“Recent developments in generative AI show models that are increasingly capable of learning and reasoning without human feedback. They are discovering unique solutions to problems that have eluded humans, training and optimizing other learning models to be better and requiring fewer resources to achieve frontier capabilities. At the same time, leading public-facing models have found a role as companions and confidants for many, helping people navigate their lives and work through social and emotional challenges.
The softer cognitive and emotional features that make us uniquely human will likely see the greatest evolution. … It’s likely that the near future will see more of us recomposing our identities around virtual personalities. Some humans are already ‘cloning themselves’ into online AIs that can represent them at scale, for example, in order to respond to thousands of follower messages on social platforms … Humans’ immersion in these virtual experiences in encounters with deepened game mechanics and lifelike virtual characters will further blur relationships, reshape socialization and erode what it means to be uniquely human.
“More of us are now encountering these capabilities online, at work and when using our smartphones. Younger generations are showing significantly greater usage and adoption. It’s obvious that frontier AI will be likely to continue to get closer to us through many aspects of our daily lives. But it’s worth noting that it won’t be universal any time soon, as access is gated by incomes and employment and understanding. Some enjoy much greater access to advanced AI capabilities. Others will soon be likely consumers of AI products and become the beneficiaries of its impacts, for good and ill.
“So, what does the advance of these tools mean for humans? Assuming that impacts and access will be unevenly distributed, the most basic needs of being human are unlikely to change much as these have endured through the past technological revolutions. Basic survival needs, shelter, the drive to reproduce, competition for resources, conflict and collaboration, socialization and identity, enquiry, ideologies and religion – each of these will persist as fundamental to the human experience. But how we pursue and attain them will surely change, and the softer cognitive and emotional features that make us uniquely human will likely see the greatest evolution.
“AI companions are a notable example. There’s plenty of anecdotal evidence emerging from people claiming that conversing with LLMs has led them to emotional breakthroughs. People are already relying on AI companions throughout the day, and roleplaying with them to compose the right texts before sending to friends and parents and lovers. The softer human traits like identity and socialization are already changing to accommodate non-humans (and have been for millennia in some ways). We seem uniquely drawn to anthropomorphizing, seeking friends and companions wherever we can. It’s likely that the near future will see more of us recomposing our identities around virtual personalities.
“Some humans are already ‘cloning themselves’ into online AIs that can represent them at scale, for example, in order to respond to thousands of follower messages on social platforms. Video game non-player characters (NPCs) – non-human characters that are built into the games’ algorithms have been part of that scene for some time now and will likely soon become freer in their interactions with human players, more conversational and improvisational. Humans’ immersion in these virtual experiences in encounters with deepened game mechanics and lifelike virtual characters will further blur relationships, reshape socialization and erode what it means to be uniquely human.
We still assume we’re the special ones, somehow fundamentally unknowable. Indeed, we do not know how we think but we defend with passion that we’re the only one’s able to do so. And yet, it increasingly looks like advanced software trained on more data than God running on more compute than most nation-states can approximate our level of intelligence feat. Our original sin is being unable to reckon with ourselves and the world. So, are we made in the image of our Gods, or are we just very complex machines that can ultimately be modeled and understood? Generative AI may force us to confront this question head-on.
“Competition and individualism can also be amplified by frontier AI, empowering some humans to be more capable in their pursuits. We could see more hyper-empowered individuals able to act in much higher orders with the help of the best models – including models that may or may not be ‘legal.’ Sociopathy could be fostered and reinforced in some individuals working closely with a nigh-omnipotent AI companion toward self-serving goals. Goal-seeking behaviors in general will be amplified by AI, for good and ill. There are already emerging challenges with criminal networks using AI to impersonate loved ones and make demands for ransoms, again showing both the duality of empowerment and the fading uniqueness of being human.
“This is all assuming the current trajectory continues. Transformer models could hit a wall, but so far, they have not. Recent developments have only enabled greater reasoning. Trillion-dollar companies are spending hundreds of billions to build out more compute, while bleeding-edge innovators find ways to do more with less, indicating that costs could go down while utility grows. For now, building and operating frontier models is extremely expensive, and the business models have not yet revealed clear paths to paying for it all.
“This may be the Big Question: Will the models establish strong enough value and relevance – and trustworthiness – so we drop our guard and give them more work to do on our behalf? Many of the changes to being human outlined here have already been underway for some time, buoyed by the previous technological revolution. Gen AI looks to be an accelerator that could amplify these trends while enabling a step-change into non-human intelligence. How much of human endeavor will be passed on to agentic AI? Who will have access to such capabilities, and who won’t? And what parts of being human will be transformed, subsumed, or simply ditched?
Indeed, we do not know how we think but we defend with passion that we’re the only ones able to do so. And yet, it increasingly looks like advanced software trained on more data than God running on more compute than most nation-states can approximate our level of intelligence feat. Our original sin is being unable to reckon with ourselves and the world. So, are we made in the image of our gods, or are we just very complex machines that can ultimately be modeled and understood? Generative AI may force us to confront this question head-on.
“Some people refer to the point in time at which a future might emerge that can do anything a human can do as the technological Singularity. That was before the breakthrough of generative pre-trained transformers (GPTs). When teenagers are communing with AI companions, nobody talks of the Turing Test. Even now the debate about artificial general intelligence (AGI) is getting fuzzy, with barriers falling and milestones being passed. If we haven’t hit that milestone yet, we likely won’t notice when we’ve passed it. Smartphones, by all accounts, are fantastic magical devices of the future, but this fact never really occurs to us.
“Any speculation about what it means to be human in an age of non-human intelligence is just that: speculation. We still assume we’re the special ones, somehow fundamentally unknowable. Indeed, we do not know how we think but we defend with passion that we’re the only ones able to do so. And yet, it increasingly looks like advanced software trained on more data than God running on more compute than most nation-states can approximate our level of intelligence feat. Our original sin is being unable to reckon with ourselves and the world. So, are we made in the image of our gods, or are we just very complex machines that can ultimately be modeled and understood? Generative AI may force us to confront this question head-on.”
This essay was written in January 2025 in reply to the question: Over the next decade, what is likely to be the impact of AI advances on the experience of being human? How might the expanding interactions between humans and AI affect what many people view today as ‘core human traits and behaviors’? This and nearly 200 additional essay responses are included in the 2025 report “Being Human in 2035.”