# mollick-confronting-impossible-futures-agi-2024-07-22

## Veille

Strategic Planning for AI's and AGI's Impossible Futures - One Useful Thing - Ethan Mollick

## Titre Article

Confronting Impossible Futures

## Date

2024-07-22

## URL

https://www.oneusefulthing.org/p/confronting-impossible-futures

## Keywords

AGI, Artificial General Intelligence, strategic planning, LLM, AI adoption, future scenarios, organizational transformation, technological uncertainty, Moore's Law, disruption

## Authors

Ethan Mollick, Professeur à la Wharton School, University of Pennsylvania

## Ton

**Profile**: Academic-forward-looking, first-person thought leader, analytical-provocative register, intermediate level.

**Description**: Mollick adopts an intellectual essayist's tone combining empirical data (2047 AGI surveys) with strategic reflection. The title "Confronting Impossible Futures" signals the willingness to address uncomfortable scenarios. Accessible language avoids excessive technical jargon while covering the complex implications of AGI. The methodical argumentative structure identifies obstacles (dystopian discussions, missing documentation, fragmentation of progress) then proposes frameworks for action. The tone of measured urgency avoids both dismissive skepticism and panicked alarmism. Typical of thoughtful essays on Substack (Stratechery, Margins style) that challenge received ideas and force readers to confront difficult questions about technological trajectory.

## Pense-betes

- Organizations plan on a 10-year horizon but ignore continued AI improvement in their strategies
- Even without new breakthroughs, current AI will disrupt organizations for years
- 2023 survey: AGI expected on average by 2047, with a 10% probability before 2027
- Three obstacles to action: paralyzing dystopian discussions, lack of clear documentation, fragmentary nature of progress
- Moore's Law as an example of a self-fulfilling prophecy: expectations become motivating targets
- Recommendation: multi-scenario planning (plateau, linear growth, exponential growth, AGI)
- Uncertainty about the timeline does not justify inaction
- Modern AI tools can facilitate scenario planning
- Impressive AI features hidden behind non-intuitive interfaces
- Critics easily downplay real capabilities by focusing on isolated failures

## RésuméDe400mots

Ethan Mollick examines a troubling paradox: despite planning horizons spanning a decade, few organizations seriously factor the possibility of continued AI improvement into their strategies. This omission is all the more surprising given that experts disagree on AI's future—some predict imminent exponential growth, others claim large language models have hit their limits.

The author emphasizes that even without further technological breakthroughs, AI would significantly disrupt organizations for years to come. AI systems require better integration, while a complete halt to development seems unlikely. Current impacts—professional automation, targeted phishing campaigns, pedagogical shifts—demand immediate attention.

Mollick draws an instructive parallel with Moore's Law, showing how technological expectations become motivating targets, creating self-fulfilling prophecies. AI leaders announce ambitious timelines (2027, five years), though opinions diverge. A 2023 survey of computer scientists placed the average expected date for AGI at 2047, with a ten percent probability before 2027.

Three main obstacles explain organizational inaction. First, dystopian discussions about superintelligence seem unplannable. Second, the striking absence of clear documentation for non-specialists perpetuates ignorance of current capabilities. Impressive features remain hidden behind non-intuitive interfaces or obscure tips.

Third, the fragmentary nature of AI progress—excellent in some domains, disappointing in others—makes it easy to downplay real capabilities. Mollick illustrates this tendency by citing critics who, while acknowledging remarkable achievements, deny their transformative significance.

The author proposes abandoning this paralysis by uncertainty. Rather than choosing a single vision of the future, organizations should plan for several possible scenarios: technical plateau, linear growth, exponential growth, or AGI. He recommends scenario planning, a process he demonstrates is facilitated by modern AI tools.

In conclusion, Mollick argues for acknowledging inevitable change and actively taking control of its direction. Uncertainty about the exact timeline does not justify inaction. Organizations must begin now to consider radically different futures and adapt their strategies accordingly.

## GrapheDeConnaissance

- Ethan Mollick —travaille_chez→ Wharton School (ORGANISATION, 0.99)
- Ethan Mollick —affirme_que→ les organisations doivent planifier plusieurs scénarios AGI (AFFIRMATION, 0.95)
- Ethan Mollick —affirme_que→ les organisations ignorent l'amélioration continue de l'IA dans leur planification (AFFIRMATION, 0.92)
- AGI —est_instance_de→ objectif des grands laboratoires IA (CONCEPT, 0.97)
- loi de Moore —est_instance_de→ prophétie autoréalisatrice technologique (CONCEPT, 0.9)
- Ethan Mollick —recommande→ planification multi-scénarios (METHODOLOGIE, 0.95)
- Arvind Narayanan —s_oppose_à→ AGI (CONCEPT, 0.85)
- sondage informaticiens 2023 —mesure→ AGI prévue en moyenne pour 2047, 10% de probabilité avant 2027 (MESURE, 0.92)
- Claude 3.5 —soutient→ capacités IA actuelles impressionnantes (CONCEPT, 0.88)
- GPT-4o —permet→ planification multi-scénarios (METHODOLOGIE, 0.83)
- Ethan Mollick —publie→ Co-Intelligence (DOCUMENT, 0.95)

---
Canonical: https://www.thekb.eu/en/fiches/mollick-confronting-impossible-futures-agi-2024-07-22/