Scotty, We Need More Power! - 10/9/24

Hey man, where do I plug this thing in?

One of the interesting challenges emerging in the race for artificial intelligence development is its huge requirement for energy. In a world full of rickety power grids and uncertain sources of supply, this is becoming the key driver for companies betting their futures on winning the AI race.

Uninterrupted power supplies are critical. Periodic rolling blackouts or energy rationing aren’t options and these things are sucking up more and more juice as companies continue to expand their computing capacity.

So how will the big players meet these challenges? Well, if you’re Microsoft, you go nuclear—literally. The company announced last week that it won’t be waiting around for a world of energy abundance from the current combinations of fossil fuels, solar, and wind. 

Instead, Microsoft has purchased its very own nuclear facility—and they’re not delaying the project for the years it would take to construct a new plant utilizing the latest technology. Instead, Microsoft plans to recommission the infamous Three Mile Island Nuclear Generating Station on the Susquehanna River in Londonderry Township, near Harrisburg, Pennsylvania.

For those who were around back in 1979, you might recall the reactor accident that began at 4:00 a.m. releasing radioactive gases and iodine into the environment. It is the worst accident in U.S. commercial nuclear power plant history. On the seven-point logarithmic International Nuclear Event Scale, the TMI-2 reactor accident is rated Level 5, an "Accident with Wider Consequences.”

After the disaster, the facility continued to operate for decades but was eventually deemed too dangerous to continue without significant upgrades which, at the time, the owner couldn’t afford. The event also heightened anti-nuclear safety concerns among the general public and led to new regulations for the industry while accelerating the decline of efforts to build any new reactors. Three Mile Island was finally shut down in 2019.

Now, Microsoft has purchased its very own source of guaranteed, uninterrupted power for the sole purpose of feeding its rapidly accelerating AI energy requirements 24/7—something unprecedented. And that brings up some interesting questions, such as how this problem might be handled by its competitors, as well as the subsequent impact on the overall energy supply. 

As energy demand ramps up significantly due to AI, will the overall consensus on nuclear change as well? And will the general push to reduce dependence on fossil fuels lose some momentum as current production supplies are squeezed? The one certainty is that, in our present state of technological capabilities, we’re barely able to keep up with the growing demand that exists even without these new requirements. Here in south Texas, we're regularly subjected to power outages when supplies become stretched.

I’ve discussed in prior letters my opinion on the knee-jerk reactions to abandoning nuclear here in the U.S. and certain parts of Europe. Although well-intentioned, it’s very clear that we are years away from fulfilling the world’s energy requirements with non-fossil fuel sources like solar and wind. It goes without saying, that everyone wants the cleanest energy possible and of course, a sustainable environment. 

Currently, nuclear still offers the most viable option. Yes, there are problems with how to handle the inevitable radioactive waste produced. And of course, systems failures are a possibility in any high-tech endeavor, famously having already occurred at Russia’s Chornobyl plant, Japan’s Fukushima facility, and Three Mile Island. 

But it’s important to remember that all of these utilized outdated technology. And in the case of Fukushima, well let’s just say it probably wasn’t a good idea to build it in an area susceptible to regular earthquakes and subsequent tsunamis. 

Weighed against the myriad of problems and exorbitant costs associated with things like the Green New Deal, nuclear makes more sense than ever before. I honestly believe it’s a foregone conclusion that we’ll advance this technology and the world will be a better place with it.  

Meanwhile, last year Germany shut down their last nuclear plant. The decision was made years prior, I’m guessing based on an overly optimistic assumption about the evolution of green energy production. But that hasn’t worked out and now Germany is dependent on France to meet its country’s needs. Ironically, France produces 68% of its energy from—you guessed it, nuclear. 

As we find ourselves hurtling toward a world where energy demand is increasing exponentially, the inevitable consequences of falling short on supply seem to indicate a need for some bolder innovations going forward. Assuming AI achieves self-awareness/sentience/super intelligence, it doesn’t appear likely it will allow itself to starve, particularly once fully integrated into the various infrastructures and systems that it will no doubt inevitably control. 

One could easily envision a scenario where it could redirect power sources in an effort to keep itself adequately fed, possibly at the expense of other critical systems we humans depend on.

Who knows, maybe this is one of the challenges artificial superintelligence will solve in an effort to ensure its very existence. Hopefully, it won’t exclude us from the solutions.

Sam Altman’s an optimistic man

The president of Open AI, Sam Altman, announced recently that the company will be transitioning from a non-profit to a for-profit corporate entity. This is particularly interesting because Sam and his crew have become synonymous with the recent AI explosion, mainly through their ever-evolving Chat-GPT iterations. 

In the past, Open AI’s public mission statement to develop AI only for the good of humanity resulted in a somewhat elevated status among their industry peers, particularly among AI ethicists and, more importantly, our elected leaders. That fact was on display when Altman was asked to testify before Congress in an effort to explore the dangers of AI and how they might be mitigated. Sam was even asked by one high-profile senator if he might be interested in leading the effort to draft appropriate legislation designed to reign in the explosion of technological innovation that seemed out of control. 

I watched those hearings closely and the main takeaway for me was that the average person serving in Congress probably doesn’t know how to turn on their laptop, much less understand any realistic strategy for mitigating the potential existential threat of AI through legislation. 

The idea that laws of any design could result in realistic controls over the myriad of potential threats from AI seems naive. But even if some type of legislative governor could be conceived and implemented that would effectively throttle the speed of the AI revolution, it seems a foregone conclusion that our adversaries overseas would ignore it, therefore gaining a significant advantage in everything from trade to military superiority. 

Now that reality has pierced the thin bubble of virtue Open AI enjoyed with its non-profit status, all those remaining with the company after a mass exodus will become even richer than before—and none will benefit more financially than Sam Altman. 

I’m not entirely sure anyone expected a different outcome and at least now we don’t have to pretend that the allure of great power and wealth isn’t a central driver in the AI race. I don’t even think that’s necessarily a bad thing. 

However, it does bring up the question of why Open AI declared itself a non-profit company in the first place. There had to be some perception that its core values of “good over profit” ensured a level of safety checks and balances that wouldn’t otherwise exist. 

As the company announced its transition, Altman again made headlines for a recent essay in which he outlined his vision of the future—one where artificial superintelligence (that surpasses artificial general intelligence as well as human intelligence) will likely arrive “in a few thousand days.” As he raises billions from investors for the newly for-profit Open AI, he’s simultaneously hyping its potential as the front-runner in the superintelligence race. 

Regardless of the company’s motivation, Altman’s grand vision seems refreshingly optimistic. The OpenAI chief predicts AI assistants will become increasingly capable, eventually forming "personal AI teams" that can help individuals accomplish almost anything they can imagine. 

He predicts AI will enable breakthroughs in education, health care, software development, and virtually every aspect of our world. Eventually, all human needs will be fulfilled for their entire lives. Working for a living will be an option, not a requirement for survival. Universal health care and income will become the norm.

And in the near term, as the population ages, the market for robotic caretakers could offer living assistance as well as companionship for millions as life expectancies grow with new AI developments. The future for humans in their last chapters of life could look very different than in the past. 

But what does a world where humans are taken care of from the cradle to the grave really look like? It might not necessarily be a net positive for many. Our innate desire to succeed, provide, accomplish, and excel are critical aspects of our human nature. They won’t just disappear no matter how pampered our lives become. It’s a theme I tried to explore in my book “Entanglement Realms.” 

Monster storm

Thoughts and prayers for those living in areas that have been ravaged by Hurricane Helene. And now, Milton appears to have developed into a superstorm, perhaps the most destructive in decades, and is heading straight toward Tampa and Sarasota. 

With all our technological advances, we’re still regularly humbled by how vulnerable we are to the incredible forces of nature.

Give me a holler and let me know what's happening in your part of the world. Always great to hear your input!

Best…..VB