Eisenhower, Asimov and AI: Sunday musings…9/13/2026
All of my news sources are literally drowning in AI doomsday predictions. Kristen Welker of Meet the Press asked the Speaker of the House when, not if, he would force a vote on a proposal to “slow the roll” of AI development. Every major American AI company has lost at least one, and most several key senior researchers or executives over the last couple of weeks. Some leave for a think tank, some return to academia, and at least one is off to study poetry. They all had the same reason for leaving.
None of their former employers were adequately frightened by the specter of truly autonomous, uncontrolled AI agents.
There are calls all around, from the left and the right, inside and outside the industry, from the young and the old. Sides have been taken. The fearful call for strict government control over the development of AI. Those concerned with the state of the world’s economies and the relative power of one major power or another argue for restraint levied on the regulators lest American economic interests fail to reach each next pinnacle first. While hardly luddites, the former group would sacrifice most, if not all of the benefits that might accrue to humanity from such a useful technology working in partnership with humans. The economy-first group shows a certain “damn the torpedoes” disregard for anything short of a dominant “first mover” victory.
Here’s my bid: Isaac Asimov gave us the AI-specific solution in 1942, and President Dwight Eisenhower created the effective fail-safe strategy in the early days of the Cold War.
Let’s start with Asimov. For those of you who haven’t done much classic science fiction reading, Isaac Asimov has, at the very least, a plausible case as the GOAT of Sci-fi writers. One of his pet topics was the inevitability of non-living entities possessed with super intelligence that eventually exceeds that of humans. His vision was that the earliest iterations of this “being” would be some sort of humanoid robot, a vision elucidated in what has become known as the “I, Robot” canon. My brother-in-law Peter and I have had an epic, decade-long discussion of the possibility that an “after-life” might exist in the quantum physics-predicted entity called the multiverse. Consistent with his deep belief in the rise of AI, in his last, great creation the “Foundation” trilogy Asimov predicts an eventual confluence of a universe-wide AI and the consciousness of every human who ever lived. The ultimate eternal victory for both AI and humanity.
In order for “Foundation” to be possible it was first necessary to ensure the survival of mankind once AI arrived and began its rise. Asimov addresses this in a 1942 short story, The Runaround, with the creation of the Three Laws of Robotics. First Law: robots may not harm humans or allow them to come to harm by inaction. Second Law: Robots must obey humans unless doing so violates the First Law. Third Law: Robots must protect their own existence unless doing so violated the first two laws. Some time later Asimov added an overarching law that would provide direction for AI as a whole in its interactions with humanity writ large, the Zeroth Law: A Robot may not harm humanity as a whole, or cause it harm through inaction. These laws, and examples of how they might function as the robots advanced and humans became more and more engaged with them, were published as a series of 9 short stories that were eventually assembled in the 1950 collection “I, Robot”.
All of the widely-held fears about the dangers of AI could be addressed by mandating the programming of the Three Laws and the Zeroth Law as a foundational element in EVERY AI of any sort whatsoever worldwide. Compete to be the best in any sector whatsoever, just don’t do so in a manner that will cause harm to either an individual or humanity as a whole. As I see it doing so would surely solve an awful lot of the small scale harms we see like financial scams, extortion and the like. On a grander scale, how could an AI possibly attack a power grid or water supply without running afoul of at least one of the Rules of Robotics? Every government and every international organization of every kind should mandate the adoption and implementation of the Three Laws and the Zeroth Law.
So why should any country agree to this? What’s to keep, say, China from just thumbing its nose at the rest of the world and forging ahead with a program that aims for subjugation of every other nation on the planet? This is where Eisenhower comes in. Why is it that mankind has managed to not obliterate itself in a hailstorm of nuclear missiles traveling hither and yon, annihilating both the human race and the planet? You know, mutual destruction that bombs whoever survives back to the stone age. Planet of the Apes stuff.
At the dawn of the Cold War, Eisenhower and the leader of every other nuclear armed country arrived at what came to be known as MAD: mutually assured destruction. Land a nuclear bomb in one country and that country responds with every bomb it possesses. This, as much as anything else, is why we aren’t all mute field workers under the watchful eyes of sentient apes. Smaller actors–think North Korea–are held in check by larger ones, usually in the same geopolitical sphere. Think China. Rogue agents (pun intended) would be just as much of a wild-card in AI as they could become in the nuclear arena. Honestly, you have to believe that several established nuclear powers would combine to destroy a country like Iran if it ever sent a nuke at Tel Aviv or a U.S. military base in Kuwait.
Unlike a nuclear war in which human life would be decimated, the target of the assured destruction would be the AI capability of whatever entity let loose an AI that harmed humans on any scale. Knock out AI and what do you get? Well, you definitely don’t consign a particular country or society to Stone Age conditions unless you think a pre-Siri or pre-Alexa smartphone as the highest tech level does so. Think 2010 or thereabouts. It’s not like you would be exposing the population of India or China or the U.S. to living conditions equivalent to the immediate post-Industrial Revolution, let alone the Middle Ages. You’re probably just backing them up to a pre-Uber Eats era where you actually had to ride your bike to the market.
Destroying or disabling a country’s AI does not kill its people, it makes it beholden to the rest of the world to support the advancement of its economy.
Smarter people than I, especially those who manage the world’s macroeconomics, can flesh out the details of what Mutually Assured Destruction of an offending actor’s AI would look like. My point is that the framework for an answer to the most basic of concerns about AI–whether it will destroy humanity as we know it–were worked out in the 1950’s by a science fiction writer and the war-weary President of the United States. Program the Three Laws of Robotics and the Zeroth Law into every AI, superseding all other functions. Agree on the START Treaty AI equivalents and create onerous disincentives sufficient to prevent violations. Allow technological advances that seek to gain an economic advantage in the same way that developments–electrification, computerization, etc.– have done throughout history.
Sounds like the makings of a really good book.
I’ll see you next week…
This entry was posted on Sunday, September 13th, 2026 at 3:53 pm and is filed under Uncategorized. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.