By Kai-Fu Lee, Chen Qiufan
Stories and analysis imagining AI's future
Subject
Themes
Ten fictional stories set across the globe in 2041 illustrate the realistic trajectory of artificial intelligence, each paired with technical analysis grounding the narrative in existing science. The journey begins with deep learning fundamentals: in Mumbai, a teenager named Nayana discovers that her family's AI insurance platform infers her classmate Sahej's Dalit caste from behavioural data patterns, using objective function maximisation to discourage their relationship. The algorithm does not know about caste directly; it simply optimises for lower premiums, and the social discrimination embedded in historical data does the rest.
Ten fictional stories set across the globe in 2041 illustrate the realistic trajectory of artificial intelligence, each paired with technical analysis grounding the narrative in existing science. The journey begins with deep learning fundamentals: in Mumbai, a teenager named Nayana discovers that…
The pairing of speculative fiction with technical analysis places this work within a growing tradition of hybrid literature that emerged in the late 2010s and early 2020s, as public anxiety about artificial intelligence intensified. Earlier popular treatments of AI tended to fall into one of two…
Every AI system operates by optimising an objective function, a mathematical goal that guides its behaviour. When that goal is narrow and singular, the AI pursues it with relentless focus, ignoring everything else. This creates what might be called the objective function problem: the gap between…
AI excels at quantitative optimisation, pattern detection in massive datasets, and customisation for individual users. Deep learning can pick a face from a million, recommend products tailored to individual behaviour, and diagnose lung cancer more accurately than most radiologists. Yet three…
Data is both the fuel powering AI and the mirror reflecting society's existing prejudices. When AI systems are trained on data generated by biased societies, they do not merely reproduce that bias; they codify it, scale it, and render it invisible behind the authority of mathematical…
A recurring pattern emerges across AI technologies: every creative capability generates an equal and opposite detection challenge, producing arms races where computing power determines the winner. Deepfakes and their detectors, autonomous weapons and their defences, quantum encryption crackers and…
Deep learning and its descendants form the technological backbone of every scenario imagined for the year 2041. At its core, deep learning is a method by which artificial neural networks, loosely modelled on the brain's layered architecture of neurons, teach themselves to perform tasks by processing…
Artificial intelligence spent decades confined to academia before deep learning transformed it into the world's most commercially potent technology. The 2016 AlphaGo victory over human Go champion Lee Sedol marked a turning point, demonstrating that neural networks trained on massive datasets could…
Computer vision teaches machines to see through convolutional neural networks (CNNs), inspired by the human visual cortex. These networks use hierarchical filters that detect increasingly complex features, from basic lines to complete objects. When combined with generative adversarial networks…
Healthcare stands at the cusp of transformation through AI-driven digitisation. DeepMind's AlphaFold 2 solved the fifty-year protein folding problem, enabling rapid determination of the 3D structures essential for drug discovery. AI accelerates the four-step process from deriving protein sequences…
Autonomous vehicles will mature through six levels of capability, from basic alerts to full self-driving without any human involvement. Level 5 remains elusive because AI must handle immense variability in road conditions and rare "long tail" scenarios for which training data is insufficient. Two…
Jobs fall on a two-dimensional spectrum where routine, asocial roles face complete displacement while creative and social roles remain safe. Robotic process automation watches workers, learns their routines, and gradually takes over tasks until entire departments shrink dramatically. The three…
Three converging revolutions drive the world toward plenitude, a state where most goods can be produced for next to nothing and work becomes optional. Solar energy costs dropped 82 percent from 2010 to 2020, wind energy fell 46 percent, and lithium-ion battery storage plummeted 87 percent. With a $2…
By Kai-Fu Lee, Chen Qiufan
Subject
Themes
Ten fictional stories set across the globe in 2041 illustrate the realistic trajectory of artificial intelligence, each paired with technical analysis grounding the narrative in existing science. The journey begins with deep learning fundamentals: in Mumbai, a teenager named Nayana discovers that her family's AI insurance platform infers her classmate Sahej's Dalit caste from behavioural data patterns, using objective function maximisation to discourage their relationship. The algorithm does not know about caste directly; it simply optimises for lower premiums, and the social discrimination embedded in historical data does the rest.