Module Check: Trajectories and Transformative AI
This module check spans all ten lessons of Track 9, Module H5. You have studied how experts forecast AI progress and why forecasting is hard, the serious debate over whether AGI is achievable and near, the surprising difficulty of even defining AGI, what transformative AI scenarios look like and how to analyze them, how to reason honestly about timelines under deep uncertainty, what transformative AI could mean for labor, power, science, and identity, how individuals and societies can prepare, and how to think carefully about the long-term future. This check tests whether you can recall key concepts precisely, apply them to new situations, and integrate them across the module's arc.
Key Terms Recap
Flashcards — click each card to reveal the answer
Module Check Questions
A research team forecasts AI progress by surveying 2,000 ML researchers and aggregating their probability estimates for when AI will perform any task as well as a human. Which forecasting method are they using, and what is its most significant limitation?
A researcher argues that the 'general learning' definition of AGI is more scientifically rigorous than the 'task-based' definition. What is the best justification for this claim?
Which of the following is the most accurate description of what 'deep uncertainty' means in the context of AI timeline forecasting?
Governments in multiple countries are actively investing in AI for autonomous military systems, intelligence analysis, and economic competitiveness. Which transformative AI consequence does this most directly illustrate?
A forecaster builds a trajectory forecast and finds that her probability estimate for AGI by 2040 is highly sensitive to one assumption: whether current deep learning architectures can be scaled to achieve robust generalization. She reports a single probability estimate without mentioning this sensitivity. What analytical error is she making?
An AI alignment researcher argues that scalable oversight is a critical research priority. A colleague responds that near-term bias and fairness in deployed AI systems deserve higher priority. Which of the following best describes the nature of this disagreement?
Module Synthesis
Integrative Synthesis: Your Position on AI's Trajectory
- This synthesis activity asks you to integrate everything from the module into a coherent, defensible position.
- PART 1 — FORECAST SUMMARY (one paragraph)
- State your overall view of AI's trajectory: where you think AI is headed, on what timeline, and with what probability. Use the forecasting language from the module — probability distributions, explicit assumptions, acknowledged uncertainty. Do not hedge so much that you say nothing; commit to a defensible estimate and own it.
- PART 2 — DEFINITION AND SCENARIO (two paragraphs)
- State which definition of AGI you find most useful and why. Then identify which transformative AI scenario you think is most likely and describe the two or three most important developments that would confirm or disconfirm your scenario choice as evidence accumulates over the next five years.
- PART 3 — CONSEQUENCES AND PREPARATION (two paragraphs)
- Identify the two consequences of transformative AI you consider most important — one positive opportunity and one serious risk. For each, explain what you would recommend to an individual, an institution, or a government to either capture the opportunity or mitigate the risk.
- PART 4 — LONG-HORIZON REFLECTION (one paragraph)
- Reflect on your own epistemic position: What is the claim you are most uncertain about in your synthesis? What kind of evidence, if you encountered it in the next year, would most update your view? What would it take to change your mind significantly?
- PART 5 — PEER DIALOGUE
- Exchange syntheses with a partner who reached different conclusions. Identify the single most important point of disagreement. Try to determine whether it is: (a) a factual disagreement about what the evidence shows, (b) a values disagreement about what outcomes matter most, or (c) a methodological disagreement about how to reason under uncertainty. Report which type of disagreement it is and why that distinction matters for how you would try to resolve it.
- This is not a test with a correct answer. It is an assessment of your capacity to reason carefully, acknowledge uncertainty honestly, and integrate complex evidence into a defensible position — the most important intellectual skill the module was designed to build.