From Acceleration to p(Doom) and Back Again: How AI may help and harm a grid in transition

Context:

At a DOE-sponsored conference  in early February 2024, we had a medium-sized room for a panel and workshop to consider whether and how much various AI technologies would help us solve pressing problems facing the North American grid, like the coming much-higher percentages of renewables and other present and looming operational and cybersecurity challenges.

My two highly credentialed panelists, Dr. Colin Ponce of Lawrence Livermore National Lab, and Dr. Christopher Lamb of Sandia National Lab, were both brilliant and eloquent. Their bios revealed a mix of deep thought in computer science, mathematics, operational cybersecurity and AI. Over the course of the 2-hours we spent together and with the 100+ folks who attended, in responses to my questions as moderator as well as those lobbed at them from the workshop participants, they both revealed themselves to be more sanguine about the prospects for integrating AI tech into grid systems (including in some cases, control scenarios) than I am. And they seemed to move those in the room, who in a show of hands at closing time revealed they were more positive about what AI will do for grid management, and/or less worried, than they were at the beginning of the session.

For you, dear reader, who weren’t in the room, or for those who were and want to partially relive the experience, I bring you the opening monologue and a few of the questions we worked through. Perhaps you will find some of what follows not only interesting, but helpful ... that’s my hope, anyway.

Opening Monologue: 

Imagine a boxing ring. In one corner is a school of thought that holds that AI research, development and deployment should proceed as fast as possible, unimpeded by excessive caution or regulation. Going by the name Accelerationism, and led by Guillaume Verdon, it:

Advocates for propelling rapid technological progress as the ethically optimal course of action for humanity. Its proponents believe that progress in AI is a great social equalizer. Followers see themselves as a counterweight to the cautious view that AI is highly unpredictable, potentially dangerous, and needs to be regulated.

In the other corner we have polymath Eliezer Yudkowsky of p(doom) fame. In a 3-hour podcast with MIT’s Lex Fridman shortly after the release of GPT-4, Yudkowsky makes a very well-supported case for AI as an existential threat to humanity, and that bad times are coming much sooner than later. As they approach the end of the interview, Lex asks him “what gives you hope?” To which Eliezer replies:

That I’m wrong.

How is this going to play out do you think? On an aging grid with many more renewables, with transmission congestion and capacity shortfalls, and according to the often smart if sometimes crazy Elon Musk, three times as much demand to serve by mid-century?

 We’re building renewables (and maybe soon, SMRs) to reduce emissions for climate reasons. We are talking about AI because our understanding of its apparent potential to help operate the grid in the presence of a much higher percentage of variable resources.  We’re also talking about AI because of what it seems poised to do for offensive and defensive cyber teams, including those defending and targeting the grid.

Questions:

Here is a subset of the questions that engendered the most conversation ... and there was A LOT of conversation. We didn’t get to all of them, and attendees volleyed quite a few of their own at us as well. You might try using these to stretch or shape your own thinking and socialize related concepts in the organizations you care about, including your own.

1. As you likely know, variations of AI/ML are already in use by utilities and other grid-serving entities to forecast weather and loads. Are you aware of other already-fielded uses of these technologies?
2. Where do you think generative AI will be most helpful in grid operations?
3. You may or may not be aware that a book is going to published at the end of this month with the title: AI: Unpredictable, Unexplainable, Uncontrollable. Does that sound right to you, or do you feel that the opposite will prove to be true, that AIs on the grid will be Predictable, Explainable and Controllable?  (Author: Dr. Roman Yampolskiy, U Louisville)
4. Are you aware of grid systems suppliers' stated intentions to begin introducing AI capabilities into their product lines? In particular, energy management systems, distributed management systems, DERMS, SCADA systems, etc.
5. Do you have security concerns about the extent to which Chinese suppliers dominate the market for inverters? And if yes, could AI play a helpful role in better securing the US grid as it comes to rely on an increasing percentage of generation, as well as ancillary services, from inverter-based resources (IBRs)?
6. Do you believe we will come to use and then rely on AI systems, trained on operator data, in control rooms as operator co-pilots?
7. In your opinion, could/should DOE be involved in testing AI-enhanced grid systems before they are approved to operate on the BES?