I am exhausted by the AI industry’s permanent state of prophecy
After four years covering AI, I’ve had my fill of confident pronouncements about what this technology will mean for humanity. There’s plenty to learn from what’s happening right now.
If you’re looking for prophetic warnings or promises about AI over the past few years, you don’t have to look far. AI companies and their leaders have long been fluent in a language that can sound downright biblical.
Google DeepMind CEO Demis Hassabis has envisioned a “golden era” of “maximum human flourishing,” where humanity could “travel to the stars and colonize the galaxy.” OpenAI CEO Sam Altman has promised “shared prosperity to a degree that seems unimaginable today.” Anthropic CEO Dario Amodei has declared, “I believe we are entering a rite of passage, both turbulent and inevitable, which will test who we are as a species.”
And in a recent interview with The Economist, Elon Musk predicted that AI could surpass the combined intelligence of humanity within five years and an “AI robot singularity” could arrive within a decade. He even cast himself as Cassandra, saying people don’t realize that what he predicts “will come to pass.”
Lately, I’m exhausted by all of it. The AI industry seems to exist in a permanent state of prophecy where either abundance is coming or catastrophe is looming, or both at the same time.
This past week alone offered plenty of examples. Meta CEO Mark Zuckerberg published a manifesto predicting that within “the next few years,” people will have access to “superintelligence beyond human capacity,” ushering in a “new era of personal empowerment” and some of “the most amazing” decades in history.
Over the weekend, Amodei reiterated his belief that AI could make it possible “to cure most human disease in ~5-10 years, as crazy as it may sound to ordinary people and frankly to biologists as well.”
Personally, after four-plus years of covering AI all day, every day, I’m increasingly wary of being told that the world is perpetually on the verge of blowing up, becoming radically abundant, curing cancer, losing its jobs or entering some unprecedented new phase of human existence.
The AI industry’s default mode
I’m not saying AI isn’t a big deal. It clearly is. I’m not saying AI isn’t changing the world. It already has. And I’m not saying none of these predictions could come true. They could.
But my problem with these confident, sweeping pronouncements is that they have become the AI industry’s default mode, often crowding out the harder work of understanding the technology that actually exists today.
None of this is new. Back in February 2023, just a few months after ChatGPT launched, I was already struck by the quasi-religious tone of OpenAI’s pronouncements about the future. So when the company published a manifesto called “Planning for AGI and beyond,” I rewrote part of it in the style of the biblical prophet Isaiah, including:
1:1 — The vision of Sam Altman, which he saw concerning planning for AGI and beyond.
1:2 — Hear, O heavens, and give ear, O earth: for OpenAI hath spoken, our mission is to ensure that artificial general intelligence (AGI) — AI systems that are generally smarter than humans — benefits all of humanity.
1:3 — The ox knoweth his owner, and the ass his master’s crib: but humanity doth not know, my people doth not consider. For lo, if AGI is successfully created, this technology could help us elevate humanity by increasing abundance, turbocharging the global economy, and aiding in the discovery of new scientific knowledge that changes the limits of possibility.
1:4 — Come now, and let us reason together, saith OpenAI: AGI has the potential to give everyone incredible new capabilities; we can imagine a world where all of us have access to help with almost any cognitive task, providing a great force multiplier for human ingenuity and creativity.
It was a funny parody to me at the time. But three and a half years later, after an endless procession of intelligence ages, golden eras, existential threats, superintelligence manifestos, imminent singularities and promises of radical abundance, I’m kind of over it.
A message from our sponsor:
Codestrap’s AI Value Factory converts your tokens into durable operating leverage that continues to deliver because it keeps shipping software solutions out the delivery bay.
Prophecy can crowd out the present
The problem is what living in a permanent state of prophecy does to the conversation about AI in the present.
The OpenAI-Hugging Face incident is a good example. I was at Black Hat when OpenAI and security researchers publicly laid out what happened, and I reported the details as they emerged. The agents found vulnerabilities, coordinated their efforts and even created their own internal message board to share information and assign tasks. The security implications are real and worth taking seriously.
But within days, the incident itself was being swallowed up by the prophecy.
Headlines declared that “rogue AI” was no longer science fiction and asked whether humanity had “lost control.” The incident was called a “warning shot” and a “wake-up call.” Gladstone AI CEO Jeremie Harris described an Uber driver spontaneously asking him about the episode as evidence that “the world really is waking up” to the potential of “rogue AI.”
Personally, I don’t think the hairstylist, Pilates teacher, family member and old friend who have asked me about the OpenAI-Hugging Face incident are waking up to anything but confusion. And all the hand-wringing about “rogue AI” risks obscuring a more immediate question, especially in an incident with so many unanswered questions about human error: Who is responsible and accountable for the consequences when AI systems behave in ways their creators didn’t anticipate?
Prophecies have power
In the case of AI, prophecies influences everything from investment to regulation to how the public thinks about the technology. And when the people making sweeping claims also run the companies building and selling the technology, raising staggering amounts of money to do it, and in some cases preparing for IPOs, the prophecies gain extra power.
It’s part of why I started Ground Level AI in the first place, to spend less time reporting on the latest prophecies from AI labs and more time reporting on what happens when the technology we actually have meets the real world. That is, companies trying to deploy it, workers figuring out how to use it, communities confronting the infrastructure required to power it, security teams dealing with risks.
I’m not saying we should ignore the future. AI is moving extraordinarily fast, and what seems speculative today can become real much sooner than expected. And we need organizations examining risks, building scenarios and developing policies around AI.
But the faster AI capabilities change and fan out into society, the more important it is to stay steeped in reality. As the technology keeps changing beneath our feet, I think it’s essential to distinguish what is actually happening from what might happen and what someone simply predicts will happen.
After all, there is plenty to deal with at ground level without constantly racing ahead to the promised land or the apocalypse.
Me too Sharon. Me too.
Sharon, this is both on point and hysterical. As a teacher working with women to embrace AI. I feel ya!