Advanced Search
May 21, 2024
It seems like almost every day there is another story about new advancements in Artificial Intelligence, or AI. By now, many of us are familiar with ChatGPT, but there is a wide variety of different models and applications for the rapidly-evolving AI technology. To wrap up Season 7 of The Climate Conversation podcast, Dan and Alison are joined by Helena Fu, director of the Department of Energy (DOE) Office of Critical and Emerging Technologies. Helena discusses how AI can help modernize the power grid for a clean energy future and shares some of what DOE’s newest office has accomplished since it opened in December 2023.
Show notes:
Sign up for our bi-weekly newsletter, Climate Change Solutions, for insight on the latest innovative climate solutions and environmental policy in action. Follow us on social media @eesionline
About this Podcast:
With all the depressing climate news out there, it’s sometimes hard to see progress. The Climate Conversation cuts through the noise and presents you with relevant climate change solutions happening on the Hill and in communities around the United States.
Twice a month, join Environmental and Energy Study Institute staff members as they interview environmental, energy, and policy experts on practical, on-the-ground work that communities, companies, and governments are doing to address climate change.
Whether you want to learn more about the solutions to climate change, are an expert in environmental issues, or are a policy professional, this podcast is for you.
Episode Transcript:
Daniel Bresette: Hello, and welcome to the Season Seven finale of the climate conversation. I'm Dan Bresette, president of the Environmental and Energy Study Institute. And with me on the solemn occasion, the end of the season is my co-host, Alison Davis. Hey, Alison, have you been?
Alison Davis: I've been good. Although I am also a little bit sad that it's the last episode of the season before our summer hiatus. Don't worry, we will for sure be back in September with Season Eight. We've already started planning for that. But today, we're going to wrap up Season Seven with a conversation about artificial intelligence or AI, which is something that's been coming up a lot in EESI’s content recently. In our previous episode, actually two episodes ago, we learned how AI and machine learning is being used for making vegetation models and mapping spatial ignition probability, which is useful for wildfire modeling. And if you follow our congressional briefings, you've also heard about how it can help enhance weather forecasting and also dam infrastructure. So for this episode, we're actually going to look at the potential of AI to catalyze clean energy deployment in the United States.
Dan: For those of us who don't work on artificial intelligence every day, you know, as scientists or technologists or engineers, a lot of times our impressions of what this actually looks like, it's been pretty warped, by the way it's been portrayed in popular culture. So for a certain generation, I'm sure they're thinking about 2001 A Space Odyssey, my generation is more of The Terminator. But in reality, artificial intelligence is not really either of those things. But what exactly do we mean when we're talking about artificial intelligence in the real world? The legal definition of AI is a machine based system that can for a given set of human defined objectives, make predictions, recommendations or decisions influencing real or virtual environments. Artificial intelligence systems use machine and human based inputs to perceive real and virtual environments abstract such perceptions into models through analysis in an automated manner, and use model inference to formulate options for information or action. An overly simplified definition is that it's a computer with the ability to mimic human ways of thinking, and it happens to learn really, really fast.
Alison: The Department of Energy launched its office of critical and emerging technologies in December 2023 to harness the potential of AI, as well as other innovative technologies like biotechnology, quantum computing, and semiconductors to advance our clean energy future. This action was prompted by an executive order from the Biden Harris administration in October, with the goal of ensuring US leadership in the responsible advancement of AI. The new office has kept busy in its first six months and just launched a major report titled, “AI for Energy: Opportunities for a Modern Grid and Clean Energy Economy,” which will be the focus of our conversation today.
Dan: Join us for that conversation, here comes Helena Fu, director of the DOE office of critical and emerging technologies. In her previous role as director for technology national security at the White House, Helena was responsible for advancing strategic technology cooperation with allies and partners. She has also served as director of international science and technology cooperation and trusted research from the DOE Office of Science and as senior policy advisor at the White House Office of Science and Technology Policy. Before that holiness spent several years at the US Embassy in Beijing serving as deputy director and later as energy attache and director of the DOE China office. Helena, welcome to the show. It's great to see you.
Helena Fu: Thanks so much for having me. It's a pleasure.
Dan: I was fortunate enough to be part of an event that you were also part of a couple of weeks ago, the AI powering the New Energy era Summit. Unfortunately, our panels were up against each other. But I did get a chance to see some of your presentation. It was really excellent. It was the US leadership at the intersection of renewable energy and AI. And I think that recording is available online, maybe we can pop a link to the show notes. But I want to begin our conversation today by kind of backing up a little bit and just asking, you know, what is it about the potential for artificial intelligence to help deliver climate solutions in the clean energy space that makes you so excited for the work that you're doing at DOE?
Helena: I think that DOE really sits at the intersection of two really interesting questions, right one around what the potential of AI could be, especially as applied to the energy sector, but also to our science and national security missions. And what is the energy consumption of this particular technology and what the potentials are there? Doe is part of the AI executive order that was issued at the end of October last year is really charged with a number of really critical tasks. There's a lot there around AI safety and security. And DOE has a lot of activities underway primarily through our National Nuclear Security Administration to evaluate models to do red teaming around chemical, biological, radiological and nuclear risks. But we're also tasked with promoting innovation. And we were really happy last week to really dive in and release a number of really exciting things around both AI for energy, which we're talking about right now, as well as other tools and foundation models that DOE is developing to advance our basic and applied science mission. So it's a really exciting time, I think, obviously, AI, really, in the last few years has really come on to the main stage with the advent of these new tools, the generative AI and the large language models, we think there's a huge opportunity space for scientific AI, the four foundation models that really speak the language of biology, chemistry, or physics to help solve some of our biggest scientific problems that will also help us address energy security, our climate crisis, and really enable the broader deployment of clean energy technologies at scale.
Dan: Artificial Intelligence seems to be, you know, sort of part of the vision of a future that a lot of people are presently envisioning. But when it comes to clean energy infrastructure in the US, we're dealing with infrastructure that's been around for a long time, we're dealing with the past and the present. And the AI for energy report specifically notes that the earliest grid components that we are currently still using in some cases were built more than a century ago. And the long lived nature of grid assets means that things that we're building today will be in use for a very long time, you know, things built 40, 50 years ago, before what we would consider modern information technology management capabilities, are still relied upon for a safe, reliable and affordable grid. How can cutting edge AI technologies available today be applied or integrated with older pieces of infrastructure currently in use across the power grid?
Helena: That's a great question. And, you know, I think the grid itself is one of the most complex machines that we've ever built. And different stages of modernity, as you just said, you know, some parts of the grid are very, very old, some parts are just coming online now. And, you know, we will need to figure out ways to address the increasing scale and complexity of the grid. And I think AI is one tool, not the only tool, but one tool that can help with a number of different kinds of applications. And the recent AI and energy report that we just put out really looks at the opportunity space here, both in terms of understanding the modeling the capacity and potential transmission capabilities, to look at how we can use AI to help with compliance and review for a federal permitting. One of the things that we also were really pleased to announce is something called this new voltaic initiative with AI as part of the the name of the initiative that really looks at how we can use these large language models to look at environmental impact statements, things that, you know, consultants spent a lot of time and a lot of energy building into, you know, a 200 or 300 page report, which then after the project is over, goes on a shelf and is not used again. So how do we make sure that we can bring those efficiencies into the process to accelerate the deployment of clean energy technologies, we think AI can be used to help forecast both in terms of renewable energy production for grid operators and weather forecasting for renewable energy production. These are things that our labs have been very, very engaged on in partnership with industry, as well as Smart Grid applications of AI. We think there's a lot of opportunity in that space to to look at great enhancing technologies like topology, optimization and dynamic line ratings. I think these are things where there's a lot of opportunity space here. I would also note that our office of cybersecurity energy security and emergency response put out a public assessment of work that they had also done in conjunction with AI executive order, because as the sector Risk Management Agency for the grid, they are really looking at how AI can be used to basically where the potential risks AI can pose to the grid, as well as how AI can be used to counter those potential risks. So they're looking at it from a cybersecurity perspective from potential on unintentional failure modes of AI as well as more intentional failure modes of AI. This is a really important line of work, and one in which our Cesar office is really looking to engage further on over the course of the year in partnership with utilities and other stakeholders.
Alison: So the report mentions virtual power plants or VPPs. And that's something that I was interested to read about because it was mentioned by one of our podcast guests last season from do we actually Jonah Wagner in the loan programs office. So for our listeners who couldn't catch that episode VPPs are collections of distributed energy resources that can balance electricity demand and supply, just like a traditional power plant. And doe estimates that they can handle potentially up to 20% of peak demand. But as of now, the process of enrollment can be a bit complicated. So I was hoping that you could explain a little bit some of the ways that AI might make it easier for consumers and utilities to enroll in VPPs.
Helena: Firstly, I will say, Jonah Wagner was at OCP, for a part of the time where we were working on this an Energy Report, he was a wonderful partner there and is a great partner now that he's back in the department. And we're really glad to have him back. When I would say on VPPs. Here, you know, the AI and Energy Report does talk about the opportunity space here. We talk about how can be used to optimize revenue for VPP providers while also providing a customized option that's based on user preferences. And I think that customization is a really important word. We think that virtual power plants will be able to enroll different kinds of users to reduce load based on the price signals. But every user will have their own different preferences for for example, you know what the temperature of the AC should be at some people run hot, and cold. And we really think that AI can help bring that personalized experience and allow for more people to participate in this process, as well as potentially help optimize how we respond to the demand response market signals. And in terms of how we get more people to enroll, I think that's also an opportunity space to educate consumers about what their options are to show that, you know, it can be a personalized experience. And that this is one way in which people can make good choices to help with a larger climate issue while still maintaining their own preferences here.
Alison: As you mentioned a little bit earlier, AI has a lot of potential to help decarbonize the US economy, but it is not itself a Climate Neutral technology. There are some concerns which were noted in the AI for energy report that there are emissions and electronic waste associated with using AI, I was wondering how your office is addressing its high energy consumption and other impacts that it might have on the environment.
Helena: So what I would say is, this is not something that my office specifically is addressing this is something that the Department of Energy as a whole is very, very closely tracking. What I will say is, you know, the US has historically managed to double the projected rate of electricity growth in the 1990s to the early 2000s. So we've done this before, what I will also say is that we have been projecting growth in power needs, right? This is not just a data center issue or an AI issue. This is broader electrification of the economy, manufacturing coming back to the United States and electric vehicles, and many different loads that are coming online. Data Center growth right now is a very small percentage of current electricity demand. It is a growing area of electricity demand. And so we really need to be thoughtful about how we address this issue. We also recognize that it's not a nationwide issue, right, there are areas and regions where they're experiencing much more load growth, we were really pleased to be able to announce a number of activities. Last week about how we are thinking about this issue, it's clear that we're going to need to get a granular understanding of where this load growth is happening. The R Lawrence Berkeley National Lab is currently engaged in a study on datacenter power demand. So that's already ongoing. Our secretary charged her energy advisory board to also look at this issue and report back with recommendations in June. So that's coming up very soon. We also were really pleased to announce upcoming series of convenings, where we look to use our convening authority to bring together different stakeholders in areas of high load growth. So planning for that is currently underway. But besides understanding the issue, I think we really see an opportunity to continue to deploy at scale, currently available technologies, clean energy technologies that are available to come online on the grid. And we have many tools that we didn't have back in the 1990s in the early 2000s to do that right through the bipartisan infrastructure law and the inflation Reduction Act, trying to remember all the acronyms here. Yes, through BIL and IRA there are tremendous opportunities. And we really want to make sure that people are taking advantage of those opportunities. And people are aware of what is possible so that we can continue to support the deployment of clean, firm power generation and advanced grid technology. So we were really, really focused on that. And there are a number of offices at the Department of Energy that are laser focused on that. And then finally, DOE has long been an innovator and driving innovation in efficient, more energy efficient AI, hardware and software. And I don't think it's very widely known that DOE operates for the fastest supercomputers in the world, including the world's fastest, and that we take a lot of care to make sure it is highly, highly energy efficient. And in fact, it was really our investments and co-development, co-investment very early on with industry that resulted in the development of critical components of GPUs that are powering today's AI revolution. And really, the limiting factor at the time was energy, we needed to be able to have a highly efficient power envelope for these kinds of hardware that still maintained performance. And so that was something that was really a huge achievement out of the Exascale Computing Initiative, where we were able to develop these deep industry partnerships to drive innovation in the sector. And so what I would say here is that, you know, we think that there's a huge opportunity for additional innovation, both in hardware, but especially in software, and our Office of Science and ARPA e are really, really engaged on this particular issue, as well as our Office of Energy Efficiency and Renewable Energy.
Alison: That is so cool and so exciting. And it seems like it's part of a cycle of technological innovation that we've seen before. New technology has reshaped the job market many times in modern history, it's happening again, for sure with AI, and this cycle can cause some anxiety about loss of employment. But on the flip side, it is so exciting to think about the new types of positions being created and all the opportunities. How do you expect the deployment of AI technology to affect career opportunities in the clean energy workforce?
Helena: You know, much of clean energy deployment requires skilled labor, offshore wind technicians, EV charger, technicians, electricians, and that need will continue to exist. And I would really point you to, you know, the do E's Office of Energy jobs, they've been doing an amazing work on this specifically. And many of the students that are learning AI today will want to use their skills and talents for real really meaningful purposes. And there are so many exciting applications for how AI can be used as a tool, not as a replacement. But as a tool to innovate in the clean energy space. One of the things that doe was charged with in the AI executive order was to coordinate with NSF but to develop, you know, a pilot program for training 500 new additional researchers by 2025, that would be capable of meeting the rising demand for AI talent, I'm really pleased to say that we already are now exceeding that 500 number. We launched a website to really detail all of the various opportunities for training and upskilling that are available hands on training at our national labs that are available to researchers to build up their skill sets in this space. Because at the end of the day, we think AI is going to be a transformative tool that will be used to amplify the work that we are already doing. We also see it as transformative for DOE's mission in science, energy and security. And so we have a lot of plans and thoughts about how we want to drive that forward.
Dan: Helena, I have a question sort of following up on Alison's about workforce. It's great that DOE is meeting this sort of near term milestone but for the future, there's a lot of talk sort of in the utility sector about you know, waves of retirements and yes, you know, backfilling and finding replacements on the skilled side. How does the Department of Energy think about you know, longer term science, technology, engineering and mathematics, education and training, sort of for the future for the people who might still be in high school or a grade squat today? How do you make sure that those opportunities are open to sort of the widest range the greatest number of people from as many different parts of the country as possible?
Helena: This is a really important issue because it and it's not just something that doe is, I mean, you mentioned utilities are experiencing this, obviously at our national labs. We are also experiencing this issue of talent shortage, and I think the White House recognizes this this is something that was in the AI Executive Order, specifically around talent and talent in government but also talent more broadly. If this is going to continue to be an issue, and what we're going to need to do is to invest in partner creatively, to make sure that we can build a pipeline that has skills that can help support all aspects of this chain. Right? It's not just the data scientists, how are we going to train our grant writers and our, you know, our technicians and utilizing AI to optimize the work that they're doing. But But I agree with you, this is going to be a really big area where we're going to need a concerted focus, the labs have already begun this, we are already talking and thinking about this within the context of the executive order. But really, I agree with you, this is an issue and it's one that is going to persist, and we will need to continue to invest here. What I would also say is that it will continue to be important for DOE specifically, but also broadly the United States to continue to be a place where we can attract the best talent from all around the world. It is one where we need to do both, we need to train the American workforce and make sure that we are highly skilled and equipped to do things at the edge at the frontier. And we are also going to need to draw from everywhere else because this talent issue is going to be important for our leadership broadly in science and technology.
Dan: So you just mentioned the edge, right? The frontier, not only is that where your work is happening, and the Department of Energy's work is happening, but it also seems to be happening at a breakneck pace. Not only do you have the AI for Energy Report, but just last week, there's also been a new initiative launched called the frontiers in AI for science, security and technology. So no wonder time seems to be flying for you and your colleagues. But this was a big deal. Deputy Energy Secretary David Turk, proud of former EESI briefing panelist, David Turk, was part of the big announcement. Can you tell us a little bit more about that and what that means for your work going forward?
Helena: Yeah, and again, this is not just my office, this is something that I know that our labs, our program offices have invested enough. And I've been working in AI for decades, right? I think they've invested enough to see the potential transformative opportunities that lie before us in our mission spaces. Right? So yes, it's energy. But it's also science. And it's also our national security missions. And so, you know, the the intent behind the fast division and proposal is really to pull together something comprehensive that leverages the existing infrastructure, we call it the enabling infrastructure that lies at DOE. So if you think about what do we and you know, I think for listeners of this podcast, you know, you may you may know, DOE because it's got the word energy in its name. So you may associate DOE specifically with wind or solar or nuclear, or fossil energy. But really, you know, DOE has at its core, then s&t agency, right. And we have this huge science mission, we have this huge national security nuclear security mission. And I think it is not often known, you know, the fact that I was telling you before, you know, that there are these that we steward design develop with industry, you know, the fastest supercomputers in the world. And we need that for our mission space. So I mean, we already have this compute. We have tremendous stores of classified and unclassified scientific data that is developed through our network of scientific user facilities across the country, as well as through our research with academia and industry partnerships. There's this tremendous store there. And of course, our workforce, right. And our workforce is comprised not just of AI experts, we have some of those, but also, you know, material scientists and physicists and biologists and chemists, so, you know, this multidisciplinary workforce, across the national labs that work in the interests of the nation, right, so we do they do work, of course, we steward them here, a DOE, but they also support many, many other agencies in the work that they do. So really, the intent around fast is to think about how we can supercharge this effort to develop a capability within the US government that will help us answer a call around these pressing science, energy and security mission spaces. So we really see it as essentially four things that are underpinned by its core partnerships, both with industry and academia, as well as our workforce. And the four things really around how we harness the data to make it AI ready and available at scale, how we advance computing and infrastructure so that we can both develop the next generation of that frontier computing capability not just what's available now. But you know, what, what the new architectures are going to need to look like how we're going to develop trustworthy frontier models that are going to not be large language models necessarily, but you know, the kinds of science based models that we're going to need for our mission. And then lastly, how we tune those models to answer specific questions to address, you know, to develop new materials that will be replacements for others, or how we're going to develop new kinds of technologies that will enable the clean energy transformation. And we have already, I mean, in some cases, begun to do small investments that point the way to what this larger thing will look like. But I think it's a really motivating and an exciting time to be a doe, because we're already at work trying to do even more. And I think there's just really a huge opportunity, space, and a lot of promise.
Dan: That’s very, very exciting, and you have a ton of stuff going on. But Helena, thank you so much for joining Alison and me on the podcast this week, it was really a great chance to finally get a chance to meet you and listen to your remarks uninterrupted, where I didn't have to, you know, head backstage for something else. So thanks so much.
Helena: Thank you so much for having me.
Dan: Alison, like we said in our intro, we're both a little bummed that the season is coming to an end, we've covered a lot of really interesting ground this season, I've learned a ton gotten to know some really, really interesting people, some neighbors, some people that I've kind of always wanted a chance to get to know a little bit like Helena. But you know, maybe it's fitting that we ended the season on something looking toward the future artificial intelligence and its potential to help us meet all these different goals. I think one thing that really comes through listening to Helena is how important it is for something like the Department of Energy and organization like the Department of Energy to sort of really have a strategic focus, and to be able to pull all of the different people from across the agency together, in this case, to advance artificial intelligence for energy security, national security, all the different reasons she elaborated, but also be able to strategically pull in people from the outside. I think that is an underrated ability of the Department of Energy. But it's really, really important. And it probably applies to more than just DOE, federal agencies have a lot of convening power. And often, something that can really help things get going is just to help people talk to each other to convene people, get them together on the same conference room, and, you know, focus their attention on some of these big challenges ahead of us. And I feel like in many cases, that is a really positive step forward. But it was really, really great to get a chance to talk with Helena such interesting work wouldn't surprise me if we come back to this topic in future episodes, or briefings or articles or all the other stuff that we have going on. What did you think?
Alison: So one thing that I kept thinking about throughout the discussion is how many moving parts that there are not just with artificial intelligence, but also with the power grid. And then also within the structure of the Department of Energy, especially with a new office, there's so many new people new programs to coordinate not to mention coordination with the national labs. And it is amazing to me that Helena is juggling all of these things at once all the time. So I don't know how much sleep she's been getting in the last six months since the office started. I hope she's getting enough but it is really an impressive amount of work that they're churning out under under her leadership. If you want to learn more about EESI's work on new technologies like artificial intelligence, head to our website at eesi.org. Also, follow us on social media @eesionline for all of our recent updates. The Climate Conversation is published as a supplement to our bi-weekly newsletter, Climate Change Solutions. Go to eesi.org/signup to subscribe. Thanks for joining us and see you next season!
Advancing science-based climate solutions
Top-Rated Climate Nonprofit – 4-Star Charity