AI Taps Employees for Thoughts and Beliefs, Plus Thoughts on Mania Hangovers

Let’s have a look at some of what’s happened this week. A lot has happened, more is happening now, and it’s reasonable to expect that the flood of significant and insignificant events will continue to overwhelm us. You can only monitor so much of what transpires, and you are only capable of analyzing and understanding a subset of everything you observe. When the imponderables of time and space meet the constraints of cognition, something has to give.

All the same, we do our best to keep pace, though others do their worst to make it difficult for us to pierce the veil of obscurity.

Speaking of obscurity, I see that Elon Musk is still under the delusion that he, in his capacity as chieftain of SpaceX, is leading an AI company rather than a satellite-communications and space-exploration company. If the market continues to reward him for this glaring category error, willful or otherwise, he will persist in the charade.

Musk’s delusion is apparently not absolute, however, because Musk seems to appreciate that his AI capabilities need considerable upgrading. Depending on one’s generosity or perspective, Musk promised or threatened SpaceX employees that they would “effectively be the parents of the AI,” because Grok, his AI alter ego, would inherit their “thoughts and ideas and beliefs.”

Proud AI Parents?

Now that we’ve opened that bottle of magical thinking, let’s give it some time to breathe. Imagine that you are a SpaceX employee hearing and processing Musk’s statement of intent regarding Grok’s ingestion of your thoughts, ideas, and beliefs. Would you be flattered and honored at the prospect, or would you be appalled and horrified at the privacy intrusion and exploitation of your intellectual and psychological essence?

Some employees might respond instinctively or reflexively, one way or the other, while others would perform a risk-reward calculation before rendering a verdict. Regardless of what you’d decide, you might agree with me when I say that Musk is asking rather more of his employees than their contracted labor during work hours, however the latter might be demarcated.

He wants their hard work and professional commitment, sure. That’s a given. But he also wants nearly everything else he can extract from them. This is one way that AI, whatever else it might be doing, is changing the relationship between employer and employee. We should mark this change. Musk is not (and will not be) the only employer to perceive paid employees as cognitive husks that can be scooped out continually to feed the voracious appetite of LLMs. A question: Should employees be considering separate contractual arrangements for this exchange, which goes beyond the traditional labor-value proposition? Give it some thought.

Revolving Door at OpenAI

OpenAI continues to suffer an exodus of key employees. Perhaps the timing could be worse, just ahead of an aspirational IPO, but I struggle to imagine how. Not all of these employees are jumping from the deck of a burning ship — some are leaving for personal or health reasons — but the overall picture is concerning for anybody with an investment, or a prospective investment, in the company. I suspect that OpenAI erred strategically and tactically by placing too much early emphasis on high-profile, low-return consumer AI use cases instead of more practical business applications. I seriously doubt that many people want to pay for an “AI life coach,” but AI fulfillment (even partial) of closed-loop, back-end processes and workflows seems more viable.

AI likely has very real limitations, as we’ve discussed many times before. Screenwriter and filmmaker Aaron Sorkin feels good about AI, but that’s because he doesn’t foresee it working particularly well in creative contexts. I broadly agree with his assessment.

Just look at what AI algorithms have done to social media, encouraging everybody to make the same sorts of posts, with the same formal and structural tendencies, the same “personalized” video formats, and the same attempt to simulate authenticity.

Think about that: simulating authenticity. Simulated authenticity is a contraction in terms. Something is either authentic or it isn’t. The attempt to simulate or fake authenticity makes your output, by definition, inauthentic. AI’s algorithmic logic tends to gravitate toward a stultifying mean, a homogenous cultural mush, weak and insipid. Maybe AI will get better at simulating human output, but it’s got a long way to go. Besides, the inherent simulation precludes meaningful approximation of human authenticity.

Elation Before the Storm?

Michael Burry has déjà vu as he considers the current AI exceptionalism. He recalls a similar sense of imperviousness and indestructibility that preceded the bursting of the dotcom and subsequent bubbles. The mood is similar, no question, but correlation does not prove causation. Even if there were a causal connection between an unreasonable, untamable optimism and market implosions, the timing of the latter would be difficult to predict.

Still, I have a confession to make. I was one of those people who believed, back in the late 90s, that the Internet (it was authoritatively upper-case back then) was a major watershed, a stark dividing line between what had been and what would be. As the Internet and online web commerce gained velocity, the old rules that applied to the market and the economy (and much else besides) seemed to have been banished to the pages of dusty historical tomes. That was delusional thinking, and I got caught up in it, even though I would say that I make every effort to remain grounded and relatively rational.

That’s why I have empathy and even sympathy for those who have fallen under the mystical sway of AI exceptionalism. When you’re living inside the exhilarating rocket ship, hurtling faster and higher toward new frontiers, you are intoxicated by the seemingly limitless possibilities.

You might even have a significant stake in the expanding commercial success of the technology, as many do today with AI. The exhilaration of the chase, combined with the prospect of life-changing prosperity, is a potent cocktail. The metaphor of drunkenness and inebriation is apt, because manias reduce your inhibitions and impair your assessments of real-world circumstances. Boom psychologies, as they have sometimes been called, are powerful forces; they result in denial or disregard of anything that doesn’t comport with a continuation or intensification of the good times. It’s obviously delusional thinking, which always puts you at risk of being sideswiped by reality when you least expect it.

Slippery Perspective

Now, let’s back up. The Internet did have a major influence on much that followed. It did profoundly change the way we live, communicate, work, and play. In the clarity of hindsight, much of that change was positive, but some was not. There are things I miss about the past, not out of some deranged nostalgia, but because there was a more tangible sense of community and togetherness in some of those rapidly receding or lost conventions and customs. In ways that I can’t fully understand or explain, many of us have become less reasonable, more adversarial and confrontational, and more petulant and small-minded as more of our lives unfold online. There’s something to be said for decorum, a word that rarely surfaces these days.

AI also appears destined to be with us for a long time. Even though AI hype has galloped well ahead of any material AI reality, that doesn’t mean that AI will not have its uses, some of which might not be as epochal as those that the Internet gave us. People will use AI in various ways, good and bad, if only because people are good and bad.

A bursting of an AI bubble, or whatever you want to call any implosion that is likely to follow the current AI mania, does not mean that AI has no future. All it means is that we will have failed to keep our perspective — yet again. I’ve been there, and I feel your elation and anticipated pain.