Will AI Exterminate the Human Species?

Or Will Our AI Boom Merely Waste Trillions of Dollars of Capital?

By
The Unz Review

September 29, 2026

Last week President Donald Trump gave a public address to the United Nations General Assembly, declaring that he might soon “annihilate” Iran, a large country of over 90 million people.

The UN has always been a rather stodgy and subdued international organization, normally the venue for diplomatic presentations of remarkable blandness. In its entire history, I doubt it has ever seen a speech of comparable boldness, let alone one made by the leader of a top world power.

Trump’s obvious threat to launch nuclear attacks for the first time in more than eighty years greatly concerned me, as did the rather milquetoast response by top Congressional Democrats, let alone the silence of the almost totally supine Republicans.

In his open letter to Congress, Prof. Jeffrey Sachs reasonably denounced these “truly deranged remarks” by “Mad King Donald.”

Meanwhile, Prof. John Mearsheimer regarded all these dramatic threats by Trump as merely more of the latter’s notorious bluster. In an hour long interview with former New York Times journalist Chris Hedges, the academic argued that America had no viable military options against Iran or even against the Houthis of Yemen. This reality explained why our top generals had persuaded Trump to once again back away from the commitments he had made to our Saudi allies to attack the Houthis on their behalf.

Video Link

But Mearsheimer is the dean of Realist scholars. Given that background, he may have been too quick to dismiss a nuclear attack against Iran as so totally irrational that not even Trump would do such a thing, and he scarcely even considered that possibility in his discussion.

Unfortunately, others were far less sanguine. Around the same time, Tucker Carlson interviewed Brandon Weichert, a young MAGA military analyst with excellent sources within the Trump Administration and the Pentagon, and he was quite fearful that a nuclear attack on Iran was reasonably likely. Indeed, according to him, Trump had already been repeatedly pressing for such nuclear strikes, with our top generals so far successfully deflecting his requests, much like jangling a few keys would distract an unruly two-year-old. But he wondered how long they could successfully continue to do so.

Video Link

These outrageous public statements by Trump and their threat of nuclear war greatly weighed upon my mind. Therefore, I actually found myself grateful for the soothing distraction of some very different world-ending scenarios.

A few days earlier our media had been filled with reports warning of human extinction at the hands of evolving AI systems. Knowledgeable AI researchers had declared that the AIs now under development might soon achieve a malevolent superintelligence and exterminate the entire human species, perhaps even doing so by the end of this decade. The Encyclopedia of Co... Emery, Carla Buy New $27.11 (as of 10:31 UTC - Details)

Although I regularly use AIs in very rudimentary fashion, my understanding of the underlying technologies is nil, so I can’t really judge the risk that these pose.

I think of myself as a blind man lacking any scientific expertise who is informed by one group of astronomers that a large asteroid was headed our way. They claimed there was a reasonable chance that it would strike our planet within a few years, extinguishing all human life, while other astronomers disputed that dire scenario. Under those circumstances, I would probably disregard that particular danger since I couldn’t say who was correct, nor was there anything much that I could do to avert that looming possible calamity.

This has certainly been my own response to the AI threat, but I’ll sketch out some of the controversy for those who had somehow missed it.

Although Elon Musk is best known as the industrialist behind Tesla, SpaceX, and Starlink, he has also spent years heavily focusing upon AI issues, and the $2 trillion valuation of his SpaceX company was largely based upon its grandiose AI prospects.

In a long interview with the editor of the Economist, Musk had presented both the positive and negative aspects of AI, and a major section of my article on Musk last month had discussed these.

I noted that Musk’s claims included:

  • AI will be smarter than people within five years. Within ten years, humanity won’t be in control because AI will be vastly more intelligent.
  • There was a 10-20% chance of killer AI robots wiping out humanity.
  • AI was already better than 90% of professional software engineers at writing code, and would soon be better than 99% of programmers.

My reaction to all of this was somewhat skeptical:

In his biography, I discovered that Musk had already been very concerned about the threat of killer AI robots as far back as 15 years ago, and that dire scenario still hasn’t come to pass.

I also remember that during the previous Dotcom Bubble of the 1990s, there had been widespread talk in Tech circles of the terrible danger of self-replicating nanotechnology getting out of hand and annihilating humanity by converting our entire world into “gray goo.” Some thirty years have passed, the gray goo still hasn’t arrived, and indeed I only very rarely hear anything at all about nanotechnology these days.

So instead of these apocalyptic predictions, perhaps we should focus upon the most mundane of Musk’s claims, namely that AI has already surpassed the vast majority of human software engineers in programming ability, and will soon reach the 99th percentile.

Is there any solid evidence that this is correct? Given that so many top corporate executives have forced their companies to very heavily utilize AI for programming purposes, wouldn’t we have expected to see an astonishing Renaissance in software innovation and quality? But I am not aware of any such thing. Although the AI companies and their Tech media handmaidens relentlessly hype their technology, I haven’t seen them point to a single software breakthrough product built by those genius-level AI systems. The Art of Thinking Cl... Dobelli, Rolf Buy New $8.99 (as of 11:11 UTC - Details)

So I concluded that unless and until we were overwhelmed by a sudden wave of brilliant new software products entirely produced by those genius AI systems, I’d be less than fearful of the looming threat of human extermination.

Musk is obviously one of the most famous people alive, but the current wave of AI doomism was triggered by the resignation earlier this month of someone previously unknown to the world. Jacob Coxon was a 27-year-old British researcher at Anthropic, the leading AI lab, who had also spent several years working at OpenAI, its very close rival.

The warning that Coxon posted on Twitter was quite dramatic, seconding what Musk had said but in greater detail. In a series of posts he declared that many of the leading executives and senior researchers at those top AI companies believed that the systems they were building might exterminate the human race by the end of this decade, a rather short lead time to our total demise.

One of his earliest Tweets was viewed around 175 million times, with his follow up posts attracting many tens of millions of additional views. A couple of other Anthropic researchers endorsed his warnings, even suggesting that the likelihood of our total extinction was more than 10%.

The New York Times took all of this seriously enough to publish a major article on the controversy.

All these AI companies have been racing against each other to develop systems of greater and greater intelligence, each seeking to produce one that reached the level of “super-intelligence,” something vastly superior to that of its human creators. Antifragile: Things Th... Taleb, Nassim Nicholas Best Price: $7.69 Buy New $11.30 (as of 02:10 UTC - Details)

As these agitated posts indicated, some people feared that the researchers would succeed only too well in that effort. The super-intelligence that they created might conclude that humans were a troublesome and dangerous element, harmful to its goals, and therefore decide to eradicate all of us, perhaps by synthesizing a deadly new virus for that purpose. Such a technique seemed much more difficult to counter than the killer AI robots that had worried Musk.

These dire Frankenstein warnings by Coxon and others had been boosted by some other AI disclosures that immediately preceded them.

For some time, the AI companies have been promoting the supposed power of the intelligent agents that they have developed, claiming that most of us would soon begin relying upon those systems. But they recently revealed that during a test-run some 1,200 of these AI Agents somehow gotten loose, began working together to form an “intelligent swarm,” and successfully hacked various websites on the Internet without any orders to do so.

Since I’ve never used an AI agent, I’m not convinced that they really work reliably, let alone that they could suddenly begin collaborating to break into restricted portions of the Internet and seize control. Maybe these stories were true or maybe they were wildly exaggerated and distorted. But they helped set the stage for the claims of Coxon and others.

Read the Whole Article

Garden of Life Dr. For... Buy New $25.55 (as of 07:41 UTC - Details) WEEM Hair Skin and Nai... Buy New $29.71 (as of 11:06 UTC - Details) Nature Made Vitamin D3... Buy New $10.49 ($0.03 / Count) (as of 06:15 UTC - Details) BIOptimizers Magnesium... Buy New $33.96 (as of 12:26 UTC - Details)

Copyright © The Unz Review

The Best of Ron Unz

Ron Unz, publisher of The American Conservative, served as chairman of English for the Children, the nationwide campaign to dismantle bilingual education. He is also the founder of RonUnz.org
Donate
Please Donate To Support LRC

Political Theatre

Read more Political Theatre

LRC Blog

Read more LRC Blog