Read from the left: AI generates many answers (cyan). They only become a good decision after passing through the six mental filters in this module. Without the filters, you just copy the first answer.
Detailed content
🔍 Critical thinking
Critical thinking is the ability to questioning information before accepting it: separate fact from opinion, demand evidence, and decide based on it—not on what sounds convincing.
New here? When AI "hallucinates", it invents a fact (a number, a quote, a link) that seems true but nonexistent—and says so with the same confidence as a correct answer. Critical thinking is what catches this.
What is
It's a checking habit: where did this come from? who benefits if I believe it? what’s missing here? Applied to AI, this means not treating an answer as true just because it’s well written. The model generates text that sounds authoritative even when it’s wrong—fluency isn’t proof of accuracy.
Why learn
People who don't think critically become at the mercy of the machine’s answer. In a field that changes every week, the skill that protects you most is knowing how to question things well — knowing which number to trust, which source to challenge, and which conclusion to reject.
📌 Why this matters (WEF data)
Analytical thinking is the world’s #1 core skill: 69% of companies consider it essential—the most cited of all in the Future of Jobs Report 2025 from WEF. It’s not a secondary “soft skill”; it’s first on the list.
✓ Who thinks critically
- ✓Ask for the source before sharing a number.
- ✓Distinguishes predictions from observed facts.
- ✓Looks for the caveat missing from the headline.
- ✓Review the AI’s answer against what you already know.
✗ Who becomes dependent
- ✗Copies the first answer because it "sounded good."
- ✗Confuses AI confidence with correctness.
- ✗Shares a statistic without checking its source.
- ✗Accepts the metric the vendor chose.
⌨️ Practice now (copy and paste into your AI)
Objective: turn AI into your critical checker instead of a blind source.
Aja como um revisor crítico. Sobre a afirmação abaixo:
1) liste cada dado factual e diga se é verificável;
2) marque o que é opinião ou previsão (não fato);
3) aponte qual evidência falta para eu confiar;
4) dê 2 perguntas que eu deveria fazer antes de aceitar.
Afirmação: "<cole aqui a afirmação que você quer checar>"How to verify: AI should separate facts from opinions and admit what no can be verified. If it only agrees with everything, try again and ask for the opposing view.
Well-written text doesn’t prove it’s correct.
AI invents facts with confidence.
The caveat that separates data from smoke and mirrors.
Analytical thinking, skill #1.
💡 Creativity
Creativity is combine references and imagine possibilities that don’t exist yet — finding a different way out, not the average of known ways out.
What is
Human creativity comes from experience, emotion, and purpose. It has two movements: diverge (generate lots of ideas, including bad ones) and converge (choose and refine the good ones). AI is excellent at the first move and weak at the second—it has no taste of its own to know which idea matters.
What does “recombine” mean? AI learns patterns from what has already been written and recombines them. It’s great for variations and drafts. But the leap—the new problem no one had seen—comes from someone who has lived the problem, not someone who has only read about it.
Why learn
When AI generates a thousand drafts in seconds, the bottleneck is no longer produce and becomes choose and originate. Your creativity is what directs the machine toward something worthwhile — and what recognizes a spark amid the volume.
The machine is good at diverge (generate lots of options, in cyan). The human work is to converge: recognize the idea that matters and originate what was missing.
📌 Why this matters (WEF data)
Creative thinking is the 4th core skill of the Future of Jobs 2025: 57% of companies consider it essential. In a world where producing has become cheap, creating value is increasingly tied to originating—not repeating.
⌨️ Practice now
Objective: use AI to diverge and you to converge.
Gere 12 ângulos bem diferentes para resolver o problema abaixo.
Inclua pelo menos 3 ideias estranhas/improváveis.
NÃO escolha a melhor — só liste. Eu escolho.
Problema: "<descreva o seu problema real aqui>"How to verify: you should come away with 2–3 angles you hadn't considered. The creative step is yours: choose one and develop it using your own experience.
Generate many options (AI can help).
Choose and refine (you decide).
The leap that comes from experience.
Creative thinking.
🕸️ Nexialism
Nexialism? It's the art of connect areas that seem separate — technology + behavior + economics + culture. A nexialist isn’t the top expert in any one of these; they’re the person who sees how they relate to each other.
What is
AI is excellent at answering vertical: deep within a well-defined field. A person with a broad perspective makes the move horizontal: connects a psychology finding to a product decision, an economic data point to a career choice. That depends on tacit knowledge — what you know but can’t write in a prompt.
Tacit knowledge is knowledge that comes from experience and can’t fit in a manual: the “feeling” that something is going to go wrong, the ability to read a room. It’s exactly what AI lacks—and what Princeton researchers point to as the limit of obsolescence predictions.
Why learn
People who only know how to ask questions within their own box get answers from that box. People who connect fields create questions AI would never come up with on its own — and that’s where the advantage that can’t be automated lies.
AI is strong inside of each circle. The nexialist lives in the center: sees the connections between technology, behavior, economics, and culture—and that’s where the important questions come from.
📌 Why this matters (Princeton data)
AI's limit is tacit and contextual knowledge. Narayanan and Mitchell (Princeton) show that predictions like "AI will replace job X" usually come from a shallow understanding of the work. Real example: in 2016, radiologists were predicted to become obsolete; a decade later, there is a shortage and salaries are high.
AI goes deeper; you connect.
The knowledge that doesn’t fit in a prompt.
People who know the work can see the limits.
What AI wouldn't suggest on its own.
📚 Breadth of knowledge
A knowledge base is the collection of experiences, readings, and examples that you bring—and that supports both good questions and sound judgment of the answers.
What is
It's your internal library: cases you've seen, patterns you recognize, analogies you have at hand. Your repertoire is what makes you notice that "this looks like that 2019 case"—a connection AI only makes if you give it the right context.
Why learn
There’s a direct link to axis 1 (Leveraging AI): your knowledge fuels the prompt. People with little background ask generic questions and get generic answers — and, worse, can’t judge whether the answer is any good. People with a strong background guide AI and correct its course.
Reading beyond your own field
Most useful ideas come from a neighboring field. Those who read only their own niche end up with the same knowledge as everyone else in that niche—in other words, no advantage.
Note examples and patterns
Knowledge that doesn’t become a note gets lost. Save cases, phrases, and counterexamples — they come back when you least expect it, including as context for AI.
Curate sources, don’t just consume them
Choose of whom learning is half the toolkit. A few excellent sources are worth more than many shallow feeds.
📌 Why this matters (Princeton/MIT data)
A knowledge base is the foundation of domain expertise. Princeton’s research shows that people who master their own field understand AI’s limits better than outsiders who are enthusiasts. MIT points out that the #1 failure when evaluating AI is lacking a reference point to question the result—exactly what a broad knowledge base provides.
Cases and patterns at hand.
Better context → better answer.
Choose who to learn from.
Know whether AI got it right.
📊 Data literacy
Data literacy = data literacy. No is being a data scientist. It means knowing how to read a number, asking where it came from, and not falling for a pretty chart. “Ground truth” is the actual answer key—the truth against which you compare a result.
What is
It's about reading, interpreting, and questioning data to make better decisions. It includes recognizing the vanity metric (a number that impresses but says nothing useful), biased framing, and confusing correlation with cause. With AI, this becomes even more critical: the machine presents numbers with the same confidence, whether they’re right or wrong.
Why learn
According to MIT, the #1 evaluation failure is trust the metric in the vendor’s sales material instead of questioning the ground truth. The question that protects you when evaluating any AI is simple: “is the answer key really true?”
Illustrative panel (not a real screenshot). “98.7%” is the seller’s vanity metric. The three questions in blue are data literacy in action — they target the ground truth, not the headline number.
📌 Why this matters (WEF + MIT data)
Technological literacy is WEF’s 6th core skill (51% of companies). MIT Sloan points to the #1 question when evaluating any AI tool: “is the answer key (ground truth) really true?” — the caveat that separates skill from hype.
⌨️ Practice now
Objective: audit a number before trusting it.
Vou te dar uma estatística. Aja como auditor de dados:
1) qual seria a "ground truth" (gabarito) por trás desse número?
2) que amostra/recorte ela exigiria para ser confiável?
3) é correlação ou causa?
4) que métrica de vaidade pode estar escondida aqui?
Estatística: "<cole o número/afirmação aqui>"How to verify: a good answer points out what’s missing for you to trust it — it doesn’t blindly validate the number.
The real answer behind the metric.
A number that impresses but isn't useful.
Moving together doesn’t mean causing.
Technological literacy.
🎯 Decide what to do
AI helps execute, but someone needs to decide what needs to be done and which problem matters. Getting things done has become easy; making the right decisions has become more valuable.
What is
It's the prioritization skill: choosing, from everything that could to do, what’s worth doing. This includes thinking about opportunity cost (what you give up by choosing this) and separating what’s urgent from what’s important. The machine optimizes the task you give it; it doesn’t choose the right task for you.
The METR illusion. In a controlled experiment (RCT), 16 experienced developers using AI on their own projects were 19% slower — and even after it’s measured, it still thought who had sped up by 20%. Perceived speed is misleading. What protects you is deciding what matters and measure the actual result, not the feeling.
Why learn
If execution has become a commodity, value shifts to whoever defines the problem and measures the outcome. Speeding up the wrong thing just gets you faster to a place that didn’t matter.
The fork that defines everything: before asking AI for speed, ask “does this matter?”. The red branch (speeding up what doesn’t matter) is the METR illusion trap; the blue branch is defining the right problem and measuring the real outcome.
📌 Why this matters (METR data)
The productivity illusion is the study’s most important finding. The METR RCT (2025, reaffirmed in 2026) shows that AI gains depend on skill + task maturity, not the tool. Knowing what doing the work and measuring the actual result matters more than perceived speed.
Self-check (optional): according to the METR illusion, what best protects your results with AI?
AI executes; you choose what.
Perceived speed is misleading (−19%).
What you stop doing.
Results, not feelings.
🧠 Module summary
Next module:
2.2 — Relationship and character skills (empathy, connection, purpose, uncertainty, learning how to learn, and ethical courage)