September 15, 2026
by Matthew Jensen
I have had a surge of people who are genuinely scared about AI (artificial intelligence) coming to me over the past week.
Not “AI might take my job” scared – I mean… “AI might take humanity” scared.
So, I sat with my own thoughts and debated myself (and ChatGPT itself) and thought about it.
I listened to the people who believe AI could ultimately become an existential threat and I argued the other side to try to find where “I” land.
And then I tried something increasingly unpopular in 2026… I tried not to pick a team.
Here is my honest conclusion.
We. Don’t. Know.
Anyone who tells you with absolute certainty that AI is going to destroy humanity does not KNOW that.
And anyone who tells you with absolute certainty that AI is perfectly safe does not know THAT either.
This has never happened before.
Zero previous civilizations created intelligence greater than their own so it isn’t something we can study to see how it turned out.
Not a great sample size.
So yes, the risks are real.
AI… may eventually improve AI!
It may become dramatically more capable than we are. Humans may use it maliciously. Companies or countries may race each other and take risks they should not take. That is a risk.
These possibilities deserve some very serious attention.
But it is important to be clear: a POSSIBILITY is not a PROBABILITY.
And UNCERTAINTY is not EVIDENCE of CATASTROPHE.
Human beings have been faced with technological advances since… well, humanity.
The automobile, aviation, electricity, nuclear technology were all dangerous or terrifying. The Internet created dangers we had never considered or imagined.
And unfortunately, tragedy HAS struck…
Humans have died in cars, airplanes, spacecraft… every type of transportation – including horse-drawn carriages.
People have even died because of smartphones.
It turns out that we, as humanity, seem hell bent on inventing increasingly sophisticated ways to kill ourselves!
Luckily, we, as humanity, find increasingly sophisticated ways to stop ourselves from doing it.
Seatbelts, airbags, air traffic control, electrical codes, drug trials, nuclear treaties, cybersecurity, and I could go on.
We generally don’t solve every problem before we invent the thing.
We invent the thing, discover the exciting parts, discover the part where it might kill us and then somebody says, “Okay. We should probably put a guardrail there.”
And – fun fact – I was actually part of Y2K remediation.
People look back now and laugh because midnight came, the lights stayed on, planes stayed in the air and civilization survived.
But Y2K wasn’t proof that everybody worried for nothing.
Thousands and thousands of us worked to make sure nothing happened.
Sometimes a catastrophe that doesn’t come to pass isn’t evidence that the warning was stupid.
Sometimes it means somebody fixed the damn problem.
And that’s the part of the AI conversation I think we sometimes forget.
When people imagine AI getting dramatically smarter over the next ten years, they sometimes pit that future AI against the humans of today.
AI gets smarter and smarter and smarter while apparently the rest of humanity stands around saying, “Well… shit.”
I don’t buy that.
We won’t be the HUMANS of 2026 in 2036.
We’ll have ten more years of discoveries, failures to learn from, laws, safeguards, research, international agreements, monitoring and ideas nobody has thought of yet.
And we will have something humanity has never had during a technological revolution before:
AI helping us solve the problems created by AI.
Computers already fight computer viruses, algorithms detect algorithmic fraud, and software tests software.
So the future probably isn’t “HUMANS versus SUPERINTELLIGENCE.”
It’s probably humans, institutions, scientists, engineers and AI – working in tandem to deal with whatever comes next.
And sometimes the solution can’t even BE invented until the problem exists long enough for somebody to look at it and see the room for improvement they hadn’t visualized before they saw the thing in existence.
That’s when somebody has their lightbulb moment.
This is what we, as humans do.
Does that guarantee everything will be okay?
No.
There is no contract with the universe promising humanity a happy ending.
Pretending there is would be every bit as irresponsible as telling you that we’re doomed.
But there’s another side of this calculation that matters.
What happens if we stop?
There are risks on both sides.
If AI eventually helps us discover new cancer treatments, diagnose diseases earlier, prevent accidents, create new medicines, make education available to everyone, help people with disabilities, and solve scientific problems humans have not yet been able to solve on their own, then STOPPING that progress has consequences too.
The person killed by an AI mistake will make the news – and they should – but the person who DOESN’T die of cancer twenty years from now because AI helped discover the treatment?
There will never be regular “LOCAL MAN CONTINUES BEING ALIVE TODAY” headlines.
Fear, on the other hand, has an excellent marketing department… it always has.
Every generation has had its version of “THIS changes everything.”
And often… it actually does.
So after everything I’ve researched, listened to and argued about, here is where I land.
I DON’T think we should blindly trust artificial intelligence and I ALSO don’t think we should blindly fear it.
AI deserves vigilance proportional to its power and the fear ought to be proportional to the evidence.
We need to collectively demand safeguards, alignment, accountability from the companies and governments building these things and we need to listen to the people sounding alarms.
The fact that we are HAVING this argument isn’t evidence that humanity has lost – or is about to lose control, but rather, it is more likely evidence that the human control system has already switched on.
Someone builds something.
Someone else says, “Wait. What happens if…?”
Someone tests it.
Someone finds the flaw.
Someone invents the safeguard.
Someone makes a law.
Someone breaks the law.
Someone invents a BETTER safeguard.
It’s all so VERY human.
And this is not the last time we will go through this.
Quantum computing, radical longevity, interstellar travel, maybe even perfect humanoid companions… who knows?
Maybe someday – time travel!
Although I suggest we make sure THAT one exists before we start regulating parking for our DeLoreans.
There will ALWAYS be new frontiers and every one of them will give humanity another reason to ask, “What if this changes everything?”
And you know what? It will.
That’s what frontiers do, after all.
The question isn’t whether change is coming, the question is what we do when it gets here, and no person can promise you that AI won’t hurt people any more than another person can promise you that everything will work out – there is no guarantee that we will get it all right.
But look around you.
Almost everything surrounding you right now exists because somebody, somewhere, walked toward something humanity didn’t completely understand yet, took the risk, discovered the danger, made it safer, and eventually… what once seemed impossible became ordinary.
So for me, I refuse to surrender the future to either blind optimism or blind fear.
I want us alert, careful, and arguing.
I want the smartest people on Earth worrying about what could go wrong while the rest of us remain curious about what could go right.
Because maybe the lesson of human history isn’t that “everything always works out,” because clearly it doesn’t.
Maybe the lesson is simpler.
We keep working it out.
And if there is one thing about the future that history actually HAS taught us, it’s this…
Humanity has never needed to know exactly what comes next to take the next step.