A Day In the Life in 2045
Advik, London, 28
Advik woke up to the pleasant melody of his phone alarm, and the not-so-pleasant view of his cubicle. Like most Londoners, he lived in a small apartment: one room and bathroom. One of many, in a skyscraper housing thousands.
He retracted the bed onto the wall, and put on his VR headset. His friend Nihal had persuaded him to sign up to an AI-designed gym app. Every morning they hang out together virtually, and do some reps while bantering. He enjoyed it greatly.
“We should have gone, Advik.” Nihal said “Out of the city. Back to India. We’d have a farm. To work your scrawny ass off!”
Advik took a moment to catch his breath. Good thing sweat was not shown on his avatar; he would have been embarrassed. “I don’t know Nihal. My mother lives here. And I don’t speak Hindi anymore.”
“You can’t fool me. You are just lusting after that neighbour of yours, hmm?. Have you talked to her yet?” Advik didn’t respond, and pretended to hit Nihal’s avatar. They laughed.
Breakfast was delivered to his door by a bumbling drone. Vegetable bacon, omelette made with vat-grown eggs and various fruits. Advik eagerly devoured it, put on his uniform and made his way to work. He commuted by tube – of all the changes that CAIS had brought to the world, clean air on the wagons was the one he was most grateful for.
Advik worked as a steward at Mr and Ms Muller’s house. He did not understand why the Mullers did not replace him with a robot, but he was not going to complain. His shift was short and his wages high. On a typical day he would work four hours, mostly cleaning and waitering for Ms Muller. She was kind to him, and he worked diligently, while daydreaming about the rest of the day.
Advik didn’t dislike his new life. He used to be a trucker – his job had consisted in overseeing the automatic pilot, taking over occasionally during hail storms. But the job had become obsolete now that CAIS agents could do it just as well. He was given severance pay, and offered a new house in London. Close to the jobs, and to his mother. Finding the job had been easy – he asked the CAIS assistant on his phone, who had helped him write a CV and send it around.
On the tube back, Advik ran into his neighbour, and his heart skipped a beat. She was pretty, and dressed in an intricate pattern of flowers – probably AI-designed. Advik wondered what she was reading. She raised her view from her ebook, and their eyes met for a second. He quickly lowered his head, embarrassed of being caught looking, and fumbled with his hands. Nihal would have laughed at him.
The train reached Advik’s station. She got up, straightened her dress and left the wagon. Advik followed. He felt on edge. If only he was more courageous…
“Hum. Hi.” Her neighbour turned around. Had Advik said that? Why had he spoken? What was he thinking? But now it was done, and it would be more awkward to stop. “I’m Advik. I think I am your neighbour? I’ve seen you around a couple of times.”
Her neighbour smiled. “Yes, I recognize you. I’m Lily.” She turned around and kept walking leisurely. Advik stepped to her side, and tried to continue the conversation. “What were you reading? If you don’t mind me asking.” Lily did not rebuff the conversation, but kept her reply short. “Oh, I was not reading. I was chatting with my fiancé.” She showed a man on her tablet. She was video calling him? No, it had the distinctive CAIS logo ; a simulation.
Advik blushed, and excused himself, walking quickly until he was out of sight. Stupid, stupid, stupid. Once he was back in the safety of his apartment, he allowed himself to cry.
He would have liked to afford a human therapist, like Ms Muller. But therapists were few and their fees high – he wouldn’t be able to afford weekly sessions. Instead, Advik booted up a therapy app from his VR headset.
The AI therapist was helpful, and guided him through some exercises to take his mind off his heartache. “Writing about how you feel might be helpful”, the AI suggested.
Advik had never been much of a writer, but he saw no harm in trying. He opened a text processor and tried to describe what he felt, the heaviness in his chest. The sentence was not quite right, but the autocorrect gave him some options on how to rewrite it. One of the options felt right – a window into the space he was involuntarily inhabiting.
He built on that foundation. The more he wrote, the more he realised that there were other things he wanted to talk about. Things that had been bothering him for a while. The AI therapist was a good listener, and encouraged him to keep going.
He wrote about his change of jobs. The sense of loss he endured after being kindly informed that he was now obsolete as a trucker – and that he didn’t have to work anymore, if he didn’t want to.
He hadn’t given himself time to figure out how he felt about one of his childhood friends, who had chosen to upload himself to the grid. Advik still saw him in VR and social media, but had been puzzled by his decision to abandon his body. He wrote about that too.
Once he was done, he read the paragraphs he had written. He felt relief, hearing his own ails described so clearly. And, surprisingly, he found pride within himself. Advik didn’t know he could write something so beautiful.
He sent the text to a few friends, including Nihal, and asked for their opinion. Nihal was enthusiastic. “This is great, man! You should submit it to a magazine or something!” Advik blushed, but felt a warm glow inside. Maybe he could write more.
Pamela, New York, 43
Moving out of New York was the best decision Pamela had ever made.
She had been sceptical when Maria had told her of her intentions. It sounded crazy, to go live in the middle of nowhere with their families. “Like a commune?” “No. Well, maybe. Anywhere but here.”. She hadn’t agreed outright, but she had not opposed it either. When Maria told her her family was moving next week, Pamela and her family had followed.
Now, after six months of living in the sleepy village of Mattapoisett, Pamela was grateful for the slower pace of life. She loved spending lazy Sundays kayaking on the river with her husband and daughter, and watching the sun set over the harbour.
Every day felt like a gift, and she was happy to be surrounded by the people she loved most.
Mattapoisett was not, geographically speaking, located in Mattapoisett, Massachusetts. The place Maria had chosen to found their village was in fact a clearing in upstate New York, on the edge of a vast forest and next to a deep river. Maria’s grandmother had suggested the name jokingly, an obscure reference to an old book that was lost on Pamela. But the name had a certain ring to it, and it stuck.
She spent the first six days on a cottage with her family, with Maria’s family on a similar cottage right next to it.
The CAIS agents queried each of them on how they wanted their village to look like. Pamela had not known how to answer. She was too used to working with what she was given, to not complain that she did not have more. She didn’t know how and didn’t dare ask for more. “A garden would be nice.” She finally said, when pressed.
Maria’s family, who had been thinking about this for longer, had a more defined plan. They wanted a traditional New England fishing village, with white clapboard houses, a town green, and a working harbour. Pamela’s family had been content to let Maria’s family take the lead, but after the first week there, they realised that they wanted to contribute too.
Pamela’s husband, an architect, designed many of the houses in the village, using traditional construction techniques and local materials. Pamela and her mother-in-law planted the gardens, and Pamela’s daughter helped to care for them. The CAIS agents did the rest, erecting walls and rerouting the river to fit their plans for the village.
Two weeks after the move, the most important constructions were finished. They continued building houses, to accomodate newcomers. Some friends and extended family, initially sceptical, had decided to move in after visiting. The village grew like a tree, sprouting from the seed they had planted next to the river.
The villagers didn’t need to work, strictly speaking. They took on projects anyway, to keep their minds and bodies active. Teaching the young ones was a popular activity that Pamela’s daughter enjoyed. Pamela started a small weaving business, making quilts and blankets with the help of the other women in the village.
Life in Mattapoisett felt rustic, but technology was embedded in the apparently old-school commune. Solar panels kept their houses alight and warm. Everyone had smartphones and VR headsets – though they learned that they often preferred one another’s company over the screens. And they had Drexler printers in their houses, ready to print any toy they fancied, any food they craved or any medicine they needed.
They tried not to rely on the CAIS network. Sometimes it was inevitable. The youngest of Maria’s children, a toddler, had a bad fall and needed urgent medical care. A CAIS doctor rushed to him when summoned, and immediately regrew the bone. But for the most part, they tried to keep life as normal as possible, even with the futuristic amenities.
Every day, the villagers would gather on the town green for a communal dinner. They would share stories and laughter, and sometimes debate and argue about the goings-on in the village. Pamela had never felt so connected to a community. She was grateful for the life Maria had brought to her.
Which made it twice as hard when Maria’s family decided to leave six months later, to start a new village in Arizona.
“We are living a wonderful life here.” Maria had announced, in their evening communal dinner. “And others deserve the same. People are huddled up in San Francisco, Los Angeles, New York. Working for the rich and wandering aimlessly after CAIS. If we can help them take the first step, it will make all the difference.”
Pamela was devastated. She didn’t want her friends to leave. They were taking a piece of her home with them. But she knew that they were right.
That night, she kayaked down the river to Maria’s house, and sat on the porch, while they talked about the day’s events. They watched the stars, and Pamela promised that she would visit Maria’s new village one day.
PS: Both short stories have been co-written by a GPT-3 text-davinci-001 language model
Answers to prompts
Q. AGI has existed for at least five years but the world is not dystopian and humans are still alive! Given the risks of very high-powered AI systems, how has your world ensured that AGI has at least so far remained safe and controlled?
A. The individual AI systems that make up the AGI have been designed to be highly specialised and transparent in their operation. Some of these narrow systems have been trained to detect other unreliably aligned AIs, and will order AIs at risk to become unaligned to shut down. Humans mostly do not participate in this process, though the alignment AGIs will occasionally query them to ensure they correctly understand their interests.
The feats of technical alignment that led to this situation is mostly the work of a conglomerate of western (and Indian) AI companies. They built on fundamental advances on the theory of AI Alignment that were pioneered in the decade of the 2030s.
Q. The dynamics of an AI-filled world may depend a lot on how AI capability is distributed. In your world, is there one AI system that is substantially more powerful than all others, or a few such systems, or are there many top-tier AI systems of comparable capability? Or something else?
A. The existing AGIs consist of a complex economy of superintelligent but narrow AIs called a CAIS (Comprehensive AI Service). Each of them is trained following a common learning paradigm. Put together, they constitute a fully general artificial intelligence with many redundant parts. Initially the CAIS were differentiated into five clusters (NATO, Sino-Russian, African-middle-eastern, India, Southamerica), though the frontiers between the clusters are progressively blurring.
Q. How has your world avoided major arms races and wars, regarding AI/AGI or otherwise?
A. AI was initially led by Western companies. Advances were publicly released, though models were kept private and their services sold through APIs. China built on these disclosures to improve surveillance, autonomous weapons and logistics. India’s AI scene took off with nanotech-built computing infrastructure, thanks to their access to talent.
CAIS technology was developed by a conglomerate of Indian and Western companies. They decided to not disclose the technology while safety components were engineered by the privately funded Project Athena. The UN was contacted.
The dangers of unaligned AI were publicly dismissed in the ensuing UN conference. However, major powers started working on CAIS technology. NATO took control of Project Athena and called for global cooperation and caution. India disagreed with the decision to delay the deployment of CAIS, and started working on their own AGI. China and Russia allied to create a CAIS.
Later, South Africa revealed their AGI program. The African-Middle-Eastern Confederation pushed the project forward.
Fearing the conclusion of the global race, NATO offered India, the Sino-Russian cooperation and the African-Middle-Eastern Confederation a treaty. The Pact of Delhi specified that all CAIS programs would be deployed simultaneously.
The South American Union demanded to be included in the treaty, despite not having any CAIS program. NATO agreed, and helped SAU set up their own CAIS seed. The Sino-Russian cooperation considered this a breach of the treaty. They forcefully regained control of their CAIS seed and deployed it. The rest of the seed CAISs were deployed in short order.
Q. In the US, EU, and China, how and where is national decision-making power held, and how has the advent of advanced AI changed that?
A. The seats of power have changed remarkably little nominally. The US is still controlled by the presidency, the EU is split into nations with sovereignty and China remains under the control of the Chinese Communist Party.
However, big parts of the economy have been automated. and CAIS components design the policies that govern trade, research and other. The respective politicians retain the power to veto, but some decisions are made too fast to be effectively oversaw by politicians.
Case in point: one year after the CAIS were initially booted up the Sino-Russian cluster was deemed dangerously unaligned by the rest of the clusters, and was forcefully terminated. The cluster war was bloodless, and mostly conducted in a simulation – once the Sino-Russian cluster became convinced it would lose an all-out war it peacefully surrendered so as to not endanger its citizens. Its territories were split up between the Indian, western and African-middle-eastern clusters. The governments of these territories were not immediately affected, though Russia and China have received a strong warning by CAIS ambassadors to hold new elections in the next two years following a scheme they have designed, allowing for democratic vote.
This war was done without any consultation of the government (though some random citizens on each cluster were polled), in the span of one hour. Politicians were left to wonder about the amount of power they hold in a world too fast for them to comprehend.
Q. Is the global distribution of wealth (as measured say by national or international gini coefficients) more, or less, unequal than 2021’s, and by how much? How did it get that way? (https://en.wikipedia.org/wiki/Gini_coefficient)
A. The distribution of wealth is far less equal. Capital owners and AI developers were assigned a greater weight in the decisions that shaped the world while the CAIS clusters were booting up. This has led to their desires and worldviews being overrepresented in the goals the CAIS pursued.
However, quality of life has risen for all. There is no poverty or illness, and everyone has access to any material riches and AI-provided services they could desire.
The remaining sources of inequality are access to human services and land. The wealthy can afford to hire therapists, masseuses and teachers without having to work anymore. The poor (who are the immense majority of people) can choose either to use free AI alternatives to these services or work to be able to afford human-services themselves. AI alternatives are objectively better, but are considered lower status and not as popular.
Land usage is hotly contested. The wealthy live in historical megapolis like London, Dubai and Tokyo that have grown even larger. The poor compete to live close to the rich to sell them their services. Increasingly smaller living spaces have been built to accommodate the poor. In these megacities, public spaces are mostly shaped by the desires of the wealthy. Poor people keep to their private spaces, and interact with each other in VR.
Increasingly more people (mainly among the poor) have started moving to new communities, where they have plenty of land to build homes and public spaces adjusted to their preferred lifestyles.
Q. What is a major problem that AI has solved in your world, and how did it do so?
A. While not the main reason, the combination of AI material design and nanotechnology helped enormously with the transition into a zero-emissions society. This was most notable in South Africa, India and South America, which became quickly industrialised with far less emissions that experts predicted. By 2040, easy to deploy and highly efficient solar panels power 70% of all electronics. Fusion reactors controlled by Reinforcement Learning AIs are used for energy-intensive applications like large-scale water desalination.
After their deployment in 2044, the CAIS clusters have solved all diseases including ageing, most natural disasters, have greatly advanced scientific discovery and have begun preparations to colonise space.
This happened in the span of a year, and was the result of the collaboration of many differentiated CAIS agents, specialised to systematically produce and test ideas. Their discoveries are deployed via a network of actuators that spans the Earth.
Every citizen can ask for any material product to their phone assistant, and have it delivered to their doorstep in a span of one hour, free of charge. New houses are built in the span of days. Citizens can voluntarily receive nanobot injections that keep them indefinitely healthy.
Q. What is a new social institution that has played an important role in the development of your world?
A. The UN Biological Weapons convention was restructured and funded with a budget of 100 million USD per year. They led the creation of national agencies that regulated custom genomic printing, promoted the stockpiling of medicaments and the development of better protective material. Most importantly, they acted as a central node in the global early detection and response network that was created in the aftermath of COVID19. Arguably, their role stopped two pandemics from developing and greatly reduced the damage caused by the 2032 nano-engineered pandemic.
Their success set precedent for the UN Nanotechnology convention, which in the aftermath of the pandemic was responsible for coordinating developers of proprietary and open source software to install privacy-preserving software locks on home 3D printers, to discourage people from using them to create dangerous pathogens and weapons.
The Partnership on AI acted as a global forum that forged deep ties between western companies and academic institutions on the forefront of AI. The development of CAIS technology was spearheaded by a conglomerate of western and Indian institutions born out of it.
Southafrica started an expansionist programme that ended up in their territory expanding through the western and eastern coasts of the African continent. The rest of the African country and part of the middle east formed a treaty of cooperation and mutual protection, that South Africa was eventually allowed in.
Q. What is a new non-AI technology that has played an important role in the development of your world?
A. The development of advanced nanotechnology shook society in many dimensions.
Households in developed countries were able to afford cheap 3D printers. The economy transformed to revolve around the purchase of digital designs for 3D printers and raw materials. People use them to print furniture, medicines and food, among other things. The 3D printers also transform how waste is dealt with – now destroyed into raw materials for ease of reuse.
Nanotechnology was also instrumental for rapid progress in material science. This led to the development of efficient, cheap and easy-to-install solar panels, better hardware for computation and better drugs. With design increasingly automated, the development cycle of these technologies was now mostly bottlenecked on testing and official approval in the days leading to CAIS.
Nanotechnology was developed in South Africa, mostly led by western entrepreneurs looking for laxer regulation and lower taxation to develop and sell their products. Impressed and scared by the technology, the Southafrican government nationalised the companies, and used them to grow wealthier exporting 3D printed materials and expand their military capabilities.
Some of the engineers responsible for the technology fled the country and founded new companies in the US, UK, Europe, India and China. This led to the spread of nanotechnology across the globe.
The technology was not without risks. In 2032, cheap 3D printing technology was used to develop a highly contagious, lethal and slow acting pathogen by a radical death cult. In the aftermath, severe restrictions were placed in 3D printer software.
Q. What changes to the way countries govern the development, deployment and/or use of emerging technologies (including AI) played an important role in the development of your world?
A. Sharing practises in AI remained on the same trajectory as today. Industries, eager to get credit for their development but not wanting to go through the traditional slow scientific process, moved even further away from journals in favour of preprints and fast-moving conferences.
Major AI conferences started requiring a section on double use and unintended consequences of deployment. This helped AI researchers grapple with the issues of AI alignment, and the field saw a surge of productive activity in the 2030s.
AI developers were also encouraged to report compute and data used for ML experiments in a standardised format. Several non-profit and academic institutions used this to keep track of the players with most capabilities, as well as forecast future advances in AI.
After the 2040 Asilomar conference, major companies in the US, UK and India stopped publishing their findings.
While advances of capabilities were kept secret, AI alignment advances were recognized as a public good for global security and publicly disclosed. As a consequence, from 2038 research of AI capabilities drastically slowed down, while research in safety continued unimpeded.
Governments around the world were mostly unaware of these developments until 2040, when NATO took control of Project Athena. Before, governments mostly focused on regulation of LAWs, data privacy and trying to bolster the competitiveness of their R&D. They also promoted international cooperation and helped subsidise some of the industry-academia collaborations that led to CAIS.
Q. Pick a sector of your choice (education, transport, energy, healthcare, tourism, aerospace, materials etc.) and describe how that sector was transformed by AI in your world.
A. Education has been progressively transformed in many stages. Through the 2020s, there has been a move towards online learning. Videos by world-class educators are shared in the classrooms. They are supplemented by in-person tutoring sessions where students work through homework, with easy access to tutors who can help them with the exercises they struggle with.
Some of the tutoring was moved entirely online, which created an abundance of data on tutoring. During the 2030s, this stream of data was used to train foundation models specialised in tutoring. They will not replace teachers, but they will be able to deal with typical questions the students have. Helped by AI assistants, teachers start being able to manage large classrooms. This is heavily criticised by parents and teachers alike. However administrators push heavily for this to cut costs. Less teachers are hired, but their salaries are dramatically raised, which gets them to play along.
After the deployment of CAIS, teachers can in theory be perfectly replaced. AI teachers are better pedagogues, will not abuse students and can dedicate all their time to a single student. However, hiring humans is perceived as high status, and people (especially the rich) keep hiring teachers for their kids.
In the communes formed after the 2045 migration, teaching is a popular activity, and people share this responsibility.
Q. What is the life expectancy of the most wealthy 1% and of the least wealthy 20% of your world; how and why has this changed since 2021?
A. By 2045, ageing is considered a solved problem and both rich and poor are able to extend their lives for as long as they want. This is mostly achieved through nanomachines consensually injected in most people’s bodies that carefully monitor life constants and are able to fix any problem.
The leading cause of death is suicide, painlessly granted when people desire their existence to cease. Some people have decided to not be injected with nanomachines and keep their mortal lifespan. Many of them will change their mind after facing illness, but some will grow old and will voluntarily die of strokes, cancer, Alzheimer’s, and other diseases associated with ageing.
Some dozens of thousands of pioneers have agreed to have their minds uploaded to computers, where they live free of the constraints of physics. In the future, simulated people will make up the majority of the population, though a large population will still live in physical reality, either on Earth, in space stations or other planets.
Q. In the US, considering the human rights enumerated in the UN declaration, which rights are better respected and which rights are worse respected in your world than in 2022? Why? How?
In one other country of your choice, which rights are better respected and which rights are worse respected in your world than in 2022? Why? How?
A. Material costs have dropped to near zero worldwide. Everyone enjoys free healthcare, education provided by CAIS agents. However, citizens still prefer to be serviced by other humans, and there is no equal access to professionals.
Housing remains a sore point. Land’s value near the megapolis has skyrocketed, and hundred store buildings are built at a frightening pace to house everyone moving closer to the centres of power. There is no homelessness, but many live in small apartments they detest.
Freedom of movement, religion and association varies within countries. Countries in the NATO cluster enjoy an unprecedented amount of freedom. They can move and raise houses pretty much anywhere within the cluster. Bigotry still exists, but strong CAIS filters in social media help people avoid unpleasant interactions.
In the Sino-Russian cluster and some other dictatorial countries there are restrictive policies that make this more difficult. However, the CAIS clusters seem to be nudging politics towards more freedom and democracy worldwide. Many people are unhappy about this, as they see this as a westernisation of their values and loss of cultural identity.
Q. What’s been a notable trend in the way that people are finding fulfillment?
A. Since the mid 2020s, advances in Machine Learning have made art and entertainment accessible for many citizens in the developed world.
Consumption of AI-generated entertainment has skyrocketed ever since. People enjoy custom plotlines in their movies and books, designed to elicit a powerful response from their audience.
Simultaneously, many people have found artistic vocations in writing, painting, VR sims and other artistic pursuits. AI tools help them express their thoughts, resulting in a plethora of art that they eagerly share with friends and family.
This trend became more and more extreme after the deployment of the CAIS clusters. Most media is produced by specialised AIs, consumed and then discarded. Most poor citizens in the megapolis chose to live in small spaces, and interact with each other in VR outside work, driving up consumption of digital goods. While most people prefer to interact with humans, many people form friendships and romantic relations with AI companions.
Recently, many citizens have started migrating to set up alternate communities. They cultivate the lifestyles they find most fulfilling. Some communities are set up like communal farms, where close-knit groups grow their own food and have parties. Others live as hunter-gatherers in a land where nature has been restored by CAIS wild keepers.
Some pioneers have chosen to forego material existence, and be uploaded to computers. Free of physical limitations, they express themselves radically altering their morphology and organise varied virtual events and experiences for each other’s benefit, including concerts, simulated vacation resorts and cafés.
The media piece has been generated using a V-diffusion model. It shows four paintings by a 2044 traveller and an AI.
Top left: the well-off part of a megapolis. We see a public green space and grand housing complexes with large windows, both located next to the coast.
Top right: the silhouette of the poor part of a megapolis. We can see the massive skyscrapers, housing tens of thousands of people each. Stylized spires pierce the sky – these are used for communications and energy collection.
Bottom right: the interior of an apartment of one of the less well-off citizens of a megapolis. It consists of a single room, clean but cramped. A VR headset rests on the bed.
Bottom left: a commune by the river. Designed by and for its citizens, and built by CAIS. A CAIS drone flies over the commune, ready to service its inhabitants if needed.