正在学习

Planning for an Uncertain Future

It would need a very insightful soothsayer to predict where AI is heading and how much our lives will change as we choose to ignore or prepare to embrace AI. Thinkers and scientists in Silicon Valley and other parts of the world are already saying that AGI is inevitable, and from there, super intelligence with all its unpredictable ramifications and impact might well be just a few steps away.

Ethan Mollick, author of the very insightful book Co-Intelligence—Living and Working with AI has, in the concluding chapter, ‘AI as our Future’, presented four possible scenarios for AI in the future.

1 The first is that AI is as close to being the best as you will ever see and future improvements will only be incremental. This can only happen if climate change activists, worried about the large power consumption of LLMs, or AI safety experts convincing governments to ban all AI development, have their way, which is unlikely and certainly not sustainable in a fast-changing and intensely competitive environment of innovation. There is already a deep distrust in anything that appears in media, particularly social media, and the easy fake replication of voice and images will erode the credibility of anything we see or hear as time passes. Chatting with bots will also reach its natural level of exhaustion, and in the absence of extreme innovation, AI will stay and be embraced by many but only as a complementary force to the ever-increasing capability of humans. Did we hear a sigh of relief at this scenario?

The second scenario is not very different from the first and envisages growth but at a slow pace. New methods of leveraging AI, as we have already seen when DeepSeek was launched by China, and more optimised power utilisation mechanisms will slow the pace of development, but innovation will continue. AI will add more value in the workplace, and coalitions of companies and governments will limit the harm that unregulated growth in AI in a country or an industry could have unleashed. AI will, however, play a much larger role in scientific research and discovery in all corporate settings and take a larger and larger role in our lives. Humans will remain in control.

The third scenario is more on the lines of exponential growth. AI companies using AI systems to create the next generation of AI software might create the flywheel effect on AI acceleration. The negative impact of this is the unpredictable bad outcomes that might accompany or even precede positive new discoveries, such as the creation of a new deadly nerve gas while the search for a new drug is in progress, and AI-powered dictators using the tool for ubiquitous surveillance on the unsuspecting masses. AI companions, as we already see in the movies, could become more compelling to speak with than humans, resulting in new types of social isolation. Though AIs will still not be sentient in this scenario, partnering with researchers, physicians and mathematicians might enable AI to develop and wield new powers on the rest of us. This could be just the beginning as AI could then reach a take-off point, a level of ability when it moves towards AGI and potentially an intelligence that is far greater than what humans can ever possess!

The final scenario is what Mollick calls the ‘Machine God’, where machines have attained AGI and sentience and help design AIs even smarter than themselves till superintelligence emerges. Will humans then cease to be of value to AI and become merely a nuisance or a potential source of molecules? As Geoffrey Hinton, one of the godfathers of AI who left the field in 2023, has said, ‘It’s quite conceivable that humanity is just a passing phase in the evolution of intelligence.’

Having painted this awful scenario, which is seen as credible by some and ignored by many, let us build our strategy on the assumption that doomsday is not near or inevitable and a carefully planned dual intelligence strategy, with the ‘human in the loop’ always in control, is most likely to prevail if articulated, designed and implemented by all of us.

练习题

Which option best describes Ethan Mollick’s overall framing of AI’s future in the source material?

A. AI has only one likely future: immediate replacement of all human work.

B. AI’s future can be considered through four possible scenarios ranging from incremental progress to superintelligence.

C. AI will certainly stop developing because governments will permanently ban it.

D. AI will remain useful only for entertainment and will not affect workplaces or research.

In Scenario 1, why does AI remain mainly a complementary force rather than becoming radically transformative?

A. Because AI becomes sentient but chooses not to interfere with humans.

B. Because future improvements are mostly incremental, while fake media distrust and bot fatigue limit excitement around AI.

C. Because AI is used only in scientific research and never in daily life.

D. Because AI companies stop using large language models entirely.

Which statements accurately describe Scenario 2 in the passage? Select all correct answers.

A. AI grows, but at a slow pace.

B. Humans remain in control.

C. Coalitions of companies and governments limit harms from unregulated AI growth.

D. AI plays a larger role in workplace value, scientific research, and corporate discovery.

E. Machines attain sentience and design smarter AIs until superintelligence emerges.

Scenario 3 suggests that AI companies using AI systems to create the next generation of AI software could produce a flywheel effect that accelerates AI development.

In Scenario 3, the passage presents only positive outcomes from exponential AI growth, such as faster drug discovery, and does not mention major risks.

Mollick’s final scenario, in which machines attain AGI and sentience and help design even smarter AIs until superintelligence emerges, is called the ‘___’ scenario.

Explain how Scenario 4 changes the status of humans in the imagined evolution of intelligence.

Why does the passage recommend a dual intelligence strategy with humans in control, rather than assuming either doomsday or purely unregulated AI development?

Which pairings of AI future scenario and organizational response are consistent with the passage and the prior knowledge points? Select all correct answers.

A. If AI grows slowly but adds workplace and research value, organizations can use AI analytics for better decision support while humans remain final decision makers.

B. If AI creates risks such as surveillance and manipulation, leaders should consider ethical implications and avoid uncontrolled deployment.

C. If a dual intelligence strategy is adopted, organizations may need to redesign structures, roles, reporting relationships, and task allocation.

D. If AI reaches the ‘Machine God’ scenario, the passage says human leadership skills such as context and emotional judgment become automatically unnecessary and harmless to abandon.

E. If Scenario 1 occurs, AI will definitely disappear from society and workplaces.

Across the four scenarios, the passage contrasts futures in which humans remain in control with futures in which AGI, sentience, or superintelligence could reduce human centrality.

Which option best combines Mollick's preferred future strategy with the earlier idea of organisational design for dual intelligence?

A. Let

A I

make all final decisions because Scenario

4

shows machines will soon surpass humans in every role.

B. Build a dual intelligence system with the human in the loop, while redesigning structures and processes to use

A I

responsibly.

C. Avoid redesigning teams because Scenario

1

implies that no workplace change is needed at all.

D. Replace human leadership with

A I

companions because Scenario

3

suggests humans will prefer talking to machines.

Which statements are consistent with both Mollick's scenarios and the earlier discussion of responsible leadership with $A I$ ? Select all that apply.

A. In a slow-growth future, humans may remain in control even as

A I

adds workplace and research value.

B. Because

A I

can augment leadership, it should always take the lead in emotionally sensitive decisions.

C. Exponential

A I

growth increases the need to consider ethical implications such as surveillance or harmful misuse.

D. A wise deployment approach is to choose

A I

application areas carefully rather than allow unsupervised action everywhere.

E. If fake media increases distrust, responsible human oversight becomes less important.

If Scenario $3$ brings exponential growth and $A I$ systems begin helping create better $A I$ , then the need for humans to focus on distinctively human capabilities such as emotional intelligence and contextual judgment becomes even more important.

In response to uncertain futures ranging from incremental progress to the $‘Machine God’$ scenario, the text recommends a dual intelligence approach in which the human remains ___, while organisations also develop frameworks for delegating goals to technology.

登录后解锁笔记、知识点解析、AI 问答

立即登录

Planning for an Uncertain Future