by Amy Webb
The G-MAFIA studied the Chinese cities where smart city initiatives were piloted—such as Rongcheng, Beijing, Shenzhen, Shanghai—and identified best practices to pilot in the United States. We now have a few American smart cities—Baltimore, Detroit, Boulder, and Indianapolis—that are testing out a wide range of AI systems and services. Networks of cubesats overhead—tiny satellites the size of a Rubik’s Cube—feed real-time data into AI systems that can recognize objects, unique light patterns, and heat signatures. This, in turn, allows city managers to predict power outages, monitor and reroute traffic, manage water reserves, and clear ice and snow off the roads. AI also helps them manage budgets and personnel throughout the year, surfacing entirely new ways to shave off fractions of expenditures at scale. Budget shortfalls aren’t gone, but they’re not nearly as bad as they used to be—and the citizens in these cities are buoyed by a sense of hope they haven’t experienced in many years.
These systems tie into public safety departments, like police and fire, which are using AI to sift through massive amounts of data, including video: if there’s no sound, pattern recognition algorithms can lip read and produce transcripts. Generative algorithms also autocomplete holes in audio tracks, and if anything is fuzzy, a stitching algorithm sharpens the focus. AI scans millions of images looking for patterns that the human eye would miss. This hasn’t been without controversy, of course. However, the G-MAFIA’s commitment to privacy means that our PDRs aren’t available to search through without a warrant. We feel safe knowing that the G-MAFIA is safeguarding our privacy.
As it evolves, AI is helping us mature into better humans. With the G-MAFIA, federal government, and GAIA taking active roles in the transition from artificial narrow intelligence to artificial general intelligence, we feel comfortably nudged.
2049: The Rolling Stones Are Dead (But They’re Making New Music)
By the 2030s, researchers working within the G-MAFIA published an exciting paper, both because of what it revealed about AI and because of how the work was completed. Working from the same set of standards and supported with ample funds (and patience) by the federal government, researchers collaborated on advancing AI. As a result, the first system to reach artificial general intelligence was developed.
The system had passed the Contributing Team Member Test. It took a long time for the AI community to accept that the Turing test, and others of its ilk, was the wrong barometer to gauge machine intelligence. Tests built on either deception (can a computer fool a human into believing it’s human?) or replication (can a computer act exactly as we would?) do not acknowledge AI for what it has always been: intelligence gained and expressed in ways that do not resemble our own human experience. Rather than judging an AGI on whether or not it could “think” exactly like we do, the AI community finally adopted a new test to measure the meaningful contributions of an AGI, which would judge the value of cognitive and behavioral tasks—different, but powerful—we could not perform on our own. AGI would be achieved when the system made general contributions that were equal to or better than a human’s.
The G-MAFIA spent many years researching and developing an AGI that could sit in on a meeting at work and make a valuable contribution—unsolicited—before the meeting concluded. They code-named the AGI Project Hermione, inspired by the Harry Potter character who always, and in every situation, knew just what to say or do. Making a valuable contribution in a group is something that most people on Earth have, at some point, had to do themselves: at work, in a religious setting, at the neighborhood pub with friends, or in a high school history class. Simply interjecting with a factoid or to answer a question doesn’t add value to a conversation. Making a valuable contribution involves many different skills:
• Making educated guesses: This is also called abductive reasoning, and it’s how most of us get through the day. We use the best information available, make and test hypotheses, and come up with an answer even if there’s no clear explanation.
• Correctly extracting meaning from words, pauses, and ambient noise: Just because someone says they’re happy to take on a new project doesn’t mean it literally makes them happy. Other cues, like their body language, might tell us that they’re fairly unhappy with the request but, for whatever reason, they’re not able to say no.
• Using experience, knowledge, and historical context for understanding: When people interact, they bring with them a nuanced worldview, a unique set of personal experiences, and typically their own expectations. Sometimes logic and facts won’t win an argument. Other times, they’re all that matter.
• Reading the room: There’s the explicit interaction and the tacit one happening beneath the surface. Subtle cues help us figure out when there’s an elephant demanding our attention.
Project Hermione sat in on a GAIA working-group session. Eighteen members of the group discussed and debated the existing standards for AI, which were developed by either those people sitting in the room or their predecessors. As the group was diverse and made up of leaders from different countries and cultures, there was a lot of subtext: certain power dynamics, personality clashes, and feelings of inferiority or superiority. The group treated the AGI as an equal member, with no additional privileges or special exceptions. Halfway into the session, the AGI pushed back on a small but growing consensus in favor of regulations. It tactfully argued against the idea and recruited another member of the group to support an alternative. Project Hermione had made a valuable contribution. (Invaluable, some would later argue.)
What made Project Hermione a success wasn’t just that it passed the Contributing Team Member Test with such ease—but rather that GAIA and the G-MAFIA saw that moment as both a warning and an opportunity. They continued recalibrating their strategies and standards to keep a few steps ahead of AI’s technological developments. They decided to limit the rate of self-improvement, adding constraints into all AI systems to keep humans in the loop. Now GAIA researchers follow new protocols: they run simulations to understand the impacts of more powerful AGIs before approving them for general-purpose, commercial, or military uses.
The G-MAFIA are wealthy, influential, powerful companies—and their success is growing. They are building exciting practical applications for AGIs to enhance our productivity and creativity, and they’re also helping to create plausible solutions for humanity’s most pressing challenge: climate change. As the jet stream shifted far north, America’s breadbasket went with it, well over the border into Canada, decimating farms and the US agricultural sector. Coffee and chocolate can’t be easily grown anymore outdoors. Citizens in Bangladesh, the Philippines, Thailand, and Indonesia have become climate refugees in their own countries. Amazon, partnered with Microsoft, France’s Groupe Danone, and DowDuPont in the United States, is using AGI alongside genomic editing to populate indoor farms with fresh produce.
Google and Facebook are using AGI to help safely and securely move entire populations, forming and shaping the Earth with new, comprehensive human communities. AGI helps them to predict which specific locations can most easily sustain life in a way that feels comfortable and preserves the cultures of affected people. Previously uninhabitable regions of our planet are either terraformed or transformed using adaptive building materials. Landscrapers—large, sprawling complexes just a few stories high—have created entirely new urban footprints. Inside, cableless elevators transport us omnidirectionally. It’s a new architectural trend that’s helped the world’s most important economic centers boom, which in the United States includes Denver, Minneapolis, and Nashville.
For a while, it seemed as though China would retreat and retrench with just a few allies—North Korea, Russia, Mongolia, Myanmar, Cambodia, Kazakhstan, Pakistan, Kyrgyzstan, Tajikistan, and Uzbekistan. Universities in GAIA nations stopped accepting Chinese applicants. Wary of surveillance and the possibility of their PDRs being hacked, China’s tourism industry dried up completely. GAIA nations relied on automated systems to produce the materials needed for manufactur
ing, repatriating factories back home. Ultimately, China’s state government determined that its exclusion from GAIA was destabilizing its economy and, as a result, causing significant political and social unrest. Reluctantly, China agreed to adopt GAIA’s norms and standards and to accept all of the transparency measures required of member nations. Communism isn’t dead—there’s still plenty of political strife to contend with, along with all the usual tensions related to different styles of governing and leading.
AGI certainly didn’t emerge without many new problems, some of which we were able to anticipate. Like other technologies that transformed human society over time, AGI has displaced jobs, led to new kinds of criminal activity, and has at times brought out the worst in us. But in the 2040s, AGI isn’t an existential threat.
In home and at work, we use a primary AGI to access information. It’s a control agent that takes on different forms and modalities depending on the situation: we speak to it, interact with it on a screen, and send it data from inside our bodies. Every family has a butler because every household has an AGI trained and attuned to its unique circumstances.
One of the biggest and most noticeable changes brought about by AGI is a sharp increase in sophistication across most facets of human existence. We can thank the G-MAFIA for how much the quality of life has improved. What used to be time-consuming, difficult challenges—like trying to schedule a time that works for everyone, sorting out an after-school activity calendar, or managing our personal finances—is now fully automated and overseen by AGI. We no longer fritter away hours attempting to hit “inbox zero”—AGIs work collaboratively to facilitate most our low-level thinking tasks. We finally have simple household robotics that make good on their promises to keep our rugs and floors clean, our laundry put away, and our shelves dusted. (We think of 2019 as a much simpler time, one full of tedious and monotonous manual tasks.)
The common cold no longer exists, and neither does “the flu.” In fact, we marvel at the naïveté of earlier doctors. That’s because IBM and Google’s AGIs helped us see and understand millions of different viroids. Now, when you’re not feeling well, an AGI diagnostic test helps determine what, exactly, is making you sick so that a treatment—one that maps to your PDR—can be prescribed. Over-the-counter medications are mostly gone, too, but compounding pharmacies have seen a resurgence. That’s because AGI helped accelerate critical developments in genetic editing and precision medicine. You now consult a computational pharmacist: specially trained pharmacists who have backgrounds in bioinformatics, medicine, and pharmacology.
Computational pharmacy is a medical specialty, one that works closely with a new breed of AI-GPs: general practitioners who are trained in both medicine and technology. While AGI has obviated certain medical specialists—radiologists, immunologists, allergists, cardiologists, dermatologists, endocrinologists, anesthesiologists, neurologists, and others—doctors working in those fields had plenty of time to repurpose their skills for adjacent fields. As a patient, you are happier. You don’t spend hours trekking to different doctors’ offices, getting conflicting messages, and you are no longer overprescribed medications. If you live in a more remote area, AGI has meant a dramatic improvement in your access to care.
We all have our genomes sequenced at birth—the process is now cheap and fast enough for everyone, regardless of income level, to participate. You decided to get your genome sequenced as well because your sequence is a vital component of your PDR. In addition to providing you a window into your unique genetic makeup, AGIs look across all of your data to detect genetic variants and learn more about how your body functions. Of course, in the United States and in other nations, there are small groups who are opposed to the practice—just as anti-vaxers once fought against vaccines. While parents can opt-out for religious or ideological reasons, few make that choice.
Because of AGI, we’re healthier—and you have new options when it comes to dating and marriage. Advanced forms of differential privacy allow a third party to look at your data (your PDR, genome, and medical records) without divulging who you are individually. That’s made AGI matchmaking providers incredibly useful, because now you can choose to optimize for family (producing children with genetically desirable combinations), wealth (projected lifetime earning potential) or fun (whether or not they’ll laugh at your jokes).
AGI assists you in other creative endeavors, beyond looking for love. The original members of the Rolling Stones died years ago, but thanks to replicating algorithms, they’re still making new music. That sensation you felt after hearing the first 30 seconds of “Paint It Black” for the first time—the melancholy guitar melody, followed by eight loud bangs on the drum and a repetitive hook that culminates in Mick Jagger singing, “I see a red door and I want it painted black”—was a singular moment of excitement and satisfaction. It didn’t seem possible you might get to feel that way again with a new Stones song, and yet their latest track is just as loud, hard, and fulfilling.
While newspapers in print are gone, the news media has adopted AGI as a means for distribution. Once the Contributing Team Member Test was passed, news organizations acted quickly to build a different news distribution model, one which still made money but had a sharper eye on the future. These days, most people don’t get or turn on the news—they have a conversation with a smart newsagent. The New York Times and Wall Street Journal both employ hundreds of computational journalists—people with strong hybrid skills sets in both traditional reporting and AI. Together, these teams report on stories and select relevant facts and data for inclusion in conversational engines. AGI-powered journalism informs us, and we can modulate it to include a political slant or more background information or a “deep cuts” version offering ancillary characters and miscellaneous facts. We’re asked to participate in news analysis and editorial feeds, debating and constructively arguing with the newsagent using our voice or interacting with screens (smart glasses and retractable tablets). There are still plenty of long-form stories told in text and video.
AGI hackers—which most often are other AGIs—are an ongoing irritant because of “no-collar crime”: nonviolent criminal acts committed by AGIs, which reveal the people who created their original source code. Local law enforcement agencies employ officers who are cross-trained in data science. With the help of China’s BAT, the Big Nine are working collaboratively on advanced hardware, frameworks, networks, and algorithms that are capable of withstanding attacks. GAIA’s partnership with Interpol has, for the most part, kept serious crime at bay.
The smart city pilots launched two decades earlier in Baltimore, Indianapolis, Detroit, and Boulder were a success and helped other communities learn best practices, which lead to the formation of the Federal Smart Infrastructure Administration (FSIA). Like the Federal Highway Administration, the FSIA operates under the Department of Transportation and oversees all of the connected systems that power our cities: wireless power transfer stations, decentralized energy generators (kinetic, solar, and wind), vehicle-to-infrastructure networks, and the fiber optics that bring sunlight into our underground farms. Sensor data is aggregated and used to model the overall health of our communities: access to clean air, the cleanliness of our neighborhoods, and our use of parks and outdoor recreational areas. AGIs predict and mitigate brownouts and water crises before they happen.
As we near the transition from AGI to ASI, an exciting opportunity has just become visible on the horizon: brain-to-machine interfaces. We’re on the precipice of molecular nanotechnology, and we hope that within a few decades, we’ll be able to record data from the billions of individual neurons inside our human brains simultaneously. Microscopic computers, the size of a grain of sand, would gently rest on top of the brain and detect electrical signals. Special AGI systems, capable of reading and interpreting those signals, could also transmit data between people. A brain-machine interface could someday allow a healthy person to retrain the brains of stroke victims who are paralyzed or have lost their ability t
o speak. Brain-machine interfaces, which we could theoretically use to transfer memories between people, might also help us experience empathy in a deeper and more meaningful way.
That possibility has us thinking about new uses for AGIs. We want to untangle thorny philosophical questions: Is our universe real? Can “nothing” exist? What is the nature of time? AGI can’t give us the answers we want, but the G-MAFIA has deepened our understanding of what it means to be human.
2069: AI-Powered Guardians of the Galaxy
The intelligence explosion, as foretold 100 years ago by British mathematician and early AI pioneer I. J. Good, begins in the late 2060s. It’s becoming clear now that our AGIs are gaining profound levels of intelligence, speed, and power and that artificial superintelligence is a near-term possibility. For the past decade, the Big Nine and GAIA have been preparing for this event—and it has calculated that once human-level machine intelligence has been surpassed, an ASI could be just a few years away.
After much consideration, a difficult decision is made by all members of GAIA to prevent ASI from being created. Some of those involved in the conversation became emotional—arguing it wasn’t fair to handicap AI’s “beautiful minds” just as they are beginning to reach their potential. We debate whether or not we are denying humanity the possibility of even greater opportunities and rewards.