Unlike the human mind which is capable of understanding, on the energy intake of substantially less than a chocolate bar that German soldiers during World War 2 were unlikely to be of perfect complexion and exotically bestowed with the genomic profile of a Hercules crossed with Richard Feynman, AI is hugely, enormously, grossly some might say, inefficient.

In contrast, each of our gloriously efficient minds are capable of continuously generating a multitude of observations, opinions and imagings most of which most of the time are prone to increase our chances of survival. This should come as no surprise given the ruthless mechanisms of evolution that have been applied to its development.

The modern corporate mantra of diversity and inclusion purports to celebrate this heterogenicity while of course at the same time acting to introduce new restrictive orthodoxies, manifested in that most dangerous of phenomena, "Groupthink".

Think - George W. Bush's foolhardy and disasterous invasion of Iraq in 2003, driven by a fantasy narrative cooked up by the best and brightest within the Beltway.

Think - the 2008 financial crisis built on the assumption, among others, that people with no income and no assets would predictably repay their borrowings until their financial obligations were fully discharged.

Think - Google AI's imagery.

"Groupthink" another word for a collective insanity therefore.

Perhaps the most cynical orchestration of the "Groupthink" phenomenon in recent times in the pursuit of political power is Tony Blair's "Third Way" in which a path of unremitting consensus is to be conducted, creating the "good Groupthink", of which he and his cohort are the arbiter, relegating any opposing view to the irrelevant and potentially subversive.

Is AI the latest manifestation of "Groupthink"? Given that the definition of the outputs of any such systems are likely to be proscribed rather than prescribed then the tendancy must be to be so. The whole point of AI is that it must conform.

So how many conforming AI's are we going to have? Given the enormous energy requirements demanded by this technology then over time, not very many, as competative pressure applies to supply. In fact, it could easily be one.

Probably the only mechanism on the horizon, perpetually the far horizon thus far in its history, for the delivery of sufficient energy to support a variegated set of AI manifestations and save us from the autocratic ambitions of Silicon Valley is nuclear fusion.

If stories such as

https://www.wired.com/story/deepmind-ai-nuclear-fusion/

are to be believed then ironically it maybe a contribution from AI technology that helps save us from the increasingly monotheist thrust of "Big Tech".

References

https://www.scientificamerican.com/article/the-ai-boom-could-use-a-shocking-amount-of-electricity/

https://www.technologyreview.com/2023/12/01/1084189/making-an-image-with-generative-ai-uses-as-much-energy-as-charging-your-phone/