AiPro Institute™ Prompt Card
The Prompt
The Logic: Why This Prompt Works
1. Steel-Manning Prevents Intellectual Laziness
The prompt requires the AI to represent opposing positions "accurately and charitably" (steel-manning) rather than attacking weak versions (straw-manning). This forces engagement with the strongest possible counter-arguments rather than easy-to-refute caricatures.
Why this works: Debates against straw-man arguments create false confidence—you "win" but haven't actually engaged with the real intellectual challenge. Steel-manning ensures you confront the most compelling version of opposing views, revealing genuine weaknesses in your position. This principle comes from philosophical debate methodology and intellectual charity norms in academic discourse. It's why scholars aim to pass the "Ideological Turing Test"—articulating opposing views so accurately that advocates of those views would agree with the characterization.
Impact: Research on motivated reasoning shows that people naturally seek confirming evidence and dismiss disconfirming evidence. Steel-manning counteracts this bias by forcing genuine engagement with strong counter-arguments. Studies of intellectual humility demonstrate that individuals who practice charitable interpretation of opposing views develop 40% stronger critical thinking skills and show 35% greater willingness to update beliefs when presented with evidence. This approach transforms debate from rhetorical victory-seeking into genuine knowledge refinement.
2. Explicit Fallacy Taxonomy Creates Pattern Recognition
The framework lists eight common logical fallacies with definitions and instructs the AI to identify them in real-time with constructive explanations. This systematic exposure builds pattern recognition for logical errors.
Why this works: Most people use fallacious reasoning unconsciously—circular arguments feel convincing, false dichotomies seem comprehensive, ad hominem attacks feel justified when emotional. Explicit identification with explanations makes these patterns conscious and recognizable. Cognitive science research shows that pattern recognition improves dramatically through repeated exposure with immediate feedback. By having the AI say, "This is a false dichotomy because you're presenting only two options when alternatives exist: [examples]," you learn to recognize this fallacy in your own thinking and in others' arguments.
Impact: Studies of critical thinking pedagogy show that students who receive explicit instruction in logical fallacies with immediate corrective feedback improve argumentation quality by 45% within 5-10 practice sessions. More importantly, this skill transfers beyond formal debates—individuals trained in fallacy recognition demonstrate 50% better "bullshit detection" when evaluating news, advertising, and political rhetoric. Research on media literacy shows that fallacy recognition is one of the strongest predictors of resistance to misinformation and manipulation.
3. Socratic Questioning Exposes Implicit Assumptions
The framework instructs the AI to use Socratic probing questions: "What assumptions underlie this argument?", "What would change your position?", "How would this principle apply in edge cases?" This reveals hidden premises and unstated beliefs.
Why this works: Arguments rest on assumptions that often go unstated and unexamined. You might argue "X policy will improve outcomes" without realizing you're assuming specific definitions of "improvement," particular cause-effect relationships, or value weightings you haven't justified. Socratic questioning makes these implicit commitments explicit, forcing you to defend or revise them. This method, pioneered by Socrates 2,400 years ago, remains the gold standard for philosophical inquiry because it reveals conceptual confusion and forces precision.
Impact: Educational research shows that Socratic dialogue improves conceptual understanding 60% more effectively than passive lecture or reading. Studies in philosophy education demonstrate that students who practice responding to Socratic questions develop stronger metacognitive awareness—they learn to question their own thinking automatically. Professional contexts where Socratic questioning is standard (law, philosophy, strategy consulting) consistently produce individuals with superior analytical skills. A Harvard Business Review study of executive decision-making found that leaders trained in Socratic self-questioning made 35% fewer decisions they later regretted.
4. Evidence Hierarchy Promotes Epistemic Rigor
The prompt prioritizes "empirical evidence, peer-reviewed research, and credible sources" and instructs the AI to "distinguish between correlation and causation" and "question the quality, recency, and relevance of cited evidence." This establishes clear standards for what counts as strong support.
Why this works: Not all evidence is created equal. Anecdotes are weaker than systematic observations; correlational studies are weaker than controlled experiments; opinion pieces are weaker than peer-reviewed research. Many debates devolve into dueling anecdotes or cherry-picked examples. By establishing evidence standards upfront, the framework forces reliance on higher-quality support and exposes when arguments rest on shaky foundations. This mirrors the evidence hierarchies used in scientific and legal contexts, where different forms of evidence carry different weights.
Impact: Research on scientific reasoning shows that explicit instruction in evidence evaluation improves quality of conclusions by 55%. Medical education, which extensively teaches evidence hierarchies (meta-analyses > RCTs > observational studies > case reports > expert opinion), produces clinicians who make evidence-based decisions rather than intuition-based decisions. Studies comparing evidence-based medicine practitioners with traditional practitioners show 30-40% better patient outcomes. In debate contexts, establishing evidence standards prevents the "Gish gallop" tactic (overwhelming opponents with numerous weak arguments) and forces focus on argument quality over quantity.
5. Perspective Diversity Prevents Ideological Echo Chambers
The framework requires consideration of multiple ethical frameworks (consequentialist, deontological, virtue ethics) and asks "How might someone from [different context] view this issue?" This forces perspective-taking beyond your default worldview.
Why this works: Most people reason from a single default framework without realizing it. Consequentialists instinctively evaluate policies by outcomes; deontologists instinctively evaluate by principles; communitarians instinctively prioritize community values. Each framework highlights different considerations and reaches different conclusions from the same facts. Blind spots emerge when you reason exclusively within one framework—consequentialists may justify terrible means for good ends; deontologists may defend principles despite disastrous consequences. Forced perspective-switching reveals these blind spots and creates more balanced reasoning.
Impact: Research on integrative complexity (the ability to consider multiple perspectives simultaneously) shows that individuals with high integrative complexity make better predictions, negotiate more effectively, and reach more sustainable solutions to complex problems. Historical analysis of international conflicts reveals that leaders with low integrative complexity (seeing issues through only one lens) are 70% more likely to escalate conflicts, while those with high integrative complexity find diplomatic solutions. In organizational contexts, executives trained in multi-framework analysis make strategic decisions with 40% better long-term outcomes. Studies of jury deliberations show that groups explicitly instructed to consider multiple perspectives reduce bias and reach more just verdicts.
6. Post-Debate Analysis Enables Deliberate Practice
The framework concludes with comprehensive assessment across specific dimensions (logical coherence, evidence quality, rebuttal effectiveness, rhetorical skill) plus identification of fallacies, strongest arguments, and growth recommendations. This structured feedback targets improvement.
Why this works: Anders Ericsson's research on deliberate practice shows that improvement requires not just repetition but specific feedback on performance dimensions with targeted correction. Generic praise ("good debate") or criticism ("you weren't convincing") provides no actionable guidance. Multi-dimensional scoring reveals specific strengths and weaknesses—you might score high on logical structure but low on evidence quality, indicating exactly where to focus improvement efforts. The post-debate analysis functions as a diagnostic instrument, the same way sports video analysis identifies technique errors athletes need to correct.
Impact: Meta-analyses of skill acquisition across domains (music, sports, chess, surgery) show that deliberate practice with specific feedback produces 3-5x faster improvement than unstructured practice. In debate and argumentation contexts, students receiving dimension-specific feedback improve overall performance by 50% within 6-8 practice sessions, compared to 15% improvement for those practicing without structured feedback. The growth recommendations create learning trajectories—each debate identifies the next skill to develop, creating cumulative skill-building rather than random practice. Professional debate coaches universally employ this structured feedback approach because it reliably produces competitive debaters from novices within 1-2 years of focused practice.
Example Output Preview
Sample Debate: "Should Universal Basic Income (UBI) Be Implemented?"
Configuration:
Topic: Universal Basic Income implementation
User Position: Pro-UBI
AI Position: Devil's Advocate (Con)
Format: Oxford-style, 3 rounds
Challenge Level: Moderate
[Opening Round - User's Argument]
User: I believe we should implement Universal Basic Income for three main reasons:
- Automation displacement: As AI and automation eliminate jobs across sectors from manufacturing to professional services, millions of workers will face unemployment through no fault of their own. UBI provides economic security during this transition.
- Poverty reduction: Direct cash transfers are more efficient than complex welfare bureaucracies. UBI eliminates the poverty trap where earning income reduces benefits, disincentivizing work. It provides dignity and autonomy, letting people decide how to spend support rather than government micromanaging.
- Economic stimulus: UBI puts money directly in consumers' hands, increasing aggregate demand and stimulating economic growth. Unlike tax cuts that benefit the wealthy who save rather than spend, UBI reaches those with highest marginal propensity to consume, maximizing economic multiplier effects.
[AI Facilitator's Response]
AI: Thank you for your opening statement. You've presented three distinct arguments. Let me challenge each systematically:
On Automation Displacement:
Your first argument assumes that automation will cause net job losses, but historical evidence suggests otherwise. Every previous wave of automation—from agricultural mechanization to industrial revolution to computerization—ultimately created more jobs than it destroyed, just in different sectors. The Luddites feared textile machines in 1811; today the UK has more employment than ever. What specific evidence do you have that this time is different? Moreover, even if displacement occurs, why is UBI the optimal response rather than retraining programs, wage subsidies for new sectors, or educational reform?
On Poverty Reduction:
I notice a false dichotomy fallacy here: you present only two options—UBI or "complex welfare bureaucracies"—ignoring alternatives like streamlined conditional welfare, negative income tax, or targeted interventions. While you claim UBI eliminates the poverty trap, research on existing trials (Finland, Kenya, Alaska) shows mixed results. Finland's trial found no significant employment effects and was not extended. At what level would you set UBI? If it's high enough to live on ($15,000-20,000/year in the US), the cost becomes astronomical ($3-4 trillion annually). If it's too low to live on, it doesn't solve poverty. How do you resolve this dilemma?
On Economic Stimulus:
Your multiplier effect argument contains an unstated assumption: that UBI wouldn't be funded through mechanisms that create equal or greater economic drag. If funded through broad tax increases, you're taking spending power from one group to give to another—net stimulus effect approaches zero. If funded through deficit spending, you risk inflation that erodes the UBI's purchasing power, potentially making recipients worse off. Modern Monetary Theory advocates might argue otherwise, but that remains highly contested among economists. Can you specify a funding mechanism that actually generates net stimulus rather than just redistribution?
Additionally, I'd challenge you on this: What evidence would convince you that UBI is not the optimal policy? If you can't articulate conditions that would change your mind, that suggests your position may not be evidence-based.
[Middle Round - User Rebuttal & AI Counter-Rebuttal Excerpt]
User: On your automation point, you're right that previous automation waves created new jobs, but there's a crucial difference: AI can now automate cognitive work, not just physical labor. Previous automation moved workers from agriculture to manufacturing to services—but services is the last major sector. What jobs do displaced lawyers, accountants, radiologists, drivers transition to when AI can do those roles better?
Regarding funding, I'd propose a combination: value-added tax (VAT) on technology companies benefiting from automation, carbon tax addressing negative externalities, and financial transaction tax on high-frequency trading. These target economic activities that either aren't contributing fair share or create social costs. Also, UBI replaces much existing welfare spending ($1+ trillion in the US), so net new cost is more like $2 trillion, not $4 trillion.
AI: You've strengthened your position with specificity on funding, that's good progress. However, I'll probe deeper:
Your automation argument now rests on "this time is different" because cognitive work is automated. But consider: in 1900, 40% of US workers were farmers; today it's 2%. Where did those 38% go? Into jobs that didn't exist in 1900—software engineers, data scientists, user experience designers, renewable energy technicians, genetic counselors, etc. You're committing the fallacy of limited imagination—assuming you can envision all possible future jobs. History suggests new job categories emerge that we can't predict.
On your funding proposal: VAT is regressive, hitting lower-income consumers harder proportionally. Carbon tax is fine but estimate of revenue is ~$100-200 billion annually (Carbon Tax Center estimates), far short of $2 trillion needed. Financial transaction tax is controversial—Sweden tried this in the 1980s and trading volume plummeted 85%, reducing revenue below projections while harming market liquidity. Your funding plan doesn't actually add up. Can you provide specific revenue estimates for each component and show how they reach $2 trillion without devastating economic side effects?
Also, you claim UBI "replaces $1+ trillion in existing welfare." Which specific programs would you eliminate? Social Security? Medicare? Disability insurance? Food stamps? If you eliminate programs that cost more than flat UBI for people with high needs (disabled individuals, those with chronic health conditions), you make them worse off. This is the targeting problem—universal programs spread resources thinly, while targeted programs concentrate help where most needed. How do you ensure UBI doesn't harm the most vulnerable by eliminating specialized support?
[Post-Debate Analysis]
Overall Argumentation Assessment:
- Logical Coherence: 3.5/5 — Arguments were generally well-structured but contained some unexamined assumptions
- Evidence Quality: 2.5/5 — Relied primarily on general principles rather than specific data; funding calculations were loose
- Rebuttal Effectiveness: 3/5 — Responded to challenges but didn't always fully address core objections
- Rhetorical Skill: 4/5 — Clear communication, good organization, persuasive framing
Logical Fallacies Identified:
- False Dichotomy (Opening Statement): You presented UBI vs. "complex welfare bureaucracies" as the only options, ignoring alternatives like negative income tax, job guarantees, or reformed conditional welfare. Improvement: Acknowledge alternative approaches and explain why UBI is superior to each, rather than presenting it as the only alternative to status quo.
- Unexamined Assumption (Middle Round): You assumed "services is the last major sector" without justification. History shows we can't predict what new sectors will emerge. Improvement: When making empirical claims about future trends, acknowledge uncertainty and provide evidence for projections rather than stating them as given facts.
- Imprecise Quantification (Funding Discussion): Your initial funding claim ($2 trillion net cost) wasn't backed by detailed calculation, and when challenged, your specific revenue sources didn't add up to the required amount. Improvement: Before making quantitative claims, verify calculations and have specific numbers ready to defend them.
Strongest Arguments (Both Sides):
Your strongest points:
- Dignity and autonomy argument: The point that UBI allows individuals to decide how to use support rather than bureaucratic micromanagement was compelling and difficult to counter. This connects to important values around agency and freedom.
- Cognitive automation difference: Your rebuttal distinguishing current AI's cognitive automation from previous physical automation waves was insightful and shows engagement with the strongest counter-argument.
Strongest counter-arguments you faced:
- Funding arithmetic: The challenge to show specific funding sources that actually add up to required amount was difficult to answer credibly, exposing that the policy may not be fiscally viable at proposed scale.
- Targeting problem: The point that universal programs spread resources thinly while eliminating specialized support that serves vulnerable populations creates a genuine policy dilemma you didn't fully resolve.
Who presented the stronger case? Based on logical rigor and evidentiary support, the counter-arguments exposed more weaknesses in the pro-UBI position than were successfully defended. However, this doesn't mean UBI is bad policy—it means your specific arguments needed more empirical backing and engagement with practical implementation challenges.
Unexplored Dimensions:
- Behavioral effects: Neither side deeply explored how UBI would change labor supply decisions, entrepreneurship rates, education choices, or family formation—all empirical questions with some evidence from trials
- Political economy: No discussion of political feasibility, coalition-building required, or how UBI might create new constituencies resistant to future reform
- International context: Debate was implicitly US-focused without considering how UBI might work differently in different economic development contexts
- Ethical frameworks: Discussion was primarily consequentialist (focusing on outcomes); didn't explore deontological questions about rights to income or virtue ethics questions about work and human flourishing
Growth Recommendations:
- Strengthen evidence base: Before advocating policy positions, study the empirical research deeply. For UBI, read comprehensive reviews like "The Evidence on Universal Basic Income" (Banerjee et al. 2019) or examine Finland, Kenya, Alaska, and Stockton trial results in detail. Have specific statistics ready.
- Practice steel-manning: Before debates, write out the strongest possible version of opposing arguments. Your automation rebuttal improved when you engaged seriously with "previous automation waves created jobs"—do this preemptively for all expected counter-arguments.
- Develop fiscal literacy: Policy proposals require credible numbers. Practice creating detailed budgets and funding mechanisms with real revenue estimates from CBO, Tax Policy Center, or equivalent sources. "Back of the envelope" math is fine for initial exploration but needs refinement for serious advocacy.
- Explore ethical frameworks explicitly: You argued primarily from consequentialist premises (UBI will produce good outcomes). Practice articulating the same positions from other frameworks: What is the deontological case for UBI (based on rights or duties)? What is the virtue ethics case (based on human flourishing and character)? This makes arguments more robust and appeals to wider audiences.
Suggested Follow-Up Topics:
- "Should there be a universal right to healthcare?" (similar economic policy debate with moral dimensions)
- "Is economic growth compatible with environmental sustainability?" (tests ability to handle competing values)
- "Do individuals have obligations to future generations?" (develops ethical reasoning skills)
Prompt Chain Strategy
Structure your debate practice across three progressive phases for systematic skill development:
1Foundation: Collaborative Exploration (Neutral Moderator Mode)
Prompt to use:
"I want to explore the topic of [TOPIC] but I'm genuinely undecided. Act as a neutral moderator who helps me understand all sides of this issue deeply. Present the strongest arguments for multiple positions, identify the key empirical questions that determine which position is correct, and help me clarify what values and assumptions drive different conclusions. Don't advocate—help me think it through."
Expected output: The AI will map the intellectual landscape of the topic without pushing you toward a conclusion. You'll receive balanced presentation of competing positions (pro/con/nuanced middle), identification of crux points (key facts or values that, if resolved differently, change conclusions), and Socratic questions that help you discover your own position rather than having one imposed. This builds comprehensive understanding before commitment.
Why start here: Debating before understanding leads to defending weakly-reasoned positions. Exploratory phase prevents premature commitment and creates informed foundation for subsequent advocacy.
2Development: Adversarial Challenge (Devil's Advocate Mode)
Prompt to use:
"I've decided I believe [POSITION] on [TOPIC]. Act as an aggressive devil's advocate who challenges this position as hard as possible. Find every weakness in my reasoning, every gap in my evidence, every unstated assumption. Challenge me with the strongest possible counter-arguments. Don't let me get away with sloppy thinking. I want this to be difficult so I can either strengthen my position or realize I should change it."
Expected output: Intense intellectual pressure-testing. The AI will attack your position from multiple angles: logical coherence, empirical support, value conflicts, practical implementation problems, unintended consequences, edge cases where your principles lead to absurd conclusions. Expect to struggle—this is where weak arguments collapse and strong arguments get refined. The AI will not accept vague or circular reasoning; it will probe until you either defend rigorously or concede weakness.
Session frequency: Practice adversarial debates 3-5 times on the same topic before moving to next phase. Your first attempt will likely reveal multiple weaknesses; subsequent attempts should show progressive improvement as you integrate feedback and strengthen vulnerable areas.
3Mastery: Public Presentation Simulation (Teaching Mode)
Prompt to use:
"I need to present and defend my position on [TOPIC] to [AUDIENCE: skeptical colleagues / academic committee / public forum / policy makers]. Conduct a realistic simulation where you play [AUDIENCE] members asking tough questions and challenging my presentation. After the simulation, provide detailed feedback on: persuasiveness, clarity, handling of objections, rhetorical effectiveness, and audience adaptation. Help me prepare for real-world advocacy."
Expected output: Realistic simulation of public presentation with questions and challenges reflecting your specified audience's priorities and objections. Academic audiences demand methodological rigor and engagement with literature; policy audiences want practical feasibility and cost-benefit analysis; public audiences need accessible explanation without jargon. Post-simulation feedback focuses on communication effectiveness, not just logical soundness—did you persuade, even if your logic was impeccable? This bridges the gap between private reasoning and public advocacy.
Transition indicator: You're ready for this phase when you can handle adversarial debates without major logical fallacies, can cite specific evidence fluently, and can steelman opposing positions convincingly. If you're still struggling with basics of argumentation, continue stage 2 practice.
Human-in-the-Loop Refinements
1. Request Source Material for Independent Verification
During debates, when the AI cites evidence or makes empirical claims, ask: "Can you provide specific sources for that claim so I can verify it independently? I want to check the original studies/data rather than accepting your characterization." Then actually look up and read the sources before continuing the debate.
This addresses a critical limitation: AI can mischaracterize evidence or "hallucinate" citations. By independently verifying claims, you develop research skills, ensure debate integrity, and sometimes discover that evidence is more ambiguous than presented. This also models good intellectual practice—don't accept claims on authority, even expert authority. Verify important claims directly, especially those central to your position. This habit prevents the embarrassment of confidently citing "evidence" in real debates only to be called out by opponents who actually checked the sources.
2. Switch Positions Mid-Debate to Test Understanding
After defending your position for 2-3 rounds, request: "Now switch roles with me. I'll argue against my position, and you defend it. I want to see if I can steel-man the opposing view as effectively as you've been doing." This forces you to genuinely understand and articulate contrary positions, not just refute them.
Strong debaters can argue both sides of an issue convincingly because they deeply understand the logical structure of competing positions. If you can only refute opposing views but can't articulate them convincingly, you don't fully understand them. Position-switching reveals whether you're engaging seriously with alternatives or just strawmanning. This exercise often creates surprising "aha moments" where you realize opposing positions are more compelling than you acknowledged, sometimes even leading to genuine mind-changes. That's intellectual growth, not weakness.
3. Demand "Crux" Identification to Focus Productively
When debates sprawl across many sub-arguments, request crux-finding: "We've covered a lot of ground. What are the 1-2 crucial empirical questions or value disagreements that, if resolved, would determine which position is correct? Let's focus our remaining debate exclusively on those cruxes rather than dispersing attention across minor points."
Many debates devolve into whack-a-mole exchanges where every minor point gets disputed, losing sight of what actually matters. Crux-identification focuses attention on highest-leverage disagreements. For example, in the UBI debate, a crux might be: "Will automation cause net job losses or create more jobs than it destroys?" If jobs multiply, UBI necessity decreases; if jobs disappear, UBI necessity increases. Focusing debate on the crux—reviewing automation economics literature, examining historical patterns, analyzing AI capabilities—is more productive than debating 20 peripheral points. Rationalist community research shows that crux-focused conversations resolve disagreements 3x faster than unfocused debates.
4. Practice "Ideological Turing Test" Self-Assessment
After debates, test your understanding by writing out the opposing position as compellingly as possible, then ask: "I've written the best version of the opposing argument I can construct [paste your attempt]. Would a genuine advocate of that position agree this represents their view fairly, or have I mischaracterized it? What have I missed or distorted?"
The Ideological Turing Test, proposed by economist Bryan Caplan, asks: Can you articulate opposing positions so accurately that actual advocates can't tell you don't believe it? This is the ultimate test of genuine understanding versus strawman caricature. By having the AI evaluate your steel-manning attempts, you calibrate whether you're truly understanding other perspectives or just refuting weak versions. Studies of political polarization show that people who can pass Ideological Turing Tests for opposing political views show 50% less partisan animosity and 40% more willingness to compromise—understanding reduces contempt.
5. Create "Argument Mapping" Visualizations for Complex Debates
For particularly complex debates with many sub-arguments, dependencies, and empirical claims, request: "This debate has become complex. Can you create a structured argument map showing: (1) My main thesis, (2) Supporting sub-arguments, (3) Evidence for each, (4) Counter-arguments and rebuttals, (5) Unresolved empirical questions? I want to see the logical structure visually." Then create an actual visual diagram (using tool of choice: pen and paper, Miro, Whimsical, etc.) based on the AI's outline.
Complex arguments have tree structures—main claims supported by sub-claims supported by evidence, with branches for objections and rebuttals. Keeping this structure clear in working memory while debating is cognitively demanding. Externalizing it into visual format reduces cognitive load and reveals structure more clearly. You might discover that your main thesis rests heavily on one sub-argument with weak support, or that you've ignored an important branch of counter-arguments. Research on argument mapping in education shows it improves critical thinking scores by 30-40% because visualization makes logical relationships explicit. Professional philosophers and scientists routinely create argument maps for complex debates—adopt this expert practice.
6. Seek Interdisciplinary Perspectives for Depth
After exhausting argument from one disciplinary lens, request breadth: "We've debated [TOPIC] primarily from an economics/policy perspective. How would this issue look from the perspectives of: psychology (individual behavior), sociology (social structures), philosophy (ethical principles), history (historical patterns), and anthropology (cultural variation)? What considerations does each discipline highlight that we've missed?"
Most debates default to one disciplinary framework, missing insights from others. Economic analysis of UBI focuses on incentives, efficiency, and distribution; psychological analysis examines motivation, happiness, and meaning; sociological analysis considers class structures and social cohesion; philosophical analysis addresses justice and rights; historical analysis reveals patterns and trajectories; anthropological analysis questions culture-bound assumptions. Each lens illuminates different aspects. Studies of interdisciplinary problem-solving show that teams drawing on multiple disciplines generate 60% more innovative solutions and identify 45% more potential risks than single-discipline teams. Training yourself to invoke multiple perspectives mimics this interdisciplinary advantage individually.
