Essay
Consider the AGI Dolphin
My girlfriend and I live 1,500 miles apart. As the well-trodden terrain of early-relationship phone calls fades into the distance, we must set our sights on new and exciting conversational horizons.
“It says here that dolphins have an IQ of 45.”
Sometimes those horizons are viral tweets from pop-sci engagement-farming accounts.
“What… does that even mean? Are they marking scantrons with their noses?”
To be honest, I had no idea what it meant. I just wanted to hear her reaction to this ridiculous factoid.
“You’re right, it should probably be lower.”
If only I had learned to apply myself properly, perhaps I, too, could make a living rage-baiting for likes and comments. For now: she trains me.
“What! No, if anything it should be higher. They basically have two brains, you know—they can turn one off while they’re sleeping.”
“I did not know that. That’s actually kind of beautiful.”
“And they certainly wouldn’t be dumb enough to fuck up the planet as badly as we have.”
My first instinct was to agree, but something in the AI discourse-tainted recesses of my mind whispered:
“AGI dolphin…”
I was going to need to put on my billionaire thinking cap for this one.
Consider the dolphin. The aquatic mammal, a public favorite perhaps for its perpetual resting smile, has been observed to teach, play, scheme, and grieve. They exhibit cultural diversity in their use of tools for hunting and mating, and even enjoy an occasional hallucinogenic high from a gentle passing around of unsuspecting pufferfish. Some animal ethicists have argued, in fact, that dolphins’ unusually high intelligence should entitle the sea creatures rights as “non-human persons,” which would make it not only morally wrong, but unlawful to hold in captivity, or to slaughter for consumption.
Imagine, for a moment, what it would mean to turn the dolphins’ intelligence up to eleven. Even by human standards, most dolphins would be far from eligible for sainthood: the flippered critters have been known to engage in necrocoitus and infanticide against their own kind, and, to the bewilderment of marine biologists, to display an aggression that transcends taxonomy. The unprovoked ramming, tossing, and drowning of porpoises by sexually frustrated young male dolphins is so well-documented that it has earned its own name in the literature: ‘porpicide.’ High intelligence often correlates with problematic behavior in other animals, too: ravens deceive with faked food caches, sea otters sexually coerce and drown seal pups, and chimpanzees wage full-on wars. A heightened awareness of social hierarchy and reward circuitry provides a master key that unlocks more doors than were perhaps originally intended by Nature. What would a dolphin be capable of were it Mensa-eligible?
Humans are unfortunate victims of evolutionary lag: our cognitive hardware was developed for small tribes and short timescales, yet our technological and cultural acceleration has far outpaced our biological evolution. At the very least, hyper-intelligent dolphins would be limited in their capacity for world domination by their anatomy. For a time.
But if they could, they would.
‘Could’ what? Anything really. Industrialized pufferfish hallucinogen harvesting would be the obvious place to start. One might pray that young male dolphins are not the ones leading the intellectual charge, though.
For most animals, an appropriately developed intelligence allows a species to carve out their niche in the world, without catching up to the consequences of being too smart for its own good. Once the bumper guards are lifted, though, the gutter is suddenly in play.
Artificial general intelligence—AGI, for short—is the term now used to describe the ever-moving goalpost of what was once considered to be “artificial intelligence,” pre-ChatGPT. A growing community of AI-enthusiasts known as e/acc (“effective accelerationism”) believes that bigger artificial brains, the kind sci-fi has been promising since before Dune’s Butlerian jihad, are the solution. To what?
Everything.
War, famine, disease, biodiversity loss, climate change—the Big Problems. This is an incomplete list. The e/acc rationale is this: the faster we create God (AGI), the faster we solve the Big Problems. In the meantime, all other priorities are secondary to this end. Trillions of dollars (and billions of gallons of water) are merely temporary inconveniences to be solved with absolute certainty by a future hyperintelligence beyond anything currently conceivable to us, a species of mid-wits.
Marc Andreessen, venture capitalist and vocal e/acc proponent, announced in 2023 that he was present to deliver good news: “AI will not destroy the world, and in fact may save it.” While we wait patiently for the birth of an AGI savior, we might also ask ourselves: why can’t humans seem to get the job done themselves? Two immediate possibilities present:
- Humans lack the raw intelligence to solve these Problems.
- We are already plenty smart. What we lack instead is something closer to a wisdom that is only gained through the trials and errors of a burgeoning intraplanetary society.
Human mid-wittery is then a best case scenario. The alternative suggests the only way out is through, and it is unclear at present whether we can hold our breath long enough to get to the other side.
An AGI born of scenario 1 doesn’t automatically lead us down a sure path, either. A God is made, intentionally or not, as a reflection of its creator—and vice versa, if that’s your thing. Not a day goes by in which the collective moral compasses of the world are not updated, however incrementally, towards new magnetic north poles. Ethical alignment is therefore a function of time; an AGI of the 1960s would surely not agree ethically with one built in 2026.
Whether an AGI submerged wholly in Beatlemania and Ed Sullivan shows would be better or worse than an actual dolphin is a matter of taste. The machine’s resulting stance on civil rights might depend on whether its training run finished before or after a given Sunday in Selma.
If we are to believe our inability thus far as a species to curtail our technological advances in favor of long term stability is a defect born of a lack of intelligence, then yes, perhaps AGI can save us. But if we end up with instead a very very smart dolphin, we very well may be fucked from both ends.
Andreessen is not the only tech oligarch-turned-prophet. In the fall of 2025, Peter Thiel delivered a lecture series in San Francisco arguing that those who wish to slow or regulate AI progress are “legionnaires of the Antichrist,” naming Greta Thunberg and Eliezer Yudkowsky as examples—a series he brought, by March 2026, to a four-day, closed-door engagement in Rome, a stone’s throw from the Vatican. Thiel’s thesis, despite hailing from a different tradition, dovetails with Andreessen’s own Techno-Optimist Manifesto, in which he states not only that “any deceleration of AI will cost lives,” but that such preventable losses are, in fact, “a form of murder.”
Unlike journalists, venture capitalists are not professionally obliged to footnote statements which may be tied to potential financial gain, or loss thereof. Investment rounds live and die by hype cycles, and what better instrument to orchestrate urgency than a bona fide apocalypse pending?
For its most ardent conscripts, ‘move fast and break things’ has been promoted from tech mantra to eschatology—an expensive one, at that.
The anticipation of a near-future Revelation event is not constrained to Silicon Valley: According to the Military Religious Freedom Foundation, there have been over 200 reported complaints of American commanders assuring troops that the war in Iran is “all part of God’s divine plan,” invoking biblical ideas about “end times.” On paper, AGI accelerationists and Christian evangelicals are not natural allies, yet they have arrived miraculously at the same operating premise: that humanity is living in its final chapter, and that anything less than hastening its End is the ultimate sin. Whether we get an Old Testament God, a New Testament Jesus, or an AGI trained on the Techno-Optimist Manifesto come Judgment Day, only time will tell.
Salvation is on a tight deadline for us humans and our earthly cohabitants. The Endangered Species Committee, known informally as the “God Squad,” has reconvened for the first time in over thirty years, amid the war in Iran. Despite a 2025 conclusion by federal scientists that extractive activities in the Gulf may jeopardize the Rice’s whale’s continued existence, the God Squad voted unanimously to exempt oil and gas drilling. There are 51 known living Rice’s whales remaining—all in the Gulf of Mexico.
The bumper guards have been lifted, and are now floating in space. Meanwhile back on Earth, we are being asked to put all our eggs into a single basket, with different names. A full-throttle acceleration towards the development of an ultimate, holy intelligence will require humans to broaden our definition of acceptable loss. A data center per diocese, a dolphin in exchange for whales. My only worry is that we have misread the scriptures: If AGI is not as inevitable as prophesied, we may be left with fire and brimstone, and no one—or thing—to grant us absolution.
“The problem with computers is that they have no skin in the game.”
“What about elephants?”
“Elephant AGI… has a nice ring to it.”
“They mourn their dead. They even hold vigils. And they are afraid of mice.”
“You make excellent points. I think they might do a better job than the dolphins. Or us.”
She yawns. End-of-world commiseration is made no easier by a two-hour time difference.
“Everything is a means to The End. Capital T, capital E.”