to be made.

Again, AI doesn't make those trade-offs. It makes the consequences of them easier to see.

AI Should Give Us More Time to Think, Not Less Reason to Think

This is where I see the real promise of AI for L&D.

We can use it to:

• Analyze more evidence.

• Consider more perspectives.

• Generate more alternatives.

• Evaluate more tradeoffs.

• Test more assumptions. 

• And get to an executable solution faster.

Then, yes, we can also use it to produce scripts, graphics, videos, slides, manuals, simulations, assessments, roleplays, job aids, and even eLearning modules at extraordinary speed. But production should come after we've decided what deserves to be produced.

That's the zig.

While much of the industry is asking how AI can help us build learning faster, I think we should also be asking how AI can help us think better before we build.

AI gives us speed. Performance consulting gives that speed direction. And that's how we get to fast and good instead of simply producing AI slop faster.