The Human Edge: Why Human Judgment Beats AI in Decision‑Making

In her book, "The Human Edge," Cheryl Strauss Einhorn argues that while answers are readily available in the age of AI, humans still need to understand what truly matters, what information to trust, and what actions to take next. These are uniquely human questions that only we can answer.

The Genesis of "The Human Edge"

Einhorn's motivation for writing the book stemmed from her observations while working with individuals, companies, and students. She noticed that much of the conversation revolved around how to use AI tools, but people struggled with understanding their own role alongside these tools. Key questions emerged: What is my unique responsibility? Which parts of decision-making should I delegate to AI, and which should I retain?

Einhorn suggests that for the first time, we need to identify our "special sauce"—our unique way of making decisions. Most people haven't consciously recognized their decision-making processes or why they rely on them. By engaging in metacognition—thinking about our thinking—self-awareness can become a competitive advantage. This involves understanding the problem we're solving, why we're solving it, and the key contextual factors that only we can grasp. By bringing these human elements forward, we can use AI to get specific answers tailored to what matters to us, rather than relying on general or readily available information.

AI as a Collaborator, Not a Decision-Maker

Einhorn emphasizes AI's role as a collaborator, not a primary decision-maker. She highlights the risks of passively allowing AI to make critical decisions. For complex problem-solving, there are eight critical steps where uniquely human judgment and information are essential:

1. Problem Definition: Clearly articulating the core issue.
2. Motivation: Understanding the underlying reasons for solving the problem.
3. Context: Grasping the broader environment and nuances.
4. Setting the Research Direction: Guiding the inquiry.
5. Conducting the Analysis to Find Meaning: Interpreting data beyond surface-level facts.
6. Identifying and Controlling for Biases: Recognizing and mitigating personal and systemic biases.
7. Including Stakeholders: Incorporating diverse perspectives.
8. Coming to Conviction: Developing a firm belief in the chosen path.

AI cannot inherently know the answers to these steps.

Two Modalities for AI Interaction: The Surgeon and The Lamborghini Driver

To illustrate how humans can effectively interact with AI, Einhorn introduces two memorable modalities:

• The Surgeon: This applies when we know precisely what piece of information is missing. We use AI like an incision to extract that specific answer.
• The Lamborghini Driver: This describes situations where we are navigating through several of the eight problem-solving steps. We, as humans, are behind the wheel of this powerful machine (AI), directing it. It's crucial to know enough about the steps to guide the machine effectively, as quickly arriving at the wrong destination is unhelpful.

Einhorn provides an example: when deciding which team member to assign to a new project, an AI tool might suggest the person with the most experience based on anonymized data. However, a human manager might recognize that a more junior person is ready for a stretch opportunity or would approach the project differently. AI often optimizes for the "best" outcome (e.g., financially advantageous), but human decisions are based on values, emotions, and relationships—factors AI cannot inherently know or account for unless explicitly communicated.

Personal Impact and Unexpected Insights

Einhorn shares that her research for the book altered her own perspective on AI. Initially, she underestimated its value in human-centric areas. However, she discovered that AI can significantly broaden her perspective. For instance, she can ask AI to articulate the biggest criticisms of an idea or to anticipate pushback from a colleague in a meeting, helping her prepare better and foster trust.

She also learned to use AI with greater intention. Instead of asking "just because I can do something, should I?", she now asks, "Is there a purpose for me doing it on my own?" She notes that discomfort often signals meaning, indicating that something important is at stake, perhaps a value or emotion.

The Increasing Value of Human Qualities

As AI becomes more capable, certain human qualities become even more valuable:

• Self-awareness (Metacognition): Understanding our own thinking processes.
• Judgment: This is paramount. Historically, experience led to judgment. Now, with AI providing answers without requiring direct experience, the source of judgment shifts. Judgment helps us identify what's missing, what doesn't look right, and where to challenge AI's answers. We must actively question AI's potential biases, assumptions, and how our own biases might create feedback loops. The goal is to seek disconfirming information, which is more robust than merely confirming a favored hypothesis.
• Responsibility: The ultimate responsibility for decisions rests with humans. We cannot blame the tool.

Einhorn suggests collecting examples of AI failures as teaching tools to help build judgment, allowing experts to point out where sensitivities or credibility issues might have been overlooked.

Misconceptions About Competing with AI

Einhorn clarifies that it's not a competition with AI. AI excels at collecting and analyzing vast amounts of research quickly. However, it cannot tell us why we are solving a problem or what problem we are solving. While AI can help refine prompts, humans must define the problem, understand their motivation, and identify key contextual factors.

She also observes that many people treat AI interaction as a transaction, when it should be a conversation. We need to cultivate the habit of first consulting ourselves, and then, when AI provides an answer, actively questioning its potential flaws and biases. Einhorn recounts an instance where AI fabricated a citation, highlighting that plausible-sounding information can be entirely false. Developing judgment to discern flaws and checking primary sources are crucial.

Even if AI becomes "perfected" and stops hallucinating, the onus remains on humans to understand its underlying assumptions and decide what answers to accept.

What Machines Fundamentally Struggle With

Machines fundamentally struggle with:

• Exercising Judgment: Only humans know which problems to solve and how to apply judgment.
• Non-Economic Decision-Making: Humans often make decisions based on values, relationships, and emotions, willingly accepting suboptimal outcomes to preserve these. This complex calculus is uniquely human and varies from person to person based on individual motivations.

The AREA Method: A Decision-Making Process

Einhorn developed the AREA method years ago, born from her experience as a financial journalist investigating companies. She realized the need for a rigorous process to ensure the truthfulness and ethical implications of her reporting, especially given the significant impact her articles had.

AREA is an acronym for a process designed to control for cognitive biases, expand knowledge, and improve judgment. Unlike classical problem-solving, which often starts with external sources, AREA emphasizes developing an internal compass.

• A - Absolute Information: This involves gathering primary source information directly related to the decision's target. For example, understanding the exact policy details of AI usage in schools. This step defines the problem without jumping to assumptions or solutions.
• R - Relative Information: This is where external sources are consulted. By first gathering absolute information, one can vet external information against primary sources, discerning relevance and how narratives might shift.
• E - Exploration and Exploitation:
  • Exploration: Goes beyond documents to understand lived experiences and qualitative factors that numbers cannot convey (e.g., a 10-mile distance taking 30 minutes due to traffic).
  • Exploitation: A unique step where assumptions and judgments are tested against the evidence collected in earlier stages.
• A - Analysis: This final step synthesizes the process, leading to conviction.

The AREA method provides confidence and conviction in decisions made amidst uncertainty, ensuring a quality due diligence process. When using AI, AREA helps ensure that all perspectives are considered, and assumptions are checked and challenged.

Einhorn notes that people have different "data preferences" (e.g., some prefer expert opinions, others numbers). AREA ensures a comprehensive understanding by guiding users through all steps, even uncomfortable ones, and constantly checking information integrity. AI can assist with all steps of AREA, but the method itself ensures that information is collected from diverse perspectives.

AI's Impact on Diverse Teamwork and Problem-Solver Profiles

Einhorn's research identifies five problem-solver profiles, each approaching problems and AI differently:

1. The Adventurer: Intuitive, strong gut feeling, optimistic bias. They quickly identify a direction and are most likely to use AI in "surgeon mode" to extract specific information.
2. The Detective: Likes research and hard evidence, prone to confirmation bias. They seek data and evidence from AI, often projecting their data-centric approach onto others.
3. The Listener: Collaborative, cooperative, trusting, with a less developed gut. Prone to social proof or liking bias. They might use AI to gather others' opinions, especially from trusted advisors.
4. The Thinker: Cautious, thoughtful, prone to relativity bias. They want all available information (data, expert opinions) and often jump over problem definition. They are uniquely able to explain their decision-making path.
5. The Visionary: Seeks unconventional and outside-the-box solutions. They might use AI to find lesser-known information or diverse voices.

Each profile has a preferred data type and quantity. AI can be used to simulate the diversity of these profiles within a team, ensuring a comprehensive understanding of a problem. It can also play "devil's advocate," generating scenario analyses and pushback, allowing team members to collaborate on solutions rather than being solely responsible for critiquing ideas.

Strengthening Your Human Edge

To strengthen one's human edge, Einhorn offers two key takeaways:

1. Question Delegation: Ask yourself: "Just because AI can do it, is there a purpose for me to be doing this on my own?" Identify the parts of decision-making that are uniquely yours and should not be delegated.
2. Cultivate Critical Thinking: Before engaging with AI, always start with yourself. Ask:
   • What problem am I solving?
   • Why am I solving it (my motivation)?
   • What are the key contextual factors that will make or break it? This ensures you have a critical framework to react to AI's answers.

Einhorn challenges individuals to embrace the invitation to better understand their own thinking. This involves identifying personal values and morals related to outcomes, which not only strengthens individual thinking but also allows for more effective and efficient use of AI. She argues that AI, rather than diminishing human value, offers an unprecedented opportunity to understand our unique problem-solving approaches and underlying values, ultimately empowering us to shape our future.