Eating the world ...

Caity PenzeyMoog

Whether AGI? Just a few months ago, the tech bros were talking about not only AGI but ASI as there was no end in sight for this open ended tech. Yours truly talked about it as well but for me, the jury's still out as there is no certitude, ever. Tech never sleeps and neither does AI, an open ended tech to remain forever unknowable as we will never know how it works as code has to write code in order for said tech to interact with the real world in real time. With this in mind, it seems GPT-5 is not the end all and be all as touted by Sam Altman, the CEO of Open AI.

Last month, the tech world turned with rapt attention to one man, OpenAI CEO Sam Altman, for the launch of the company’s new AI model, GPT-5. The atmosphere resembled the scientists waiting to see the first atomic bomb test at Los Alamos. GPT-5 was supposed to be another great leap forward toward the promised godlike telos of artificial general intelligence (AGI), a machine smarter than any human and possibly smarter than the sum of human civilization itself.

But the release of GPT-5 was a dud. Despite being relentlessly hyped for years, the most significant part of the announcement was the release of a model router that dynamically selects the correct “level” of computation power spent toward answering questions. OpenAI pitched the router as a tool for faster, cheaper answers for simple queries. But as writer Ed Zitron detailed on his Substack, this adds little value for the customer, while limiting most access to the best models, thereby lowering OpenAI’s cloud bill. To paraphrase Peter Thiel, we wanted AGI and all we got was a model router.

The AGI dream was pushed heavily by Altman, who wrote in one blog post in early 2025 that “we know how to build AGI as we have traditionally understood it,” and in another that the Singularity, a moment of apocalyptic transformation in which superintelligent machines would surpass humans, had already started, albeit in a “gentle” way. Yet online observers dubbed GPT-5’s release “Gary Marcus Day,” in honor of Dr. Gary Marcus, an eminent AI scientist who has been writing about the structural limitations and fallibility of AGI since his 2001 book The Algebraic Mind.

Why is this happening? Simple, AI's running out of content to analyze and scaling said content without knowing the problem the scaling is supposed to solve is a fools errand at best and the tech titans know it so ... in order to keep revenues flowing and cover up the problems of scaling, AI's creating it's own experiences based on the content already brought in system, leading to problems similar to inbreeding when said AI hallucinates and becomes rather unraveled to say the least. The other issue is power as processing analog neural net data with digital is ridiculously inefficient, leading to some pundits stating, as AI grows in power, the notion of eating the world in order to power same comes to mind.

An ouroboros in a 1478 drawing in an alchemical tract[1]

Resources like energy and water. Generative AI is using a lot, and demanding a lot more. I’d say that the growing appetite of Generative AI may prove to be unsustainable, but there’s a scary ambiguity in that word “unsustainable.” Will AI models have their supply of resources curtailed, or will we? Who gets fed first? What, or who, isn’t sustained?

Gen AI’s demand for energy, data, storage, and compute power is growing at a rate that can only be described as out of control. Meta used ten times the data and a hundred times the compute power to train its Llama 3 models compared to Llama 2, according to Big Technology. It swallowed up so much data, “it considered buying the publishing house Simon & Schuster to find more.” Eating the world.

AI “already uses as much energy as a small country. It’s only the beginning. The energy needed to support data storage is expected to double by 2026,” according to this article in Vox. The massive amount of energy AI requires for computation and data storage is matched by the “millions of gallons of water to cool the equipment at data centers,” and regulators are starting to talk about accountability.

Again, is this pace sustainable? Is anyone even pretending that it is? For that matter, do we have figures on how much energy AI is using? Per The Verge:

“Estimates do exist, but experts say those figures are partial and contingent, offering only a glimpse of AI’s total energy usage.

Curvature