Last week I wrote a column that tried to distinguish what we know (for sure) versus what we don’t know about the Jensen Huang scandals (the sex scandal, the corporate scandal, the national security scandal).

Alas, it seems as though my attempts at a careful taxonomy were hopelessly naive. In the face of this still unfolding affair, conspiracy theories immediately appeared from the near horizon like a swarm of locusts. I suppose the best that can be said about them (or at least some of them) is that they were and are remarkably inventive.

Perhaps (with the benefit of rose-tinted hindsight) there was a time when a scandal required a certain amount of patience. First came the revelation, then the investigation, followed by the discovery of facts and the publication of formal conclusions.

Only then it was a green light for the inevitable conspiracy theories explaining why the conclusions were wrong. It was a cumbersome arrangement, involving journalists, lawyers, investigators, occasionally courts, hearings and evidence.

Fortunately, technology has made the entire process considerably more efficient. We can now dispense with most of these steps – just make up a theory before the whole thing gets to first base, and stick with it no matter what.

Within days (hours?) an impressive collection of fanciful explanations had emerged. Huang had been removed by powerful interests seeking control of the AI industry. The American government was manoeuvring to nationalise artificial intelligence.

 The Times of India documented an even more ambitious effort at explanatory engineering – a mysterious Microsoft outage was connected to his departure, Chinese hackers were involved, perhaps even a rogue AI had something to do with it.

Exceeded my curiosity

I haven’t been able to find one implicating extraterrestrials yet (give it time), but there is apparently at least one blaming the Zionists (the tiresome prospect of trying to find it exceeded my curiosity).

Gizmodo reported that right-wing commentators had resurrected a 2024 interview in which venture capitalist Marc Andreessen spoke to Joe Rogan about government ambitions in AI. This was promptly offered as evidence that Huang’s downfall was part of a larger conspiracy to seize control of the industry.

The reasoning was – how do I say this – economical. Andreessen had discussed government involvement in AI, Huang had now lost his job, and therefore the two events must be connected. That nothing in the intervening two years corroborated the link, and that several simpler explanations were available, was apparently a petty administrative detail.

Of course, there is always the deep state, which, according to one prominent accelerationist quoted by the writer Robert Wright, was conducting “a full coup of Jensen”. Others on the same wavelength read the ouster as a warning to every other technology boss to fall into line. The deep state’s alleged motive varies by narrator. Its competence, unusually for a government body, is never in doubt.

Oh, and there is the one where Trump secretly has a piece of the action when chips make their way to China.

UnHerd reported that some in San Francisco are pointing at Dario Amodei of Anthropic, on the theory that the closed-model laboratories wanted rid of a man who favoured open ones. No public evidence supports this, a detail that has slowed the theory’s progress not a whit.

The beauty of this approach is its almost unlimited versatility. Should the price of Bitcoin start to rise, somebody will undoubtedly discover a connection to Huang. Should a Chinese satellite malfunction, the same theory can be adjusted accordingly. A conspiracy theory is rather like a modern software platform – once the basic architecture is established, additional features can be added without disturbing the underlying product.

There is, of course, a serious story underneath the entertainment.

Axios reported that prominent figures on the American right were using the scandal to challenge Donald Trump’s embrace of Silicon Valley. Steve Bannon declared that he had called Huang a Chinese Communist Party agent of influence “a thousand times”, describing the matter as an open secret. This is a particularly useful form of retrospective clairvoyance.

One knew it all along

Once something has happened, one can announce that one knew it all along. The absence of publicly demonstrated proof becomes evidence not of uncertainty, but of how successfully the truth had been concealed.

Meanwhile a petition to reinstate Huang has gathered a reported 189,000 signatures (backed loudly by Musk, and subsequently inexplicably endorsed by Trump). The Washington Post found that a good deal of the online chorus defending him appeared to be fake. Even some of the indignation at Huang’s fall from grace, it turns out, was itself an automated conspiracy.

It is perfectly reasonable to ask who benefits from Huang’s removal, or whether American intelligence assessments are accurate. It is considerably less reasonable to conclude that the presence of beneficiaries establishes a plot. Every large upheaval produces beneficiaries, and a beneficiary is not evidence of an author. Nvidia’s rivals gain, China hawks gain, short-sellers gain, regulators gain, and so, briefly, does anybody with a podcast or even a post like this.

And yet. Conspiracy theories have existed for centuries, and history contains enough genuine conspiracies to prevent their wholesale dismissal. Governments have lied, intelligence agencies have manipulated events, corporations have concealed wrongdoing, and powerful people have collaborated in secret. The problem is not the proposition that conspiracies occur. It is the conviction that every sufficiently interesting event must be one.

Social media has created an efficient marketplace for explanations that require no verification. Artificial intelligence has made the production of plausible narratives cheaper still. A few suggestive facts can be assembled into an apparently coherent account, furnished with historical parallels, decorated with technical terminology and flowcharts and images, and distributed to millions of people before the first investigator has finished interviewing the witnesses.

The economics are irresistible. Investigative journalism and congressional hearings are expensive, slow and frequently without a pithy one-line conclusion. Conspiracy production is cheap, immediate and offers the consumer something much more satisfying than uncertainty – an answer, a scapegoat, an instant revelation.

Political prejudices

Better still, the answer can be tailored to the customer’s existing political prejudices. In the attention economy, a tentative and nuanced conclusion is a display of weakness. Only wimps are tentative and nuanced.

It is hard to escape the obvious irony that the machinery now mass-producing confident, fluent, unverified explanations is often software running on the very hardware Huang spent three decades selling. He built the world’s most efficient engine for generating plausible sentences. It has now generated a great many about him.

Nvidia’s investigators (corporate and congressional) will report in due course. Their findings will be careful, hedged and footnoted, and will have the considerable disadvantage of arriving after everyone has finished reading the opinions of the uninformed, the devious and the lazy.

Whatever the investigators conclude, it will be taken as proof – if not of the facts, then of the thoroughness of the cover-up.

[Image: Ignacio Carreia for Unsplash]

The views of the writer are not necessarily the views of the Daily Friend or the IRR.

If you like what you have just read, support the Daily Friend


Steven Boykey Sidley is Visiting Professor at the Centre for Postgraduate Studies at Rhodes University, a partner at Bridge Capital and a columnist-at-large at Daily Maverick, Daily Friend and Financial Mail. His new book "Checkmate: How Shoprite Checkers rewrote the rules and won” is published by Jonathan Ball. His columns can be found at https://substack.com/@stevenboykeysidley