Welcome to the Uncanny Valley
Whenever I read a newsletter or blog post these days, the first thing I do is scan for the seams. Does the writing feel too earnest and overconfident, like a sixth grader’s essay on a subject they briefly googled? Does it invoke cringy puns? When you sniff at it, does it give off that telltale stink of synthetic authenticity?
I’m not personally in the camp that hates AI outright, or hates the smooth stream of human language that can be effortlessly squeezed out of large language models like ChatGPT. But there is something about the whole thing that feels deeply uncanny. The question I’m asking myself when I’m filtering for AI “slop” is not that different from the question many of us increasingly ask ourselves as we watch the news, or the question philosophy students invariably ask themselves in college dorm rooms at two in the morning: “Am I living in a simulation? Is this all fake?”
It’s not especially provocative to observe that we’re now deep in the Uncanny Valley as a society, caught uncomfortably in the liminal space between not just technological paradigms, but also geopolitical regimes, demographic cliffs, and geological eras. Historic observers looking back at the early decades of the 21st century are unlikely to find more clarity in our current moment than we do, but will instead confirm that 2024 was inherently blurry, like a snowball in an avalanche, somewhere between the 20th century’s end of history and whatever awaits us on the other side of the Valley.
So what should we expect to find there? In 2034, ten years from now, will AI have taken over the world, and reduced us to paperclips? Or will it turn out to be another overhyped dud, like the metaverse and crypto are currently perceived? Or could there…perhaps…be a third, more interesting road waiting for us?
This October 23rd and 24th, IFTF will host its first in-person Ten-Year Forecast Summit since 2019, where business leaders, technologists, visionary artists, and veteran foresight practitioners will come together in San Francisco to explore the opportunities, challenges, and mind-stretching oddities of the coming decade. Together we will confront six provocative forecasts about the future of AI and our society, wading deeper into the waters of the Uncanny Valley to see if we can get a peek at the world waiting for us on the other side.
To prepare for this journey, let’s unpack the first forecast:
By 2034, there will be more AI bots than humans, and in digitally intermediated forums they will be nearly indistinguishable from us.
This is a bold statement, and at first blush might sound like overwrought AI hype. So let’s take a step back for a moment with 2024 eyes. Are we in an AI bubble that’s bound to pop sooner than later? Probably. To some degree, almost definitely. As profound a shift as AI ultimately represents, there’s nothing that the memestock markets of the 2020s like more than to find a good bet and overhype it to the moon. Regardless of its ultimate potential, the reality is that AI is not going to change the world tomorrow, or next year. Institutional change happens at a much slower pace than technological innovation. Social hierarchies and cultural norms change over decades, not months.
As Institute for the Future co-founder Roy Amara famously said (though it’s often misattributed to Bill Gates), “we tend to overestimate the effect of new technologies in the short term, and underestimate the effect in the long term.” By 2026, AI might well seem like an overspent NFT relative to the gobsmacking sums of money being invested in its promise today. But by 2034, there will be more AI bots than humans.
We will not think of most of them as “bots.” If the consumer robotics market begins to make headway, we might reserve the term “bots” for those AI agents situated in plastic and metal bodies, acting as porters and farm labor for those who can afford them. Most of the AI entities we encounter will likely be faceless and voiceless, working behind the scenes to fill out our paperwork, coordinate our accounts, and resolve minor bureaucratic issues on our behalf. We mostly won’t think about them at all, any more than we routinely think about the electricity coursing through the walls of our house.
But many will have faces, and voices, and we will speak with them like they’re intelligent. Some will be photorealistic, beautiful, and impossibly charming. Others will be magical cartoonish beings summoned from children’s imaginations. Some will just look exactly like you, regardless of whether you want them to or not.
We can say this confidently because this is something generative AI already excels at. Outside of whether or not Artificial General Intelligence (AGI) is around the corner, or never coming at all, even the smallest, most cost-efficient generative AI models that exist today are already incredibly fluent with natural human language. As software interfaces and operating systems integrate this new affordance into their designs over the next few years, it will become only clearer what a pivotal milestone this new capability represents in the history of technology. Since the advent of personal computing a half century ago, computers have been highly technical systems that required specific literacies and skills to use. Even as laptops and smartphones have become more accessible and intuitive, it’s still true today that technical applications require technical users. But that won’t be nearly as true in the coming age of relational computing.
The Dawn of Relational Computing
What is “relational computing”? We use the term to describe the subtle-but-profound shift in what computers and computer software will soon become relative to what they’ve been until now.
In the age of relational computing, it will be more important to have clarity about what you want to accomplish than technical expertise in the specific tools for accomplishing it. Humans will still need to leverage their experience and acquired skill sets to identify useful solutions to well-defined problems (and this specifically will become a more valuable skill than ever). But in the age of relational computing, those who can articulate a clarity of vision will find a bustling ecosystem of AI bots waiting to support them in translating their high-level goals into tasks, plans, deliverables, and workstreams. Some tasks that take months or years to undertake today will still take months or years to accomplish in the future. But other month-long tasks will suddenly be achievable in a week. Or a day. Or a matter of seconds.
As digital systems begin to meet us more on our own terms — speaking the way we do, accommodating their understanding to our personal and professional contexts, translating our roughly worded intentions into precisely executable steps — we will think of them less like complicated machines, and more like an expanded set of relationships that we’re continuously cultivating, managing, and shaping to translate the ideas in our heads into actions in the world. And by empowering people to think more about the higher level outcomes they’re trying to achieve, these technological capacities might ultimately free us from having to think so much about computers or AI at all, instead focusing on the human relationships we are always invariably seeking to improve: our relationships with each other, with our local communities and environment, and with ourselves.
We should not assume that there will be one universal reaction to these synthetic relationships. Some people will find themselves in constant conversation with their personal synthetic assistants, like Jarvis in Iron Man. They will perceive their bots as an extension of their own internal dialogue. We will all end up knowing people who find it difficult to talk to anyone who isn’t an AI bot. Some people will date their AI bots, and we’ll spend a long time debating how we feel about that.
Others will shun human-like AI interactions as much as they can, especially during the more awkward first generation. Human-to-human services will be considered a premium value in a world soaked in AI, and pro-human efforts will find eager audiences. But in ten years, nobody will be able to entirely avoid the presence of AI bots woven into their lives, whether on the YouTube videos they watch, the interactive billboards they happen to walk by, or when they’re ordering fast food directly from Ronald McDonald, brought charismatically to life by the large language model embedded in the drive-through screen.
The point is that we should not be preparing for a world of AI takeover, but a world with millions of new AI entities whom we will form different kinds of relationships with, in similar ways that we form relationships with the people around us today. Some of these relationships will be transactional and professional, others will be informal, personal, and even intimate. There will not just be AI bots that we trust and those that we don’t (though there will be many AI bots we don’t trust). There will be AI bots that we trust with our finances and legal matters, which might be different from the AI bots we trust with our vacation planning and grocery shopping, which will be different from the bots we trust with our medical and therapeutic assessments.
There will be ongoing efforts to mandate labels for AI-generated interactions, but this will prove impossible to enforce on the margins. Perhaps your insurance company will let you know when the agent you’re speaking to is an AI bot (or more likely, the agent your personal AI assistant is speaking to). But when a family member calls you at 2 in the morning asking for money to help them out of a jam, you’re going to need some kind of two-factor authentication for their identity.
If things get really out of hand, we might even decide that important agreements and transactions should probably happen in person. Maybe just for a decade or two, just to be safe. And we’ll probably see some upside to that if it comes to it.
What’s Next
So how will this new paradigm of computing play out? How will its cascading consequences gradually transform our relationships with our employers, our governments, and our communities? What else might we discover about ourselves and our world on this journey through the Uncanny Valley?
For that we’ll have to wait for IFTF’s Map of the Decade 2024-2034, which will be unveiled in October 2024, for IFTF Vantage partners and invited guests. And over the next several months, we’ll be publishing research perspectives that offer more glimpses into the provocative relational futures of our industries and institutions.