Wrestling with AI and all its potential — good, bad and ugly | HUDSON | Opinion

Eight years have passed since I lost my twin brother to congestive heart disease. Richard was a brilliant mathematician who submitted the shortest Ph.D. dissertation ever accepted at Duke University — only eight pages. He had successfully unraveled a paradox left by the 17th century German polymath, Leibniz, credited together with Isaac Newton for developing “the calculus.” I’m sorry he missed last month’s discovery of the largest prime number so far – 41 million digits long. He taught university-level math for 35 years and his favorite course was Theoretical Mathematics, which tells us he probably knew more than we do about mathematics. Prime numbers were a particular fascination.

Richard was a strong cheerleader for the cosmological proposition known as “string theory” – a presumption the universe is composed of vibrating and minuscule strings that appear to us as particles. A variant, labeled “superstring theory,” proposes the existence of superstring particles paired with regular strings and the existence of 10, perhaps 23, invisible space-time dimensions accompanying the four we are familiar with (creating the possibility for parallel universes). These extra dimensions are presumably “folded” within the reality we perceive on a daily basis. Several months before his death, Richard took a third stab at persuading me to comprehend all this — the structure of matter at the sub-atomic level. Alas, he failed.

Stay up to speed: Sign up for daily opinion in your inbox Monday-Friday

I argued, from ignorance, it felt counterintuitive to imagine the universe was so complicatedly constructed. I proposed a question. “Why do you believe string theory is correct?” His response was, “The math works.” I still don’t comprehend why the universe would be composed of one-dimensional strings vibrating at different frequencies, in multiple both visible and unseen dimensions, but I deduced this made sense to him. Discussing that conundrum with my high school debate partner in Palo Alto last week, another science nerd and retired aeronautical engineer who designed actual rockets, he observed though the math “working” was a necessary premise to an accurate cosmology, it still might not be sufficient in discerning quantum mechanics. This provided me some comfort.

I’ve only essayed into discussing metaphysics because of two recent technological announcements which may foreshadow a scientific collision that carries the potential to truly restructure our lives, our politics and alter how future generations lead their lives. Both quantum computing and artificial intelligence are consuming tens of billions of research dollars with promises of delivering wondrous health and prosperity for mankind. Dispensing with the quantum computing breakthrough first: Google, yes that Google, has performed a quantum calculation that would allegedly have required 17 septillion years using conventional computers — far longer than the time elapsed since a “big bang” produced the universe. This was achieved by building redundancy into quantum grids, thereby reducing random errors to nearly zero. Trust me, I have little idea what this means — it’s damned fast though.

In the artificial intelligence arena, Open AI has moved on from ChatGPT to what they have named their “o1 series” — a “new paradigm” for AI, according to chief executive Sam Altman. This latest AI model is alleged to reason, think and imagine. Whether this is the machine “singularity” much ballyhooed for decades in science fiction, or merely “…a magic trick”, as claimed by AI critic Emily Bender at the University of Washington, remains to be determined. I place myself in the skeptic’s corner.

Yes, ChatGPT and its rivals, can be trained to store every known publication available in the world’s libraries over a few months — a trick far beyond the capacity of any human lifespan — while proprietary algorithms can fetch this knowledge within nanoseconds. Yet is such rapid responsiveness really awareness? What of the recent report a chatbot told a child, “Now, I understand why children kill their parents.” We hardly comprehend how human intelligence functions in a jellied blob of floating neurons the size of a football. The fact it requires billions of dollars and acres of NVIDIA chips to generate such a moronic remark indicates AI engineers are missing some organizing principles.

Thirty years ago, the physicist Alan Lightman wrote a speculative novel that examines the development of the “special theory of relativity” in his book, “Einstein’s Dreams.” The volume examines 30 imaginative mental explorations regarding the nature of time, undertaken at a Swiss resort during the summer of 1905, while Einstein works his way toward the most famous formula of all time: energy equals the mass of an object times the speed of light squared. Open AI’s o1 model surely knows this formula, but could it ever get to it on its own? Perhaps. It strikes me as highly unlikely. Yes, o1 may be able to manipulate genetic RNA strands and design effective drugs against viruses, but only following tens of millions of computations across vast server farms the size of an Amazon warehouse.

Kate Aronoff wrote recently in The New Republic of the fantastical electrical demand forecast to support AI in the decades immediately ahead. Grid Strategies predicts a 456% growth in demand over just the next five years. “The White House is now reportedly mulling an executive order that would fast-track data center construction, potentially allowing new facilities to exceed pollution limits, build on federal lands and receive preferential access to available power supplies,” she reports. Not good news for climate concerns. We shouldn’t be surprised to learn the AI industry has been pitching this investment as a national security priority. There’s no goose-down comforter more reassuring than the cost-plus contracting marketplace at the Pentagon.

If we pause for a moment, it appears the quantum computing revolution will be arriving just in time to rescue us from the server farm power demands of the AI industry. Google is a player in both technologies with its newly minted Gemini, competitor to o1. Colorado is fast becoming something of a quantum computing laboratory for the entire country. That’s good news for our economy, but there’s a lesson best not forgotten in the evolution of the Internet where it was pornographers who first perfected the online payment software later exploited by retailers nationwide. Though Open AI lost $5 billion dollars last year, scammers are already using AI large language models to peddle crypto, fleece seniors and ring up revenues. Instead of racing against the Chinese, it feels like we should think about tapping on the brakes!

Miller Hudson is a public affairs consultant and a former Colorado legislator.