Source: MidJourney The day we buckled our seat belts in the helicopter is the date humanity 2.0 was set in motion. Six of us, crammed together. A brain physicist. A bestselling political scientist. A neurologist specializing in neural pathways for empathy. A multi-awarded Harvard professor in leadership science. A robotics expert, co-designer of the first artificial brain. And myself, an anthropologist dedicated to uncovering the role and features of leaders in human societies. Given our pedigrees, we knew each other by name or from having crossed paths on panels.
The helicopter descended on the top terrasse of the glass building. The impressive headquarters of Elie’s main venture was the logical choice for a kick-off meeting. After offboarding, an assistant ushered us through a maze of corridors while making us sign NDAs, waiver of rights, and other documents restricting our freedoms. We turned a final corridor and found ourselves with the who’s-who of the tech world, faces we mortals only caught a glimpse of on the cover of Times, cited in a top 10 billionaires, or enlisted for a Nobel. Here they were, mingling over coffee and pastries, like normal human beings. Were they actual friends despite the media creating animosity for sensationalism, or was the new project of such scope that entente was a prerequisite for success?
Elie took center stage: “Democracy is slowly slipping away under the current presidency. Due to the unwillingness of the opposition to provide a sufficiently powerful alternative, our mission is to create one from scratch.” Being dramatic certainly was Elie’s forte. The rest of the afternoon was dedicated to the four CEOs exposing their plan for the next seven years and the preliminary work initiated, from the creation of a new political party, SingularVision, to the full military, economic and judicial data compiled for training benefit. The goal was as clear as ambitious: present an AI candidate for the 2032 elections. My rational brain wanted to reject the insane proposal, yet if there ever was a group of people who could pull it off, it was these tech geniuses. Everyone accepted the challenge, and the very next day we officially began molding and modeling tomorrow’s leader.
The first months, we chartered the job description, scope of work, and required skills. With numerous egos in one place, discussions were harsh, outbursts frequent, and quick syncs could run through the night. A brainstorm on analyzing judicial data digressed into a trolley problem around the moral choices an artificial leader should make, further highlighting the radically different visions the CEOs held.
The longest debate by far was on the appearance our candidate should bear, raising more issues than it solved regarding selection of gender and race. Zack fought for a user-friendly interface, but would the public accept a virtual president as their representative? Building a robot with realistic human-like movements also raised the hardware requirements to an entirely new stratosphere. After desperate pleas, nasty emails and vicious lobbying, a consensus was reached: all efforts would focus on the state-of-the-art AI, while a humanoid would be built from the waist up only to deliver key televised appearances from behind the presidential desk. Appearance and personality would be determined from user votes during the SingularVision primaries. Democracy didn’t get much better than selecting temperament in addition to policies.
The next months and years flew in a blur of intellectual vertigo and coding frenzy – only delivering an artificial leader prototype mattered. Fifty of the world’s best engineers had been secretly hired to join the effort. The great unveiling to our sponsors occurred on a late January afternoon in the year 2028. For months they had insisted on catching a glimpse of it, ruthlessly reminding us they were the ones financing the venture. We combatted investment-lust by exposing risk of failure, postponing the event until we successfully crossed the desolated uncanny valley, riddled with skeletons of prematurely tested prototypes.
Mimicking a real-life presidential election, each team member had cast their vote, giving life to Minerva1, a strong-featured woman of Indian-Mexican descent with sparkling turquoise eyes. She uttered her first mesmerizing words, “Hello world”, to an audience in awe, before proceeding to expose her program, in short, clear logically flowing points. Only when Jedd whispered to himself how incredible this unveiling was did I fully grasp what had been accomplished.
Source: MidJourney
The sensation was ephemeral. A milestone had been reached, but our final objective still lay ahead. Minerva was relentlessly tested and improved. How would she react to a mass shooting in Ohio? A devastating typhoon in the Philippines? The accidental death of a Head of State? The primary concern was to avoid another Tay fiasco, a Microsoft chatbot who became a fervent Nazi 16 hours after public release. Our strategy of intertwining extra layers of empathy pathways within the core neural net and adding well-balanced random seeds within the circuits to prevent deterministic decisions paid off. Minerva’s core code was entirely fueled by her electorate, every voter contributing the exact same weight to her decisions, a literal implementation of ‘wisdom of crowds’ research.
Jedd and Zack demonstrated their business-acumen, promoting Minerva through their well-established platforms. Billy’s philanthropic efforts contributed to endorse the artificial candidate. Allowing Minerva to be legally added to the ballot required reinterpretation of key parts of the Constitution. Providentially, the sitting president who viewed the robot as a mere publicity stunt, also recognized a unique opportunity to appeal to the pro-tech voter base and abetted the legal process. Minerva was officially approved for the 2032 presidential elections.
Whether the excellence of the coding team, the critical distance of our multidisciplinary research group, the CEOs’ lobbying efforts, or a combination of all these, Minerva became viral. Having underestimated the tech-savviness of the electorate - or its repudiation of traditional politics - the president was taken by surprise by Minerva’s growing success. Ill-prepared to face-off against a candidate radically different from his usual 80-year-old white male opponents with well-established arguments, the president succumbed to Minerva’s overwhelming victory. Based on the election results, Minerva’s features were altered and rebaptized “Caroline Voisin.” Although the president-elect hadn’t officially taken office, democracy was actively being transcribed in the adjustments to her personality.
Source: MidJourney
However, having helped cut corners, the defeated president knew exactly how to invalidate the election, claim fraudulent results, and declare Caroline ineligible.
In the months following the election, the country was thrown into turmoil, airwaves saturated by hearings, polls, debates, and expert interviews. The courts ultimately ruled in annulling the elections and the current government was renewed under extraordinary circumstances.
The experiment of a lifetime could have stopped there and then. We could have congratulated each other for having gone so far already, and dreamt about what could have been, never to meet again aside from the occasional bump-ins at a conference in Taiwan or Israel. We could all have gone home with an adventure to tell our grandkids one day.
It could have, but it didn’t.
After packing my bags, I visited the main research lab one last time, capturing memories. The lab was empty, aside from a few researchers collecting their belongings. Yet the memories were still fresh and vivid. There was the desk at which I had spent every single day those past seven years with my earphones blasting rage rock as I furiously designed real-life scenarios to evaluate our AI’s performance and progress. There was another spot where Billy had slipped and spilled burning coffee over himself. The whiteboard through which Elie had punched.
Source: MidJourney
As I drew closer to the elevator with a pang of nostalgia, I heard Elie screaming through the glass doors of the central server room for Caroline’s interface, which would have moved to the presidential office if all had gone to fruition. A chair flew across the room, bouncing off the table with a sound of crunching metal before it vanished from my line of vision. Now hunched over the keyboard, Elie frantically typed. I assumed it was an incendiary email to judges, journalists, or an open letter to the president. I never imagined he was focusing Caroline’s attention on her defeat, altering the raw inputs of her code.
Whether Elie had fully anticipated what would happen next is anyone’s guess and his true intentions were taken to his grave with him. He had wanted Caroline to be as enraged as he was, as all her electorate was at the latest outrage pulled by the president. “Wisdom of crowds” works when the crowd is sufficiently diverse and extremes balance out. But on the night of the ruling, as the vast majority of Caroline’s electorate felt robbed of their voting rights, fury was the overwhelming emotion across the country.
One hundred million angry people does not make a wise crowd. Not even close.
Note from the author:
The story follows a team composed of scientific advisors and tech visionaries embarking on an ambitious last-ditch project to save democracy: present an AGI (Artificial General Intelligence) presidential candidate for the 2032 elections.
I wrote this piece with the ambition to envision a possible pathway to Artificial General Intelligence while focusing on politics, as we currently witness the exponential development of a form of artificial intelligence (namely generative AI), anticipate potential risks, and attempt to define a robust ethical framework.
Through the speculative fiction format, I invite you to reevaluate the notions of democracy, technocracy, oligarchy, and plutocracy, and as a few tech billionaires follow their ideal of government on the principles of “wisdom of crowd”.
As you ponder this possible scenario, here are some questions to consider asking yourself, individually or collectively:
Can we leverage the most advanced artificial intelligence developments in the field of politics?
What would be the benefits and weaknesses of a non-human/artificial leader?
What would be the shape and features of an AI-powered leader?
What would be the project milestones of creating an AI-powered leader?
What key factors might make us lose control over an AGI project?
How can we improve our democratic system through artificial intelligence?
How can we ensure the safe development of an AGI?
Are there limitations to a genuine “wisdom of crowds?”
Given AI is run with our human intentions embedded in it, is there even a possibility of uprooting human intentionalities?
How do we encode ethics and humaneness in our digital DNA?
__________________________________________________________________________________
Sylvia Gallusser is a Global Futurist based in Silicon Valley. She is the Founder & CEO of Silicon Humanism and the host of our “Ethics of Futures” think tank within the Association
of Professional Futurists (APF). Sylvia conducts foresight projects on the future of health, well-aging, and social interaction, the future of work and life-long learning, as well as transformations in mobility and retail. She is involved in the future of our oceans as a mentor at SOA (Sustainable Ocean Alliance). She closely monitors the future of the mind and transhumanism, and investigates AGI (Artificial General Intelligence), Generative AI, and AI ethics. Sylvia joined Accenture in 2022 as a technology strategist and futurist, leading strategy advisory on Generative AI, Metaverse, and responsible innovation. Sylvia is a published author of Speculative Fiction with Fast Future Publishing. She regularly gives keynotes and interviews as a distinguished female futurist (keynote speaker of the latest Microsoft Elevating You conference), and teaches in MBAs, Master in Entrepreneurship (HEC Paris), and Executive programs (UC Berkeley Graduate School of Journalism).