We write this message in 2024, the biggest election year in history. The governments of over four billion people — half of humanity — are up for election, and citizens are preparing to march to the polls under a fog of disinformation. Two days before Slovakia’s most recent parliamentary election, a fake audio recording apparently exposing a major party leader’s plan to buy votes spread like wildfire on social media. The recording could not be debunked in time, and that party lost. AI can now act as a propaganda machine, with outputs so readily generated, disturbingly convincing, and highly personalized that their circulation is impossible to tame through existing means. 2024 is only the canary in the coal mine: if AI continues to hijack reality, our generation’s trust in information and institutions could soon collapse.

It’s not just politics. All across the Internet, AI-generated non-consensual pornography has exploded. How can we feel safe at a time when strangers can produce and distribute synthetic pornographic images that attempt to defile our personhood?

Eventually, we may even witness the total degradation of the collective bonds that sustain societies. Amidst an epidemic of loneliness, young people are turning to chatbots in lieu of friends, family, and mental health professionals — whether they’re looking for advice in a crisis, seeking companionship, or just passing time. We want to live in a world where people enjoy meaningful relationships with other people — not one where that role is passed onto machines.

Without intervention, the lines between real and artificial, between human- and machine-generated, and between truth and deception will completely blur — and it is our generation that will suffer the most.

AI is guiding high-stakes decisions in hiring, education, criminal justice, and more. But these often-opaque systems have further entrenched discrimination. In the United Kingdom, students flooded the streets after an exam score prediction algorithm used in university admissions amidst COVID-19 was found to penalize schools in high-poverty communities. In the United States, Black communities have seen a string of wrongful arrests driven by faulty facial recognition technology, which disproportionately fails on darker-skinned faces.

But even if AI systems were technically unbiased, the structural context of deployment could still produce harmful outcomes. For instance, law enforcement agencies often use AI in ways that mirror existing patterns of overpolicing. In the hands of authoritarian regimes, AI can be used to surveil populations and chill free speech. Similar technologies are already entering school campuses. It is not enough for AI to be fair — we must reimagine how, where, and on whom we choose to use it.

Our generation is coming of age in a world where seemingly neutral AI is supercharging surveillance and discrimination, with little accountability for its failures. Whether we’re standing trial, seeking healthcare, or applying for a job, this will reshape how we experience both public and private life.

The future of work for our generation hangs under a cloud of uncertainty. In the long run, AI may create jobs and help renew our sense of purpose — or it may send us hurtling towards a post-work reality for which societies are entirely unprepared. This is a treacherous gamble. As AI begins to match and even eclipse human performance across a range of tasks, young people look ahead with anxiety. How will we continue generating economic value, and what will become of our livelihoods? A student today cannot be sure that their dream job will still exist when they enter the workforce. Even if fears of the worst turn out to be overblown, we are undeniably on the brink of far-reaching economic upheaval.

In the near future, we will see inequality rise as AI produces winner-takes-all economic dynamics and further concentrates market power. A landmark Massachusetts Institute of Technology study found that since 1980, automation has been the primary driver of income inequality in the United States, accounting for more than 50% of the increase in the wage gap between more and less-educated workers. We can expect labor markets in the Global South to be hardest hit. In every country, low-wage workers may be the most gravely affected, though white-collar workers, too, are at serious risk of displacement. 

AI advancement could send shock waves that obstruct pathways out of poverty in the Global South, disrupt economic systems, and disempower human workers globally. If we instead aspire to a world in which AI drives economic gains for all and unlocks time for the activities we find most meaningful — while uplifting, rather than superseding, humans — leaders must intervene.

AI may make future wars exceptionally lethal. This is no longer science fiction. We have already seen the use of semi-autonomous attack drones in conflict zones, including in the Russo-Ukrainian War. Drones are not the end game. Autonomous weapon systems are being primed for use at every stage of the kill chain—not just the final judgment call—creating a dangerous cascading effect on military decision-making. While these systems are purported to enhance efficiency and reduce casualties, they raise profound ethical concerns.

We cannot and must not recklessly move forward with the deployment of offensive autonomous weapon systems. When AI-enabled weaponry operates without human oversight, assigning responsibility in the event of a misfire becomes nearly impossible. These systems undermine the laws of war, nullifying the established principles of human dignity and proportionality in combat. What’s more, autonomous weapon systems can behave unpredictably, malfunction, or be hacked or misused, potentially resulting in unintended civilian casualties or indiscriminate attacks on non-military targets. At worst, AI could power weapons of mass destruction, inflicting large-scale casualties.

The specter of autonomous killers looms large. Our inaction promises a future where the instruments of war are unaccountable and potentially uncontrollable.

As AI becomes easier to access and more powerful, with fewer mechanisms for human control and interpretability, the potential for both intentional and unintentional misuse will multiply. New models might aid non-state actors in creating biological weapons. Human-like conversational AI might help conduct large-scale persuasion campaigns and meddle in elections. Rogue systems might be used to attack critical infrastructure and sabotage key networks.

There are some disturbing warning signs. Early reports indicate that AI intended for drug discovery can be repurposed to suggest tens of thousands of lethal chemical warfare agents, and that GPT-4 can be jailbroken to generate bomb-making instructions. Frighteningly, additional threats we cannot see coming may still be on the horizon. To be sure, many of AI’s harms — including those we have described earlier, from disinformation to algorithmic bias to labor displacement — are already with us, and they are well-documented. We see a moral imperative to confront existing risks and future-proof for oncoming ones.

Moreover, the energy-intensive nature of training and operating large-scale AI models threatens to jeopardize our progress on climate goals. Global AI governance must address AI’s growing carbon footprint and mitigate other environmental harms.

One thing is clear: our generation’s collective safety now hangs in the balance. But no one actor can prepare or protect us. AI development and its risks are increasingly borderless, and so too is the responsibility of AI oversight. The UNESCO Recommendation on AI Ethics, a promising first step, was endorsed by almost every country on the planet; the Bletchley Declaration, another powerful springboard for governance, resulted from unprecedented dialogue between the world’s major AI players. In the face of this challenge of global proportions, we must stand united. As with climate change and nuclear nonproliferation, we need international coordination to govern AI.