Why these three
Critical thinking, communication, and collaboration aren't a slogan we picked because they alliterate. Here is the data we're reading — including what it doesn't say.
What this evidence does — and does not — show
Everything below establishes two things: that these three abilities are in demand, and that AI puts them at risk rather than making them unnecessary. That is the case for why these three.
None of it is evidence that the Dojo builds them. That is the open question, and it's what we are trying to measure with real students. We would rather say so plainly than imply a result we don't have.
Critical Thinking
Judging what's true, and what's worth doing.
The #1 core skill employers name
DemandAnalytical thinking is the most sought-after core skill, with roughly seven in ten employers calling it essential. The same survey expects 39% of key job skills to change by 2030.
→ World Economic Forum, Future of Jobs Report (2025)~90% of employers look for problem-solving evidence
DemandWhen reviewing a graduate's résumé, problem-solving is the single highest-rated attribute employers say they look for — ahead of teamwork, communication, and technical skills.
→ NACE, Job Outlook (2025)The more people trust AI, the less they think critically
RiskA survey of 319 knowledge workers covering 936 real AI use cases found that higher confidence in generative AI is associated with less critical thinking, while higher confidence in oneself is associated with more. Workers reported skipping critical evaluation precisely when they lacked the skill to inspect and guide the model's output.
→ Lee et al., Microsoft Research & Carnegie Mellon, CHI (2025)The work shifts, it doesn't disappear
ShiftThe same study found generative AI moves critical thinking away from producing an answer and toward verifying information, integrating responses, and stewarding the task. Different work — not less of it.
→ Lee et al., Microsoft Research & Carnegie Mellon, CHI (2025)Communication
Making your thinking legible — to a person or to a model.
Rated important by more than 75% of employers
DemandCommunication sits just behind teamwork among the competencies employers rate most important in new graduates.
→ NACE, Job Outlook (2026)Students think they have it. Employers disagree.
GapEmployers rate new graduates well below how graduates rate themselves on communication — a gap of roughly 25 percentage points, with a similar gap on critical thinking. This is the clearest signal that these abilities need deliberate practice rather than assumed competence.
→ NACE, Job Outlook (2026)Written and verbal both clear the two-thirds bar
DemandMore than two-thirds of employers look for verbal communication skills on a résumé; written communication matters to at least 70%.
→ NACE, Job Outlook (2025)Durable skills are rising, not falling, as AI spreads
Demand76% of job postings now request at least one durable skill and 47% request three or more — up 13 points since 2021. Eight of the ten most-requested skills are durable ones like communication, leadership, and problem-solving.
→ Lightcast (2025)Collaboration
Exploring an idea deeply with a thinking partner.
The top-rated competency of all
DemandTeamwork is rated important by more than 80% of employers — the highest of any career-readiness competency — and around 80% look for evidence of it on a résumé.
→ NACE, Job Outlook (2025 & 2026)One person with AI matched a team of two without it
ShiftA pre-registered field experiment with 776 professionals at Procter & Gamble found individuals working with AI matched the performance of two-person teams working without it. Read carefully, this is not a case for working alone — what AI reproduced was the idea-improvement benefit of having a thinking partner, which is exactly the half of collaboration we train.
→ The Cybernetic Teammate — Dell'Acqua, Mollick, Lakhani et al., NBER Working Paper 33641 (2025)AI dissolved the functional silos
ShiftIn the same experiment, R&D specialists normally proposed technical solutions and commercial specialists proposed commercial ones. Working with AI, both groups produced balanced proposals — the model pulled each of them into territory they would not have explored alone.
→ The Cybernetic Teammate — Dell'Acqua, Mollick, Lakhani et al., NBER Working Paper 33641 (2025)"Working with others" is one of five core skill domains
DemandThe WEF groups the skills of 2030 into technology, cognitive skills, self-efficacy, management, and working with others — with collaboration among the fastest-growing human skills of the next five years.
→ World Economic Forum, Future of Jobs Report (2025)Three honest qualifications
1. We define collaboration more narrowly than these surveys do
When employers rate "teamwork" at 80%, they mean something broad: coordinating with others, dividing work, handing it off, showing up. We train a narrower thing — exploring an idea deeply with a thinking partner through conversation. Our argument for the overlap is this: the coordinating half of teamwork is increasingly handled by tools, while the part that doesn't commoditize is two minds arriving somewhere neither would have alone. The Procter & Gamble experiment is suggestive here — what AI reproduced was the idea-improvement benefit of a partner, not the logistics. But this is our reading of the evidence, not something the surveys themselves claim.
2. Some of the strongest findings are self-reported
The Microsoft and Carnegie Mellon study is a survey of what knowledge workers said about their own thinking, not a controlled measurement of it. The association between trusting AI and thinking less critically is a correlation. It is consistent with what we see in classrooms, and it is not proof of cause.
3. This data has a shelf life
These are surveys of a labor market that is changing quickly. Figures on this page were last checked against their sources in August 2026. If you are reading this much later, treat the numbers as a starting point and follow the links — which is, after all, the habit this whole project is trying to build.
What we're measuring
If these three abilities are the claim, they're also what our build-measure-learn loop has to measure against.
- ✓Session reflections. The Reflector walks students through an honest self-assessment at the end of a session — what they achieved, how much of the thinking was theirs, and where they took shortcuts.
- ✓Depth of engagement. The DIKW level a student reaches and their Creating-vs-Consuming balance are visible during the session and summarized at the end.
- ✓Classroom pilots. CST395 at CSU Monterey Bay, and a planned certificate program with De Anza College, give us cohorts to observe over a full term rather than a single session.
- ○Still missing. We do not yet have a clean pre/post measure of any of the three abilities, and we're not going to pretend otherwise. Building one is the most important open item on this project.
Practice the three
Reading about an ability doesn't build it. The Dojo is free.