🚀 Content in 3 sentences
- AI will transform many areas of our lives until 2041. There are many applications for AI we can’t even imagine yet, some of which go to the very bottom of human existence.
- AI is not merely a technological challenge – it’s also a social challenge. From job displacement to the end of money, anything could happen.
- AI is not bad or good. It is up to us what we use it for.
🎨 Impressions
I love the combination of fiction and non-fiction. The short stories let the reader travel into the year 2041 and the explanations provide the necessary technological analysis. I’m curious to read more books from the writers. Truly an inspiring work.
☘️ How the book has helped me
The book has helped me expand my horizon about AI and the future in general. I like the optimistic view of the book. I feel like AI can make our lives better.
📒 Summary & Notes
Introduction by Kai-Fu Lee: The real Story of AI
While the future with AI is often portrayed as dystopian in contemporary art (movies, novels, etc.), Lee has a more optimistic outlook: He believes that the AI-revolution creates wealth, relieves us humans and leads to an overall better world. Lee believes to be able to predict the future as he has a long history of AI-related jobs and is an insider to the industry. He wants to look beyond the near future and see what the world will look like in twenty years time. For the book he worked together with his former colleague at Google, who is now a science-fiction writer, Chen Qiufan. By adding fiction to his science, he hopes to create scientific fiction rather than science fiction. He wants his readers to picture the future for themselves.
The book consists of ten stories total. The first seven are focused on AI applications in different industries and the societal and ethical implications. The last three stories focus on the societal and geopolitical implications of AI. Finally, there are four different outlooks on 2041 to underline, that the future has not yet been written…
Introduction by Chen Quifan: How we can learn to stop worrying and learn to embrace the future with imagination
For Quifan, science fiction is a magical portal to the future: It frees us from our current way of thinking about our lives and enables us to reflect on our surroundings from the outside. In his mind, science fiction is not providing answers but raising questions.
Chapter One: The Golden Elephant
It’s the year 2041 in Mumbai, India. Nayana, a teenager has a crush on Sahej, a boy from her class. They want to get to know each other, but an AI that controls the premiums for insurances does not want the two adolescents to be together. As he is from a family with worse circumstances, and the his therefore more likely to get sick or endangered, the premium for Nayana’s family increases if they spend time together.
Deep Learning, Big Data, Internet/Finance Applications, AI externalities
Within AI, there is a field called machine learning. One of the ways to do machine learning is called deep learning. Deep learning works best without many external rules. The machine is “fed” data and the according solution and trains numerous layers between input and output. With huge numbers of data, the AI gets better over time. Deep learning algorithms need three things to function: Massive amounts of data, a narrow domain and a concrete function to optimize. These algorithms outdo the human brain in tasks like quantitative optimizing (i.e. recognizing a face in a million faces) or customizing (i.e. showing the most relevant ad to a user) but fail to perform at cross-disciplinary tasks as well as anything creative.
This is one of the main flaws of deep leaning AI. As they can only be optimized for one goal (i.e. keeping you watching YouTube videos to generate more ad revenue), they automatically leave other things uncontrolled (i.e. the user’s (mental) wellbeing). The second major flaw is the one described in the short story: The AI is only as good as the data it is provided. If the data suggests that people from a certain group are more likely to be sick or robbed, the AI will “keep that in mind” and existing discrimination might be implemented on accident.
Chapter Two: Gods Behind the Masks
It’s the year 2041 in Lagos, Nigeria. Amaka, a deep-fake expert is tasked to fake a video of a politician by a shady customer. He has no choice but to do it, even though his video could have dire consequences for the future of Nigeria.
Computer vision, convolutional neural networks, deepfakes, generative adversarial networks (GANs), biometrics, AI security
While understanding what happens based on vision comes easy to us humans, computers have a much harder time. This field (”teaching computers to see”) is called computer vision.
One way to use AI for computer vison are convolutional neural networks, or CNNs for short. CNNs are inspired by how the human brain processes images. Essentially, CNNs consist of multiple layers that analyze the image on different levels. When analyzing a picture of a leopard, for instance, on the lowest level, the AI may see dots on the fur, the next level might see the pattern of the fur, as well as ears, eyes and so on and on the highest level, the AI can tell whether it thinks that there is a leopard on the photo or not.
The deepfake technology does not analyze, but create a fake version of an existing person. This is a huge ethical and social threat.
Chapter Three: Twin Sparrows
It’s the year 2041, somewhere in Korea. Silver and Golden Sparrow, twin orphans each get an AI-companion to support their development. Silver Sparrow, a shy boy with Asperger’s Syndrome becomes a successful artist while his brother, Golden Sparrow struggles with the stress that is put upon him.
Natural Language Processing, Self-Supervised Training, GPT-3, AGI and consciousness, AI education
The ability for computers to process human speech is called Natural Language Processing, or NLP for short. Supervised leaning means, that the AI is given a pair of answer and solution (i.e. a picture of a cat and the label “cat”). This approach works fairly well for documenting what was said (speech recognition), but not for understanding what was meant, as this is to broad of a field to train for with supervised learning. With unsupervised leaning, the AI teaches itself a language from scratch, learning not just the content but also the context. The most prominent example is GTP-3, published by Open-AI in 2020. As of today, the AI doesn’t do much more than rearrange what it knows when asked a question. It is not truly intelligent. Still, it’s only version 3 of many more to come.
One of the areas that haven’t changed much in the last centuries is school. This might change when virtual tutors for each student become a thing.
Chapter Four: Contactless love
It’s the year 2041 in Shanghai, China. Since 2019 COVID-19 has become a part of the daily life on earth, with new mutations and strains evolving regularly. Chen Nan isolates herself from the rest of the world in order to stay away from potentially infected humans. She meets Garcia, a Brazilian man in an online video game but is still hesitant to meet him in person. After Garcia comes to Shanghai to surprise her, he gets infected with a new COVID variant, which makes Chen Nan leaver her flat for the first time in years in order to see the love of her life just once in person. A trip through Shanghai in 2041 starts.
AI healthcare, alphafold, robotic applications, COVID automation acceleration
With massive amounts of data available, AI will assist medical personel with accurate diagnosis and treatment options. Also, tasks like developing drugs which often require a 3D-model of the protein can be done using AI to accelerate the process. After building a 3D-model of a protein, the AI can help find a target to attack. Then, it can run through existing medication to try to find a drug that could prove useful or it can help develop a new drug. In clinical trials, a part of the process can be done in-silico, digitally, to speed up testing of a new drug. Aditionally, AI will enable patient-specific drugs that unlike the one-drug-fits-all-approach can treat deseases more effectively.
A field that we see evolving rapidly, even today, is robotics. By 2041, many hosehold chores will be done by robots. The same goes for factories and warehouses.
Chapter Five: My Haunting Idol
It’s the year 2041 in Tokyo, Japan. Aiko, a Japanese writer goes on an immersive AR-adventure to find the reason for the death of her idol, the j-pop artist Hiroshi-kun.
virtual reality (VR), augmented reality (AR), mixed reality (MR), computer-brain-interfaces (CBI), ethical and societal issues
Immersive simulation technologies produce an alternative reality – the so called X-reality or XR for short. There are three versions of XR: Virtual reality, or VR for short creates an entirely synthetical virtual environment. AR, which stands for augmented reality adds on to the existing real environment the user is in, for example by placing a virtual dinosaur in your real living room. Mixed reality, or MR, combines AR and VR: In MR, the product takes the existing environment, but does not just add on to this environment. Rather it builds a new environment by a complete decomposition and interpretation based on it. XR is more than just the sum of AR and the real world. To achieve MR, the lens needs to understand its environment, not just see it. By 2041, smart lenses in the form of glasses or even contact lenses will likely be a reality. Through haptic gloves or even suits, you will be able to feel the virtual world as if it were real.
These technologies will also make us face new obstacles. One is privacy. With XR in our lives, a lens can literally see the world through our eyes. Also, capturing every single moment of a live makes us digitally immortal, as all memories are stored somewhere.
Chapter Six: The holy driver
It’s the yearn 2041 in Colombo, Sri Lanka. Most cars in China are now driverless on designated streets. However, in special cases, like a bomb threat or a natural disaster, real drivers take over, controlling the cars from a safe simulator somewhere around the world. Chamal, a Sri Lankan boy is drafted as a driver and faces challenges in a strange place between reality and the virtual world.
Autonomous vehicles, Full Autonomy and Smart Cities, Ethical and Social Issues
Autonomous driving requires a wide range of tasks. From perception over navigation all the way to decision making. Thus it is a hard thing to achieve.
Autonomous driving is split in to five levels:
- L0: no automation
- L1 (”hands on”): AI can do one specific task with the driver tuning it on, such as holding the lane.
- L2 (”hands off”): AI can do multiple tasks (steering, accelerating and breaking), but the human must still be able to take over at all times.
- L3 (”eyes off”): AI can drive the car, but might sometimes ask the driver to take over in difficult situations. ← We are here
- L4 (”mind off”): AI can take over driving completely, but only in a known environment such as Highways and City streets, that have been mapped before.
- L5 (”steering wheel optional”): AI can drive the care safely everywhere, the human taking over is entirely optional.
L5 is incredibly hard to achieve. One way to shorten the way to L5 AV is to simplify the task by building separate AV-roads with sensors integrated into the road and without obstacles that produce unpredictable situation such as cyclists or pedestrians.
L5 AVs will have huge consequences: The idea of owning a car will be uneconomic because ridesharing will be so much cheaper. The price of driving services like Uber will only be around 25% of the current price as the costly driver is no longer required. This also means that tons of people around the world will be our of a job. L5 autonomous driving also poses ethical questions: Should an algorithm decide about life and death?
Chapter Seven: Quantum Genocide
It’s the year 2041 in The Hague, Netherlands. A scientist has gone insane over the loss of his wife and son in an climate change related tragedy. To seek revenge, he wants to send humanity back to a pre-digital time by disabling the world’s power grid and the internet. To achieve his goal, he uses autonomous weapons among other terrifying technologies.
quantum computers, bitcoin security, autonomous weapons and existential threat
While conventional computers use bits (on or off; 1 or 0), quantum computers use quantum particles such as photons or electrons, these form the qubits (quantum bits) and can hold more than 2 states at once. While this allows for huge synchronous data processing, it is also very sensitive to disturbances like changing temperature, radiation or vibrations. One thing that could be accomplished with a 4,000 qubit QC is breaking RSA – the algorithm that encrypts bitcoin and most of the internet.
Autonomous weapons are slowly becoming a reality despite posing almost unanswerable ethical questions.
Chapter Eight: The Job Saviour
It’s the year 2041 in San Francisco, United States. Millions of workers become obsolete because of automation. It’s not just a couple of people losing their jobs, it’s millions losing their occupation and with it their hope, dreams and sense of purpose. As a result, many resort to drugs, gambling, the virtual reality or even suicide. Synchia, a company determined to restore the lives of the ones who have lost their jobs, is now trying to negotiate a deal for one of the largest American construction firms.
AI job displacement, universal basic income (UBI), what AI cannot do, 3RS as a solution to displacement.
AI will gradually be able to perform many of today’s professions. It’s not going to happen overnight, not just blue collar but also white collar jobs are at risk.
With AI replacing millions of workers, the invisible hand of the market will no longer work. Higher productivity at lower costs will be a dangerous combination as it will widen the gap between poor and rich to an unprecedented level. In addition to the economic dimension, there is also a emotional dimension to be considered: The loss of a job is more than merely losing a source of income, it’s loosing one’s purpose in life.
One idea to counter this problem is universal basic income (UBI). As the name suggest, every human being gets a universal amount of money every month to spend however they want.
There is one important question for people entering the job market: What can AI not do? There are three main areas that have shown to be difficult for AI: Empathy, creativity and dexterity. So what about the people who lose their jobs to AI? The three Rs are part of the solution: Relearn, recalibrate and renaissance. Relearning in order to perform a more sustainable profession is the most straight-forward answer. We need to recalibrate our brains to embrace an AI powered future. Many jobs will be easier or less repetitive thanks to AI. Many other jobs we can’t even imagine now will be available on the job market. Finally, AI may lead to a new renaissance: Taking the focus away from careers and putting it on creativity, empathy and the human experience in general.
The job market in general will evolve in another way: The simpler, more repetitive positions are the first to be automated. These jobs, however, are usually the entry-level jobs for young workers and equip them with the necessary expertise to perform well at the more complex jobs. Therefore, “Made-up-jobs” and an extended time for studies and training have to be expected.
Chapter Nine: Isle of Happiness
It’s the year 2041 in Doha, Qatar. Russian tech-billionaire is on a quest to find true happiness on an island designed by the crown-prince of Quatar. His journey goes to the bottom of questions about what makes us humans truly happy.
AI and happiness, general data protection regulation (GDPR), personal data, privacy computing using federated learning and trusted execution environment (TEE)
Creating happiness is a fundamentally challenging task. The first problem is happiness itself: What is happiness, where does it come from and how can it be measured? To find answers to this problem through AI requires massive amounts of data and it requires the most personal data about ourselves. To better understand happiness, AI could analyze our micro and macro expressions, our blood pressure and our eyes.
Not just with happiness is data collection an integral part of AI technology. How can we manage, store and process sensitive data to enhance our lives without sacrificing our privacy? One option would be to create a central, personal and protected personal data hub, that will or will not provide the data to for-profit companies, if necessary. With future breakthroughs in encryption and privacy computing, a decentralized approach could be possible by 2041.
Chapter Ten: Dreaming of Plentitude
It’s the year 2041 near Melbourne, Australia. Advances in AI and technology have brought down the cost of living to nearly zero, making working to “make a living” unnecessary. The Australian government tries to counteract joblessness by introducing Moola, a virtual currency that rewards voluntary work. Keira, a young aborigine, takes care of Joanna, a former marine biologist who saved the Great Barrier Reef. In the process, she discovers the problems with Moola and decides to take matters into her own hands.
Plentitude, new economic models, the future of money, singularity
Plentitude, as Kai-Fu Lee calls the new chapter of humanity when all basic needs are met at nearly no costs, will pose new challenges to humanity. Plentitude challenges today’s purpose of life for many people: Working to make a living.
There are many new technologies in the making: Wind and solar in combination with batteries will provide the energy needed to power the world, bringing the cost down to about a quarter of today’s price per kWh. Advancements in synthetic biology will enable non-toxic fertilizer and plastic-eating bacteria. Automation, powered through AI will bring down the cost of goods to slightly above material costs.
This will make our economic models redundant in many aspects. Scarcity has been one of the driving factors in human development. Where demand exceeded supply, innovation occurred. The new “currency” would be status and honor, achieved through voluntary work in science, social facilities or art. In the story, the Australian government tries to reward it’s inhabitants with a new currency: Moola. It is intended as an incentive to do good. The idea backfires as people try to cheat the system for more social status.
The transformation from our current to the post-scarcity model will be a challenge, as corporations will try to keep the prices high through artificial scarcity.
Final words
AI is – like any other technology – neither inherently good, nor is it inherently bad. The question is what we as individuals and as humans make of it.