10 Priorities for AI’s Present and Future

🎯 Defined Learning Outcomes

After studying this material, students should be able to:

1. Understand the dual nature of AI advancement (opportunities vs. risks)
2. Recognize the importance of multidisciplinary approaches to AI governance
3. Distinguish between short-term, long-term, and lasting AI priorities
4. Evaluate why balanced stakeholder collaboration matters in AI development
5. Apply a framework for thinking critically about AI's societal impact

---

📚 Step-by-Step Teaching Guide

---

STEP 1: Setting the Context — Why Does This Matter?

The Core Premise:

AI is not just a technology upgrade — it may be a civilizational turning point

Think of it this way:

Key Insight: The article compares AI's impact to the Industrial Revolution — meaning we are not talking about incremental change, but fundamental restructuring of how society works.

---

STEP 2: Understanding the "Now, Later, and Lasting" Framework

This is the central organizing concept of the article.

TIMELINE OF AI PRIORITIES
─────────────────────────────────────────────────────
NOW          →    LATER         →    LASTING
(Immediate)       (Mid-term)         (Permanent/Enduring)

Current harms     Emerging risks     Civilizational
& benefits        & opportunities    implications
─────────────────────────────────────────────────────

Why does this three-part framework matter?

• Focusing only on NOW = missing catastrophic future risks
• Focusing only on LATER = ignoring people being harmed today
• Both are critical — this is the article's central argument

💡 Alan Turing Quote to Remember: "We can only see a short distance ahead, but we can see plenty there that needs to be done." — This supports acting on immediate problems while staying alert to future ones

---

STEP 3: The One Hundred Year Study on AI — Who Is Speaking?

Understanding the Source:

ONE HUNDRED YEAR STUDY ON AI (AI100)
├── Founded: ~2014 (a decade before this 2024 article)
├── Mission: Perpetual, ongoing evaluation of AI's effects
├── Approach: Multidisciplinary expert panels
└── Output: Recommendations for research, policy, and practice

Why "One Hundred Years"?

Key Takeaway: This is not a one-time report — it is a living, evolving project designed to grow with AI itself.

---

STEP 4: The Multidisciplinary Imperative

Core Argument: AI cannot be understood or governed by engineers alone.

The Required Disciplines:

         ENGINEERING
              │
    ┌─────────┼─────────┐
    │         │         │
 SOCIAL   BEHAVIORAL  ECONOMIC
SCIENCES   SCIENCES   SCIENCES
    │         │         │
    └─────────┼─────────┘
              │
         AI GOVERNANCE
         (Complete Picture)

Real-World Example to Understand This:

Imagine an AI hiring tool that is technically flawless but discriminates against women.

• An engineer might say: "The algorithm works correctly"
• A sociologist might say: "It reflects historical bias in training data"
• An economist might say: "It will reduce workforce diversity and long-term productivity"
• A behavioral scientist might say: "Candidates will change behavior to game the system"

You need ALL perspectives to understand the full impact.

---

STEP 5: The Tension in the Field — Short-Term vs. Long-Term

This is a real, active debate in AI:

Why does this divide exist?

SHORT-TERM THINKERS argue:
"People are being harmed TODAY by biased AI systems.
Focusing on sci-fi futures distracts from real victims."

LONG-TERM THINKERS argue:
"If we don't solve alignment now, future AI could be
catastrophic at civilizational scale."

THE ARTICLE argues:
"This is a false choice. We must do both."

---

STEP 6: The Three Domains of Action

The article calls for action across three interconnected domains:

┌─────────────────────────────────────────────┐
│              DOMAINS OF ACTION              │
├─────────────┬──────────────┬────────────────┤
│  RESEARCH   │    POLICY    │   PRACTICE     │
├─────────────┼──────────────┼────────────────┤
│ Scientific  │ Regulations  │ How AI is      │
│ understanding│ & governance │ actually used  │
│ of AI models│ frameworks   │ day-to-day     │
├─────────────┼──────────────┼────────────────┤
│ "How does   │ "What rules  │ "How do we     │
│ it work?"   │ should exist?"│ deploy it      │
│             │              │ responsibly?"  │
└─────────────┴──────────────┴────────────────┘

Key Insight: All three must advance together — research without policy = ungoverned power; policy without research = uninformed rules; practice without both = chaos.

---

STEP 7: Neural Models — The Scientific Understanding Gap

A specific technical concern raised:

"It is crucial that we engage in efforts to advance our scientific understanding of these models and their behaviors."

What does this mean in plain language?

CURRENT PROBLEM:
Modern AI (neural networks/large language models) are
often "black boxes" — even their creators don't fully
understand WHY they produce certain outputs.

THIS IS DANGEROUS BECAUSE:
├── We can't predict failures reliably
├── We can't explain decisions to affected people
├── We can't guarantee safety properties
└── We can't correct problems we don't understand

THE CALL TO ACTION:
Invest in INTERPRETABILITY and EXPLAINABILITY research

---

STEP 8: Stakeholder Ecosystem — Who Must Be Involved?

The article emphasizes "fostering dialogue, collaboration, and action among various stakeholders"

AI STAKEHOLDER MAP
                    ┌──────────────┐
                    │  RESEARCHERS │
                    └──────┬───────┘
                           │
    ┌──────────────┐        │        ┌──────────────┐
    │  GOVERNMENT  ├────────┼────────┤   INDUSTRY   │
    │  & POLICY    │        │        │  & BUSINESS  │
    └──────────────┘        │        └──────────────┘
                           │
    ┌──────────────┐        │        ┌──────────────┐
    │  CIVIL       ├────────┼────────┤   AFFECTED   │
    │  SOCIETY     │        │        │  COMMUNITIES │
    └──────────────┘        │        └──────────────┘
                    ┌──────┴───────┐
                    │INTERNATIONAL │
                    │    BODIES    │
                    └──────────────┘

Why does diversity of voices matter?

• Different groups experience AI differently
• Power imbalances mean some voices are systematically excluded
• Good policy requires legitimacy, which requires inclusion

---

STEP 9: The Goal — Human Flourishing

The ultimate purpose stated in the article:

"Maximize AI's potential for contributing to human flourishing"

Unpacking "Human Flourishing":

HUMAN FLOURISHING IN AI CONTEXT MEANS:
├── ✅ AI enhances human capabilities (not just replaces them)
├── ✅ Benefits are distributed equitably
├── ✅ Human dignity and autonomy are preserved
├── ✅ Future generations inherit a better world
└── ✅ AI serves human values, not the reverse

This is a normative claim — it tells us AI development should be value-driven, not just efficiency-driven.

---

STEP 10: Synthesis — Putting It All Together

The Complete Mental Model:

PROBLEM: AI is transforming society at unprecedented speed
         with both massive opportunity AND serious risk

DIAGNOSIS: The field is divided between short/long-term
           thinking and lacks multidisciplinary integration

SOLUTION FRAMEWORK:
┌─────────────────────────────────────────────────────┐
│                  10 PRIORITIES                      │
│                                                     │
│  NOW (Immediate)  │  LATER (Mid)  │  LASTING (Perm) │
│  ─────────────── │  ──────────── │  ────────────── │
│  Address current  │  Build better │  Ensure AI      │
│  harms & gaps     │  governance   │  serves humanity │
│                                                     │
│  ACROSS: Research + Policy + Practice               │
│  WITH: Multidisciplinary + Diverse stakeholders     │
│  TOWARD: Human Flourishing                          │
└─────────────────────────────────────────────────────┘

---

✅ Knowledge Check — Test Yourself

Question 1: Why does the article compare AI to the Industrial Revolution?

Question 2: What is the danger of focusing ONLY on long-term AI risks?

Question 3: Name three disciplines (beyond engineering) needed for AI governance.

Question 4: What does "human flourishing" mean in the context of AI development?

Question 5: Why is the "black box" nature of neural models a governance problem?

---

📌 Key Takeaways Summary