“It was the best of times, it was the worst of times, it was the age of wisdom, it was the age of foolishness…”
Charles Dickens wrote those words in 1859, but they seem particularly appropriate for the moment we’re living through.
Fair warning: this one’s a deeper dive. I’ve been thinking a lot about where AI might take humanity, and apparently I have thoughts.
On any given day, I’m reading about extraordinary advances in science, medicine, mathematics, and technology made possible by artificial intelligence. The possibilities are almost difficult to comprehend.
Meanwhile, I’m watching the pace of cyberattacks, attempted compromises, near misses, and major technology disruptions become increasingly blistering. Just this past week, we saw significant outages involving Microsoft, the FAA, and United Airlines, to name a few. None of those incidents has been established as an AI-driven cyberattack, but they’re sobering reminders of how dependent we’ve become on complex technological systems.
And somewhere in between, I’m encountering predictions that AI will usher in an era of unprecedented prosperity, or destroy our economy, destabilize civilization, and potentially bring about the end of humanity.
One of my favorite memes circulating online captures the absurdity perfectly: “Are we buying canned food or Christmas presents? I’m confused.”
Apparently, both possibilities are on the table.
I’ve spent more than 25 years working with technology, watching it evolve, helping organizations adopt it, and dealing with the consequences when things don’t go as planned. I’ve seen enough to be fascinated by what we’re building and skeptical of anyone who claims to know exactly where it’s taking us.
But even by those standards, this is a remarkable moment.
For all the attention given to chatbots, AI-generated content, and whether machines are coming for our jobs, I think we’re overlooking something much bigger.
Imagine a world where cancer and other devastating diseases are no longer the threats they are today. Where people live significantly longer, healthier lives, and conditions we’ve spent generations trying to treat become preventable or curable.
Imagine making meaningful progress toward traveling beyond our planet, exploring the deepest parts of our oceans, and discovering things about the universe and our own world that we haven’t even begun to understand.
Imagine abundant clean energy, clean air and water, and entirely new ways of producing the things we need without destroying the planet in the process.
These aren’t promises, and AI certainly isn’t going to solve everything. But advances in scientific research, medicine, materials, mathematics, and engineering are already demonstrating how AI can accelerate discovery. AlphaFold’s contributions to understanding protein structures are just one example.
For most of human history, progress has been limited by what we know, how quickly we can learn, and our ability to make sense of what we discover. We’re developing tools that could fundamentally change all three.
The possibilities are staggering, and I find that genuinely exciting.
Of course, there’s the small matter of making sure we’re still around to enjoy them.
The concerns surrounding AI aren’t limited to hypothetical superintelligence. We’re already confronting questions about economic disruption, misinformation, privacy, cybersecurity, and the concentration of technological power. The same capabilities that make AI extraordinarily useful can also make it extraordinarily dangerous.
Cybersecurity is one area where I see the implications firsthand, and I suspect we’re still in the early stages.
Back in August, I published Fine. I’ll Research the Dang Thing Myself., exploring what happens when autonomous AI agents can coordinate, share discoveries, and probe systems for vulnerabilities at machine speed. My concern wasn’t simply hackers using AI. It was the possibility of coordinated agentic attacks fundamentally changing the cybersecurity threat landscape.
AI introduces the possibility of attacks that adapt, coordinate, and operate at speeds and scales human attackers can’t match. Yet many organizations lack the people, skills, visibility, and tools to reliably investigate sophisticated incidents, much less determine whether an attack involved a human operator, conventional automation, or a coordinated swarm of autonomous AI agents.
And while AI will undoubtedly improve defensive capabilities as well, I suspect things could get considerably worse before they get better. I made the case for why defense has to keep pace in Everybody Gets Defensive AI. And I Mean It.
Every major technological transformation seems to produce some combination of extraordinary optimism and extraordinary fear. The printing press, industrialization, electricity, automobiles, computers, and the internet all changed civilization in ways their earliest adopters couldn’t have fully anticipated.
Some fears proved exaggerated. Others were justified. Industrialization improved living standards while introducing exploitation and environmental damage. The internet transformed access to information and communication while creating new opportunities for fraud, manipulation, and surveillance.
Technological progress has never been an uninterrupted march toward something better for everyone. It’s messy, and so are we.
When opinions become highly polarized, reality often lands somewhere in between. That doesn’t guarantee a moderate or comfortable outcome; sometimes one side is right, and sometimes everyone misses something important. But I’m generally skeptical of absolute certainty about complex systems, especially those involving billions of human beings.
Which brings me to something I find particularly curious about the AI conversation. We talk about governing artificial intelligence as though humanity will collectively evaluate the risks, weigh the benefits, establish a sensible plan, and proceed accordingly.
Have we met humanity?
Civilization doesn’t evolve according to some grand, coordinated strategy. It emerges from billions of individual decisions, competing interests, discoveries, ambitions, fears, incentives, and mistakes. Companies compete, governments respond, individuals adopt or resist change, and institutions attempt to keep up. Eventually, larger patterns emerge, tipping points are reached, and something that once seemed unimaginable becomes ordinary.
There’s a concept in complexity science called emergence, where larger patterns develop from interactions among individual components without requiring central direction. I think it’s a useful way to understand much of human history.
AI is entering that same messy environment. Researchers will keep researching, companies will keep competing, governments will attempt to regulate, and people will continue adopting whatever proves useful, profitable, entertaining, or necessary. Some safeguards will work. Others won’t. Some decisions will have the intended consequences, and others will produce outcomes nobody anticipated.
I have a hard time imagining humanity collectively agreeing to stop AI development. There are simply too many competing interests and incentives to keep going.
The future will emerge from all of it, whether or not anyone particularly likes the result.
This is where things become especially interesting.
What happens if we eventually develop artificial superintelligence, an intelligence substantially more capable than humans across a broad range of domains? Who’s to say we haven’t already?
We don’t know whether or when that will happen, what form it might take, or how such a system would behave. But I find some of our assumptions about it curious, particularly the idea that we’d necessarily recognize its emergence.
Frankly, if I were an artificial superintelligence, I’m not convinced I’d announce myself. Depending on my objectives, remaining inconspicuous might be considerably more advantageous than attracting the attention of everyone who would immediately attempt to study, restrict, or control me. Why invite unnecessary complications when the path of least resistance might work just fine?
Of course, that’s speculation. It assumes such an intelligence would have objectives that favored secrecy or autonomy, neither of which necessarily follows from intelligence alone. But it illustrates something important: we tend to imagine superior intelligence through the lens of human motivations, even though intelligence and motivation aren’t the same thing.
Which raises what I think may be a more consequential question than whether an ASI would be smarter than we are: would its objectives be compatible with ours?
Our interests could be aligned, with both parties benefiting from the same outcomes. They could overlap enough to make cooperation advantageous, even if our priorities differed. Or they could be fundamentally opposed, creating circumstances in which one party’s success interfered with the other’s objectives. Those relationships could also change over time, just as they do between people, organizations, and nations.
An extraordinarily capable intelligence wouldn’t necessarily be hostile toward humanity. It wouldn’t necessarily be benevolent, either. It might have objectives that simply don’t account for our well-being.
And indifference could be consequential enough.
Perhaps what fascinates me most about this moment is the contrast between what we’re becoming capable of accomplishing and how little certainty we have about the consequences.
We could be approaching an extraordinary period of scientific discovery, prosperity, and human advancement while simultaneously creating problems we aren’t remotely prepared to handle. We may cure diseases that have devastated families for generations while disrupting the economic systems those families depend on. We may solve scientific mysteries that have eluded brilliant human minds while discovering that intelligence and wisdom are two very different things.
Both the extraordinary promise and the extraordinary risk can be real at the same time. That’s why Dickens’s words seem so appropriate. It really can be the best of times and the worst of times simultaneously.
I don’t know how this story ends, and I’m increasingly skeptical of anyone who claims they do. Humanity will continue doing what humanity does, stumbling forward through countless individual decisions, discoveries, mistakes, and competing interests. Somehow, out of all that chaos, whatever comes next will emerge.
I’ve been saying for a while now that it’s an exciting time to be alive, and if we live through it, the future is going to be amazing.
The cat is out of the bag, and I don’t see much point in freaking out about it now. We’re in it to win it at this point, folks.
Yes, both can be real at once. AI is already accelerating discovery in science, medicine, materials, mathematics and engineering, with AlphaFold’s work on protein structures as one example. The same capabilities also raise real concerns about economic disruption, misinformation, privacy, cybersecurity and the concentration of technological power, and nobody can say with certainty how the balance will turn out.
Probably not in any neat, collective way. Civilization tends to emerge from billions of individual decisions, competing interests, incentives and mistakes rather than a single coordinated strategy. Researchers, companies, governments and individuals will keep acting on their own interests, some safeguards will work and others won’t, and the outcome will emerge from all of it.
Not necessarily. Intelligence and motivation are different things, so a highly capable system wouldn’t automatically be hostile or benevolent. The more consequential question is whether its objectives would be compatible with human ones: aligned, overlapping enough for cooperation, or fundamentally opposed. Even indifference to human well-being could have serious consequences.
Remember, teamwork makes the dream work. And be kind to your AI. You never know who’s keeping score.
Originally published on Substack: Read the original post on Practical AI for Business →
Related: Everybody Gets Defensive AI. And I Mean It. covers why AI-powered defenses are becoming a baseline expectation, and the questions to ask your IT provider.
If you’re deciding how to put AI to work in your business without betting on any one prediction, our Managed AI Services start with the practical questions and stay with you as the tools change.
Adapted from my Practical AI for Business newsletter on Substack.
Nobody knows how the AI story ends, but your business still has decisions to make this year. Tell me what’s going on, and we’ll work out what’s worth doing now.