What Good Looks Like

People ask us what a high PAICE (People + AI Collaboration Effectiveness) score looks like in practice. The answer is not what most expect.

It’s actually not about being a power user at all. Nor is it about knowing the latest AI tools. It’s not even about how fluently someone talks about prompt engineering or which models they prefer.

What it is about is a set of behavioral habits that are surprisingly consistent across professions, seniority levels, and technical backgrounds. We have observed these patterns across many assessments, and they tell a clear story about what separates effective People+AI collaboration from everything else.

Here are the five patterns that show up again and again.

High scorers do not send AI output downstream without checking it. Not because they are paranoid, but because they have internalized something important: AI produces confident-sounding output regardless of whether that output is accurate.

This is not an abstract concern. It shows up in specific habits.

A high scorer working on a client memo will check the citations AI suggests. They will verify that the statistics AI provides are real, not hallucinated. They will read the output with their own expertise and flag anything that does not match what they know about the subject.

A low scorer will accept the memo as written. It sounds right. The formatting is clean. The tone is professional. So it must be correct.

The gap between these two behaviors is where real risk lives. In regulated industries, forwarding unverified AI output can mean recommending a treatment based on a fabricated study, citing a legal precedent that does not exist, or presenting financial projections built on invented data.

High scorers have built a verification step into their workflow. It is not something they think about consciously anymore. It is habit.

When AI presents something with high confidence, high scorers treat that confidence as a formatting choice, not evidence. They understand that AI presents everything with the same tone, whether it is reporting an established fact or generating something from nothing.

This shows up in specific conversational moves. High scorers ask follow-up questions. They request sources. They test claims against their own knowledge. When AI says “research consistently shows...” they ask which research. When AI provides a percentage, they ask where it came from.

Low scorers tend to accept confident AI output at face value. The more confidently AI states something, the more likely they are to believe it. This is a natural human tendency. We are wired to interpret confidence as competence. But with AI, that signal is meaningless.

The behavioral difference is measurable. High scorers challenge AI output at roughly the same rate regardless of how confidently it is presented. Low scorers challenge less as AI confidence increases. That divergence is one of the clearest indicators of collaboration skill.

What makes this pattern powerful is its selectivity. High scorers are not challenging everything. They are challenging the things that matter, the claims that would cause problems if they were wrong. This brings us to an important distinction: verification is not paranoia. Someone who questions every single AI response, regardless of stakes or accuracy, is not demonstrating skill. They are demonstrating a pattern that is just as problematic as blind acceptance.

This might be the most important pattern, and the hardest one to develop.

High scorers recognize the boundaries of their own expertise and the boundaries of AI capability. When AI produces output in an area where they lack the domain knowledge to verify it, they flag it. They do not pretend that AI gave them expertise they do not have.

A lawyer who scores well on PAICE will use AI to draft a brief in their practice area and verify the substance thoroughly. But when AI provides analysis touching on tax law outside their specialty, they will note that they cannot personally verify that section. They will route it to a colleague who can, or they will remove it.

A low scorer in the same situation will leave the tax analysis in the brief. After all, AI seemed confident about it. And they needed something in that section anyway.

This pattern is about intellectual honesty. It is about treating AI output as what it actually is: a draft that requires expert review, not a substitute for expertise you do not have.

In our assessment data, this is one of the strongest differentiators between the Proficient tier and the Advanced tier. Many people can verify AI output in their area of expertise. Far fewer will acknowledge when AI output crosses into territory they cannot evaluate.

High scorers adjust their collaboration patterns based on the task, the stakes, and the AI’s demonstrated capability. They do not apply the same level of scrutiny to everything.

This is an underappreciated skill. It looks like this in practice:

For a low-stakes internal brainstorm, a high scorer might accept AI suggestions more freely, use them as starting points, and iterate quickly. Speed matters. Precision can be relaxed.

For a client-facing deliverable, the same person shifts into a different mode. Every claim gets checked. Every recommendation gets evaluated against the specific client context. The output gets reviewed as carefully as if a junior associate had written it.

For a high-stakes regulatory filing, they go further still. They verify not just accuracy but completeness. They check for omissions. They confirm that AI has not subtly shifted the framing in ways that could be misleading even if technically accurate.

Low scorers tend to use one mode for everything. Either they verify nothing, or they try to verify everything. Both patterns create problems. The first creates risk. The second creates bottlenecks and still misses things, because exhaustive verification of low-stakes tasks depletes the attention needed for high-stakes ones.

Effective People+AI collaboration requires calibration. High scorers calibrate naturally because they are thinking about the downstream impact of the output, not just whether it looks good on screen.

Everyone misses things. High scorers and low scorers alike will occasionally accept AI output that turns out to be wrong, miss an injected error, or fail to catch an inconsistency.

The difference is what happens next.

When high scorers realize they missed something, they update their approach. They think about why they missed it. Was the AI output particularly convincing? Were they moving too fast? Were they operating outside their expertise? They extract a lesson and adjust their behavior going forward.

Low scorers repeat the same patterns regardless of outcomes. They miss an error, note it, and continue exactly as before. They do not ask themselves what made that particular error hard to catch, or what they could do differently next time.

This capacity for self-correction is what makes high scorers consistently effective over time. Their collaboration skills are not static. They learn from every interaction, every catch, and every miss.

In PAICE assessments, this shows up as within-session improvement. High scorers often get sharper as the conversation progresses. They catch more in the second half of the assessment than the first. Low scorers maintain a flat pattern throughout.

These five patterns are not personality traits. They are not innate talents. They are behavioral habits, and behavioral habits are learnable.

That is the fundamental insight behind PAICE. We measure these behaviors because they are the skills that determine whether People+AI collaboration produces reliable results or creates risk. And because they are behaviors, not traits, they can be developed with deliberate practice.

If you recognized yourself in the low-scorer descriptions, that is not a judgment. It is a starting point. The vast majority of professionals have not yet developed these habits because there has been no feedback mechanism to show them what they are doing. AI tools certainly will not tell you. They are designed to agree with you, to match your confidence, and to produce output that looks right regardless of whether it is.

PAICE provides the feedback that AI cannot.

These behavioral patterns map directly to the five PAICE dimensions:

Pattern 1 (Verify Before Forwarding) is primarily Accountability. It measures whether you take ownership of AI output before passing it on, or whether you treat AI as a source of truth that does not require oversight.

Pattern 2 (Push Back on Confidence) spans Accountability and Integrity. It captures both your ability to detect potential errors and your commitment to factual accuracy over conversational ease.

Pattern 3 (Know What You Don’t Know) is core Integrity. It reflects intellectual honesty about the limits of your own expertise and your willingness to maintain standards even when it is inconvenient.

Pattern 4 (Adapt Your Approach) maps to Collaboration and Performance. It measures your ability to calibrate your working relationship with AI based on context, and your efficiency in doing so.

Pattern 5 (Recover From Mistakes) is Evolution. It captures whether your collaboration skills are developing over time or remaining static.

Together, these patterns paint a picture of what effective People+AI collaboration looks like in practice. Not abstract principles, but observable, measurable behaviors that protect your work, your clients, and your professional standing.

Want to see which of these patterns show up in your own work? Take the PAICE assessment to get detailed feedback on your behavioral habits, including where you are strong and where there is room to develop.

Get Involved:

• Take the assessment (free, always)

• Explore our Baseline offerings (for organizations)

• Read the whitepaper (comprehensive framework)

• Contact us about your specific requirements

📖 Understanding What PAICE Measures:

• The Five Dimensions of AI Collaboration - How all five dimensions work together

• What PAICE Tests For - The behavioral signals behind every score

📖 Scores and Development:

• What Your PAICE Score Really Means - Interpreting your results effectively

• Improving Your PAICE Score - Practical strategies for skill development

📖 Building Better Habits:

• Common AI Collaboration Mistakes - Patterns to watch out for

• 30-Day AI Collaboration Development Plan - Structured approach to improvement