Phil Sansom – Producer

---

Transcript

Giles Whittell: It might seem bizarre, but until a century ago, the world was a bigger place. For most people the only possible method of travel was walking, and a simple trip from one city to another could take days.

Lucy Yu: The only things that could make travelling quicker and overcome those limits were horses. So for millennia, horses were central to the way we lived…. 

Adam Dorr: …and our cities and our lives were built around that mode of transportation. 

Lucy Yu: That’s Adam Dorr from the think tank RethinkX. He’s talking about horses as an example of something so entrenched in our society that it seemed we would depend on them forever. 

But when a better technology came along… the old world disappeared astonishingly quickly. 

Giles Whittell: In the late 19th Century the motor car was invented, and by the early 20th it began to rival horses. 

Adam Dorr: … horses were disrupted by cars, in just 15 years.

Giles Whittell: When something more advanced comes along, it can be completely and utterly disruptive to what came before. In this case, cars were so much faster and more convenient that they immediately outcompeted horses. So people made the switch.

Lucy Yu: A similar example is the advent of digital technology almost a hundred years later. In this case, what made information goods overwhelmingly competitive was their cost.

Adam Dorr: …before that digital revolution, you had to pay for information. If you wanted music or or movies or something like that, you had to buy a physical recording of it. If you wanted to read a new story, you had to buy a copy of the paper or the magazine. But once you’ve got a computer, it costs virtually nothing. 

Giles Whittell: At zero marginal cost to produce, information goods had an extraordinary edge. And their explosion ushered in a whole new set of industries.

Adam Dorr: …literally trillions of dollars of new economic opportunity. And it was built largely on business models that did not exist before – business models that were based on information being super abundant and effectively free, which was unimaginable before the age of the internet. 

Giles Whittell: So there is enormous potential for rapid change as a result of technology. And Adam thinks we’re about to see something directly comparable when it comes to energy – a revolution that brings about an era of super-abundant clean energy, unlike anything we’ve seen before.

***

Lucy Yu: Hello and welcome to Inside the Energy Transition – a podcast about the bright future of green energy. I’m Lucy Yu, CEO of Centre for Net Zero, an impact-driven research unit founded by Octopus Energy. 

Giles Whittell: And I’m Giles Whittell, the editorial lead on climate and sustainability at Tortoise. In this series we’re exploring the biggest questions, and debunking the most commonplace myths, about the energy transition: one of the defining goals of our time. 

Lucy Yu: Here at Centre for Net Zero, we work at the intersection of tech, climate and energy to advance a fast, fair, and affordable transition. 

Giles Whittell: If the right steps are taken, we can achieve this kind of transition and get to net zero. And over six episodes, we hope that this podcast series will enable you to share this view, and go away with a positive sense that the future of energy is bright. 

***

Giles Whittell: Lucy, each week you and I are going to take a common myth about the energy transition, we’ll analyse it with help from our guests, and if it turns out to be wrong, we’ll bury it. So what is the first myth we’ll be looking at? 

Lucy Yu: Well this week we’re examining the myth that the transition to clean energy will be slow and inefficient. 

Giles Whittell: Okay. So that’s the myth. We’ll go into the specifics later, but broadly, what is the truth then?

Lucy Yu: Well, as Adam says, we’re about to be hit by a really full system sustainability revolution. 

We’re going to see exciting technologies about to enter people’s homes, everything from electric vehicles, home batteries, home solar panels, automation in the home such as through smart thermostats, intelligence in other parts of the energy system.

This will lead to a massive paradigm shift. But in order for this to happen it’s going to take unprecedented cooperation between tech, energy, governments, and many other actors.

Giles Whittell: Is this just wishful thinking? Because obviously that would be a good thing if we had an energy revolution in a matter of a few years, because that’s all we’ve got. But do you seriously think that could happen?

Lucy Yu: I think if we look at other sectors, there’s very good reason to believe that it will do. Data and digital are not new things, so they’re things that have already massively disrupted sectors such as finance, such as healthcare, such as transport. 

And in fact I think what we’re about to see in energy is really just the impact of those technologies moving across to a sector which has traditionally been very slow moving, and slow to respond to technology and innovation. 

So we can expect to see practices and principles that have already been demonstrated and proven in other sectors, with an additional external driver of this incredible need to decarbonise as quickly as we can. So I do believe that this will happen extremely quickly, and I’m looking forward to hearing from our guests!

Giles Whittell: Well let’s speak to the first of those guests. Greg Jackson is here – he’s the CEO of Octopus Energy, a company that’s been described as an ‘energy tech unicorn’. 

It was only founded seven years ago but it’s already stepped up to challenge the legacy ‘Big Six’ energy companies, while providing 100 per cent renewable electricity to its customers. 

So we’re going to find out how they’ve done it and what this disruption means for the future energy market. Welcome Greg!

Greg Jackson: Hi, how are you doing?

Giles Whittell: Great to have you here. Look, let me ask you just to start by telling us, in a nutshell, the Octopus story and how you’ve grown so quickly. 

Greg Jackson: Yeah. So Octopus is a company whose job is to drive the renewable energy revolution across the globe, cheaper and faster than anyone imagined. 

And the way we’re doing that is first of all using technology to drive down the cost of an energy company and make it dramatically more agile. That enables us to simultaneously start bringing people cheaper energy while it’s greener energy, and give them an outstanding experience. So instead of the renewable revolution being done to people, it’s been done for them and with them. 

Where we are today is we operate in over a dozen countries. We’ve got over 3 million direct customers. Our technology is licensed to over 20 million customers. 

And we’re trying to join it all together, because fundamentally when the wind blows and the sun shines we now have abundant, super cheap energy. And what we have to do is create a system where we can use it when it’s cheap. The more of it we use when it’s cheap, the less of a problem we’ve got when the wind’s not blowing so much or the sun’s not shining.

Giles Whittell: You’re a tech company too. You’ve got this platform called Kraken – tell us about it. 

Greg Jackson: So look Kraken… when we started this business, we identified that 14 per cent of the revenue of a traditional energy company was spent on overheads and administration. The equivalent figure for Amazon is 2 per cent. And a typical energy company’s got two products. Amazon has 300 million. Now if you can do 300 million products for 2 per cent, it kind of says something about your technology if you’re spending 14 per cent for two products. 

And so today Kraken powers the Octopus Energy business in the UK and five or six other countries, and it enables us to look, for example, looking at a wind turbine, how much energy is it going to be generating; and then be deciding up to two weeks ahead what you’ll be doing with that energy. 

So we’ve got a whole load of people with electric cars. If we know it’s not going to be windy for a few days, let’s reduce the amount we’re charging the cars – always leave enough in the battery for the uses they need, they can always override the machine intelligence that does this – and then we’ll fill the cars up when it’s windy. 

That kind of optimisation is the way that we can dramatically reduce the cost of energy as we make both generation and consumption more renewable. 

Lucy Yu: And is technology the silver bullet here? You’ve talked publicly about your plans to bring a digital revolution to the energy industry. Is technology all that’s needed to achieve that?

Greg Jackson: Technology is critical, in the same way as in a minicab office, you’ve got someone picking up the phone and clicking on a mouse, trying to move cabs from one place to another; and you compare that with Uber, which in real time has massive amounts of data on the location of every car and every potential passenger, and it’s got really good predictive stuff, so you know if it rains what’s going to happen. 

It’s the same in energy – we need to do this real time matching and forecasting of generation and consumption. But it’s nowhere near enough. 

There’s a couple things we should be really aware of. The first is, we’ve really got to understand that countries around the world have made some big mistakes in this transition when it comes to how we work with the citizens of the world. 

For a long time, renewable energy was: a bunch of anonymous construction companies would come and whack massive great pylons up on a nearby beauty spot, and in return local residents would get an extra line on their energy bill to be told it was to tackle a problem that was decades away on the other side of the world. 

Instead, the kind of thing Octopus has done is: we now for example have something called the Fan Club, where if we have a wind turbine in an area that’s part of the Fan Club, people who live nearby can join the Fan Club and they will get cheap energy when the wind’s blowing. So in return for hosting a wind farm, they’re actually getting this benefit of cheap green energy. 

Now that kind of thing that says, “how do we make the renewable revolution good for citizens,” I think is just as important as the technology. 

Lucy Yu: So it’s technology and co-design with the people involved. 

Greg Jackson: Yeah. And it’s the same thing with everything. Uber wouldn’t work if it just had the tech. It’s also created a user experience – in the vernacular – that people enjoy, the bit where people point at the map and say, “I see the car coming”. That kind of thing makes Uber’s tech meaningful. 

And we’ve got to be doing the same in energy.

Giles Whittell: Can the grid handle the energy world that you describe, especially five, ten years out where you might have a widely distributed generating system with turbines in every backyard, as it were; with a huge fleet of EVs that can push energy in as well as draw it out; is the grid that we’ve got anywhere near ready for that?

Greg Jackson: You know, the physical grid we’ve got is already massively underutilised. I think on average it runs at about 60 per cent of capacity. Now if that was airlines or hotels, entrepreneurs and travel agents would be springing up to sell the spare capacity and give us all cheaper holidays. 

In energy, because we’ve got these monolithic ways of running the system that are bureaucratic and regulated and run on a ten year consultation process, we kind of just keep putting more copper in. Even though the existing copper is so underutilised the vast majority of the time. 

So the first thing we need to do is digitise the grid: use the kind of technology I’ve been talking about to be able to, at times when we’ve got abundant green energy, give it really cheap access to that spare capacity on the grid and use it to charge electric vehicles, heat homes, run factories, do baking… I mean, it doesn’t matter, but let’s use it, and let’s make it very cheap, just as we do with spare hotel rooms and spare airline seats. 

Similarly, the same technology would then enable us to work out the bits of the grid that do need reinforcing. So instead of sticking enormous amounts of new copper in the ground or in pylons, we’re able to target it where it’s needed. 

So you can imagine in a local network in a cul-de-sac, just to get a picture in your head; as people start getting electric cars, so that local grid will end up getting overloaded if they’re all plugging in that same time. So the first thing we should do is, let’s use smart pricings and intelligent optimisation so that your car charges at a different time than next door, which charges at a different time than their next door. 

That way we can create a much smoother load, making the most of the capacity before we ever have to put any more in. 

Giles Whittell: So you’re saying that this digitisation can end the kind of peaks of demand that presumably were the reason we built a grid that you say is broadly underutilised. Do you think, or do you know, that we can even out that demand curve?

Greg Jackson: I’ll tell you what, we’ve spent four or five years repeatedly proving this. And it’s fascinating when you’re talking to legacy industries. Time and again, you show it, and every time you show it, they go, “ah, but what if!” 

And anyway, finally, the results of a major trial we did, tens of thousands of households demonstrating in the real world reducing peaks by 23 per cent. And by the way, by more than that if people had electric vehicles. Demonstrating that if we put an alert through to people on an app or a notification by email, they will turn up their electricity or turn it down if we reward them for it – in droves. 

Now this isn’t saying, by the way, that everyone has to do it. It’s a bit like Tesco. Most of us go into Tesco and we buy the mince we were going to buy. But some people will spot there’s another pack of mince with a yellow label on half price, about to hit its sell by date, and they choose they’re gonna have a different meal that night. 

Now because of those people, we get less food waste and all of us pay less for our food in general, because Tesco are able to monetise something that would otherwise be thrown away. 

That’s the way we need to see energy – which is the people who enjoy a bargain should be able to grab one! If the wind’s blowing and there’s electrons we’re otherwise going to throw away, let’s make the most of them! 

And that’s happening, by the way. Last year, I think in the first quarter of 2021, the national grid threw away 300 million pounds of green electrons when the wind was blowing, because we had nowhere to put them.

Giles Whittell: So what’s the technological solution to that? I mean, you see it coming, and you let people know that the oversupply is coming so energy will be cheap, and please use it now?

Greg Jackson: Exactly. Now will that reduce the whole of the 300 million to zero? Well, if it makes it 150 million, that’s better. If it reduces it to zero, even better. 

It may even be that you can charge a little bit for it, thus not only reducing the waste, but reversing it. 

And we have great examples. Octopus Energy has got thousands of customers on a tariff that varies every half hour. And at times prices go negative. We pass those onto consumers. And you find that people will say, literally, “come and charge your electric car in my driveway. That way you get free electricity in your car and I get paid for it.” Right? 

Now these things are happening already. What we need to do is open up the mechanism so it happens at scale. 

And electric cars today are still only a single digit percentage of the car fleet. But last month, 25 per cent of all cars sold in UK were electric! The rate of growth is phenomenal. And what we’ve got to recognise is an electric car holds five days electricity for a typical house. 

So this isn’t about going around and turning off a three watt light bulb. This is about automating the times at which we shift 7,000 Watts of electric car charging. 

***

Lucy Yu: But the green transition needs to take place at huge scale all around the world. So we still need to build thousands of solar panels, wind turbines, and batteries. 

Cost is a commonly cited issue, but our next guest is here to explain how quickly the cost of emerging technologies can fall and how this can catalyse extremely rapid change. You’ve already heard a clip from him: it’s Adam Dorr from the think tank RethinkX!

Adam Dorr: Thanks for having me.

Lucy Yu: Adam, you’re currently researching the disruption of the global energy sector by new energy generation and storage technologies – tell us more about it. 

Adam Dorr: Sure. Well we are on the cusp of the most profound disruption of the energy sector in over a century. 

And like other disruptions, this one is being driven by the convergence of several key technologies, not just by one technology, but by several. And in this case, the main ones of interest are solar power, wind power, and batteries for energy storage. 

And these were expensive – not long ago these were expensive technologies – but the cost has improved relentlessly and consistently. So that now, the combination of solar, wind, and batteries is a cheaper way to generate electricity than conventional fossil fuels, than early generation nuclear power, and even hydropower in many areas. 

By 2030, the economic competitiveness of these new technologies will simply be overwhelming almost everywhere. And so all of the evidence shows that this is going to be a textbook case of technology disruption. 

So based on their current trajectory and our understanding of the historical pattern of disruption, we fully expect solar, wind, and batteries to outcompete and replace the older, incumbent energy technologies over the course of just the next fifteen to twenty years. 

And this is an important consideration because these disruptions, when we look at history, they happen much, much faster than seems intuitively plausible, intuitively likely.

Giles Whittell: And you’ve written about parallels between this disruption in energy and comparable ones in the digital information space. Where are we on the parallel curve, as it were? I mean, if you can put it in terms of, say, uptake of smartphones in the last decade of the last century and the first decade of this… can you put it in those terms?

Adam Dorr: Sure. The key parallels are: number one, that the fundamental underlying pattern of disruption is that a new technology emerges that is simply overwhelmingly competitive compared to what has come before. 

It offers a combination of capability and affordability – so capabilities and cost – that the incumbent technology simply can’t compete with. And so they are outperformed and outcompeted very quickly. These are largely driven by economic forces, and they were driven by economic forces even before we had a modern global market economy. So that underlying pattern holds. 

But one of the most striking features of this particular disruption now in energy, and the parallels it has for the digital disruptions of the past, is that these new energy technologies have extremely low operating costs, especially solar and batteries. 

So once a solar plant is built and installed, it just sits there, right? It just sits there and happily makes electricity, no need for fuel, there’s virtually no maintenance; even the labour requirements are low, it doesn’t need security guards like a nuclear power plant does; if there’s rain once in a while to wash off the dust, it doesn’t even need cleaning. 

What this means is that once you have the assets, the marginal cost of generating electricity is very close to zero, with solar and batteries in particular. And wind also is very low marginal cost. 

So what happened in the world of bits is about to happen in the world of electrons. And we’re just starting to enter this new era where we will have super abundant, clean energy. And that is going to change everything, very much the way that the internet changed everything for us.

Lucy Yu: There’s a fascinating central idea in RehinkX’s work on energy and energy futures, which is what you’ve just described, that we will encounter a situation in the future in which actually renewable energy will become so economically favourable that we will actually have an abundance of it – we will have this idea of what you call ‘super power’. 

I’m interested to know how often you think that we will effectively have too much power in the future? And do we have the right regulatory framework and market structures in place to actually really leverage this super power?

Adam Dorr: Well it’s a great question. And this was one of the most striking findings in our recent research: that any energy system that’s based on solar and wind power, it has to be built to get through the worst time of year, right? So that means cloudy weeks of winter when the days are shortest. 

What follows is that for the rest of the year, that solar and wind capacity generates a surplus of energy. And in this case, it’s a surplus of nearly zero marginal cost electricity. That surplus is very nearly free, and of course it’s clean. And that’s why we call that surplus ‘super power’. 

So this is extraordinary. And to answer your question, in sunnier regions – say for California, for example – super power will be available on over 90 per cent of all of the days of the year, because you build it to get through the worst time and all of the rest of the year you have this surplus capacity. 

Now, what we found, which is extraordinary, is that even in cloudier regions at higher latitudes – we studied New England in the United States, which has weather that’s comparable to much of Western Europe and the United Kingdom – and we found that super power is still available in substantial quantity for about two thirds of the year. 

So if the system is optimised – if the balance, the mix, of solar and wind generating capacity combined with the right mix of energy storage capacity and batteries – if that is optimised, the amount of superpower you can generate in total in a region is enormous. It’s up to twice as much as all of the existing electricity demand.

So this is like the early days of the internet, the regions and nations that embrace this sooner rather than later, will reap huge benefits.

Giles Whittell: To get to that point though, there is a period of infrastructure build out, if you will. There have got to be a lot more wind turbines and solar panels than there are now. So to what extent are we currently being held back, whether by government planning policies – local, state, federal – or barriers to entry in the market? 

Adam Dorr: Well, there are obstacles. And we do have a large build-out ahead of us. But an important consideration is that solar, wind, and batteries are now mature enough that they no longer need direct government support to be competitive. That’s what’s driving the disruption: their competitiveness. And that competitiveness will only rise over the next decade. 

But what that means is that the best thing governments can do at this point is to ensure that we don’t prop up the incumbent industries that are based on older technologies. 

And there is going to be a temptation to do so. There will be pressures from that incumbency. And it’s no surprise that they will do everything they plausibly can to secure, to retain, government support as their ship is sinking. 

That will only delay the inevitable. Nothing can stop this disruption, just as nothing has stopped disruptions in the past. But the social and environmental cost of delay would be enormous. 

So we, the public, we must be vigilant and not allow our governments to be captured by those obsolete industries. 

Now there’s no one-size-fits-all solution. There’s no one set of government policies that either isn’t working in one place or will work in every place. So we have a lot of learning work to do, experimenting and learning to do. 

But I think the bottom line is that these technologies are ready to deploy, and the focus of government needs to be on paving the way, clearing the obstacles, and making sure we are not shackled to the technologies and the industries of the past.

Giles Whittell: Is it the case that prices come down so far that the return for investors in this infrastructure comes down, and that you therefore have a risk of the investment drying up? Is that a problem? Or am I just looking for problems?

Adam Dorr: No, this is a legitimate concern, and those dynamics are very complicated, and they vary quite substantially from one market to another, one region to another, one country to another. 

For example, if you are a business. It will continue to make sense to put solar panels on the roof of your property and batteries on your property, even if the return on investment to those assets at the grid scale has declined. 

So we have a situation where there’s a lot to optimise. There are a lot of opportunities to obtain a return on investment that depend on factors like: where will these assets be utilised? When will they be utilised? And by whom? 

Let’s imagine a scenario, say a pick a region. Say one city in Texas decided that it was going to make a large investment in solar and wind and batteries. And as a result of that, it fully anticipated that there would be a very large quantity of what we’re calling ‘super power’ produced in that region.

The situation that would be present there after the build-out and the investment in those assets is in that region, in that city, electricity would be very, nearly free for a lot of the time. 

So that city could then attract businesses – new, innovative startups, perhaps older industries that are looking to pivot, and anybody who thought they could find a new way to utilise ultra low cost, nearly free clean energy, especially if they could utilise it flexibly, like at different times of day and in different quantities throughout the year. 

And if that city made that investment – had the vision and the initiative to do so – it would attract those new industries. 

And so investment in super power could be something very similar, it could run directly parallel, to something like regions offering tax incentives to attract businesses, or infrastructure incentives to attract businesses. Free electric vehicle charging is an example of an infrastructure investment that could attract businesses and the public, and new business models could be built around that, right? 

So this is the sort of thing that we’re talking about, where you aren’t building solar, wind, and batteries to make money selling electricity; you’re building them to facilitate the creation of value in other ways, built on top of those and the extraordinary condition that they enable.

It’s the same way that Starbucks offers WiFi for free, to get people in the door to buy their coffee, right? Well it wasn’t that long ago that we paid for internet dial-up access by the minute, right? That was not that long ago. In the 1990s, you paid to get on the internet by the minute. But within a decade, coffee shops were giving away access to the internet for free because they were obtaining value in other ways.

Those are the sorts of new, innovative forms of value creation that we’re going to see.

Lucy Yu: If you had a blank piece of paper in front of you right now, how would you design the future energy system?

Adam Dorr: Yeah, this is a fun thought experiment, but I think the key takeaway message here is that there really won’t be a single, perfect, one-size-fits-all design that will work everywhere. 

And so what I would do is incentivise experimentation: large pilot projects across the globe. See what works, and what doesn’t work, across a wide range of different geographic and social and economic conditions. 

I think the important thing is that we cultivate a mindset of innovation, a mindset of agility, and an openness and transparency as well, because those are the things that we need to learn as quickly as possible. 

And that’s what we need to focus on. We need to focus on learning, and then spreading that knowledge far and wide and as quickly as possible across the entire planet.

***

Giles Whittell: At the very cutting edge of all this technology is AI, which is what powers Octopus Energy’s Kraken platform that Greg was telling us about. 

Now clearly, data analysis and AI are going to be fundamental to a future that’s powered by renewable energies. 

That’s why our third guest this week is Priya Donti, a computer scientist from Carnegie Mellon University who is also the Co-founder and Chair of the group Climate Change AI. Priya, welcome!

Priya Donti: Hi, nice to be here.

Lucy Yu: Priya, what are some of the ways AI can help when it comes to the energy transition?

Priya Donti: Yes, so there are three big categories of ways I like to talk about. One of these is turning raw information into insight. 

For instance, as we manage our electric power systems, if we can get a better sense of how much electricity supply, for example, solar power will have in the future, this can help us to better manage these systems. And initiatives like Open Climate Fix are using a combination of satellite imagery and weather data in order to create these kinds of forecasts. 

Another category is making real world systems more efficient. 

And then a third big category AI can be used for is to accelerate scientific and engineering workflows. 

For example, as we try to invent the next generation of batteries for use in many different use cases like electric vehicles or for storage on electric power systems, usually what you do is you look at the outcomes of your past experiments, you try to figure out, “okay, what battery do I create next,” and then you synthesise that battery, it takes several months to test it out, and then you try the next one over again. And this whole process can be really time intensive. 

So AI has been used to, for example, analyse the outcomes of past experiments and more intelligently recommend what you can try next, in order to cut down the number of design cycles. 

So to summarise the ideas here: it’s turning raw data into actionable insights, helping improve the efficiency of real-world operational systems, and accelerating science and engineering for clean tech. 

Giles Whittell: Broadly speaking, how long does it take to go from the drawing board to AI actually being used to speed up our responses to these problems?

Priya Donti: Yeah, how long it takes AI to be used – it tends to depend a bit on the AI, but also on the specific use case for which it’s being used. 

For example, in the electric power sector, AI is already very widely used for forecasting. And this is because of a combination of things. One, the data for forecasting… while there’s a lot of work to be done to improve that data, to some extent that data already exists, so you can already start to apply AI algorithms to actually analyse that data. 

And there’s already a bit of an established procedure for when you create a forecasting algorithm, how it actually gets integrated into, for example, the electricity control room. 

Whereas if you’re trying to use AI to more dynamically optimise electric power grids… so there are lots of procedures on electric power grids that try to make sure that roughly the amount of electric power going into the power grid is exactly equal to the amount of power being consumed, and these processes are traditionally slow and inefficient, and AI has been used to try to speed those up and make them more dynamic. 

There’s both an aspect to which the data for that is only partially there, which means that you can’t necessarily today create AI algorithms that can do all of this for you; and it’s not necessarily easy to test these algorithms out, because there isn’t an established procedure for actually deploying them on either the real system or a test system to see how well they work. 

So these questions around data, and is there a place to actually test these things out, can really largely determine this question of how long it actually takes to get from idea to action. 

Lucy Yu: AI is something that I’ve worked on in different shapes or forms for a number of years now, and sometimes it feels as though we’re always talking about its future potential. 

And I’m really interested, in terms of the energy transition, are we harnessing its full impact as much as we could be right now? 

Is AI just another tool in the toolbox, or is it an active force accelerating us towards net zero?

Priya Donti: I think that AI can play different roles in different places. 

In some cases, it can be a way to accelerate or provide additional insight to an existing workflow. And in some cases it can really unlock new potential that wasn’t there before. 

You know, analysing policy documents in order to help decision makers make better policies… policymakers have always been making policy, but AI can be a tool that allows them to maybe make higher quality decisions based on more information. 

Whereas in another example – of more dynamically optimising electric power grids based on the data that you’re getting from sensors, in order to foster the integration of renewable energy – this is something that existing methods are really failing to do, so AI can be a really critical component of that aspect. 

And I would say that we are, in some cases, I think, starting to really realise the potential of AI from an algorithmic perspective; but there are some practical considerations, like is there enough data available, or is there data sharing between different entities? So is your algorithm really able to leverage a larger set of data to make better quality insights? 

There’s still a lot of potential to fix those kinds of infrastructure related aspects in order to increase the potential of AI. But then in some of these other workflows, like real-time optimisation and power grids, I think we are very, very pre- seeing that potential impact just because we really don’t have ways to validate these methods on the ground. 

Giles Whittell: Right. Okay. So we’ve talked about AI as a route to more and quicker solutions to climate change. But can it also make it worse? 

Priya Donti: Yes. In the same way AI can be used for applications that are explicitly good for climate change, it can also be used in ways that are explicitly counteracting climate change. 

So for example, if you make an emissions intensive industry more efficient, for example, making it cheaper to extract oil and gas, then that potentially impedes the transition to other forms of energy by making oil and gas cheaper. 

But there are also broader systemic effects that I think we also often don’t think about. So AI is for example, being used broadly in recommender systems in order to advertise to people and get them to consume more. And we don’t think of this often as a directly climate relevant application of AI, but it has huge climate impacts, potentially, if we’re increasing consumption across society. 

There are other places where AI is sort of a key part of a broader technology, and the sort of trajectory of that technology, how it’s actually developed – depending on how that happens – can either have good or bad impacts for the climate.

So if we think about, for example, autonomous vehicles, where AI is a critical component in the development of autonomous vehicles, if you develop autonomous vehicles in a way that entrenches the role of private and fossil-fueled transportation, that’s potentially counterproductive to decarbonising the transportation sector. 

Whereas if you develop autonomous vehicles in a way that makes it easier to use low carbon and public transportation, and to link up between different forms of public transportation, then you’re potentially helping decarbonise the transportation sector. 

And then of course we’ve talked about applications here, but AI algorithms themselves can have a carbon footprint based on the computational infrastructure that they run on, the electricity that they use to actually run. 

So really thinking about: do we always need the biggest state of the art AI model that is using a lot of energy, or can we get away with a smaller model? Doing that holistic assessment is really, really important. 

From a macro perspective, data centres currently make up about 2 per cent of global greenhouse gas emissions, whereas the electric power sector currently makes up about a quarter. Though of course these numbers could change. 

We want all sectors to go net zero, so it’s important to think about these numbers holistically when we’re doing that.

Giles Whittell: So what’s the solution here? Can you orient AI to minimise the negative impact, or is it ultimately a sort of old-fashioned human regulatory problem? 

Priya Donti: A lot of it comes down to human regulatory problems: where are the incentives in terms of where we apply our technologies; where it’s easier to make a profit under current profit-driven models; where is talent incentivised to go, in the sense of where are there fewer risks or maybe more established pathways to being a data scientist already. I think there are lots of regulatory, financial kinds of questions to answer there. 

Giles Whittell: Okay, well I’m reassured in a way – at least there’ll be something left for us to do. What does popular discourse often get wrong about AI? 

Priya Donti: Popular discourse often assumes one of two things. One of them is that AI is going to be the saviour that all of us need. It’s going to solve every single problem across society. 

And on the other side, you have the skepticism around: AI is not going to do anything. It’s a completely over-hyped technology. 

And the truth is really somewhere in between. There are things that AI is well scoped to do. There are places where its strengths can really shine. And there are other places where it’s wholly inapplicable for practical or technical reasons. 

So it’s really important to think about AI part of that broader toolkit, rather than being in some sense, anything exceptional that deserves a lot of terror or a lot of skepticism in that way.

Lucy Yu: And Priya, what’s one thing that excites you most about applying AI to the energy transition?

Priya Donti: As I mentioned earlier, I think it’s really important that everybody across society contribute all of the skills and toolkits and knowledge that they have in order to address the energy transition and the broader climate crisis. 

This is really an all-hands-on-deck kind of situation, where we really need to make really rapid advancements across technology and policy and social systems in order to address this issue. 

So I think the thing that excites me about using AI for the energy transition isn’t necessarily so specific to AI, as it is saying that: it’s one additional way for more people to get involved in enabling the energy transition, by using that specific toolkit – in the same way that I hope many other committees that are developing these kinds of toolkits also look inwards and say, “how can we use our specific skills and talents in order to address the energy transition?” 

Because the more that every single community does that, the more people we have on board trying out different things and getting us closer to actually addressing this.

***

Lucy Yu: So Giles, what have you learned? 

Giles Whittell: I’ve learned a lot. It’s a bit of a mental snowstorm. Three things come to mind. 

First of all, a very fast energy transition is possible, but it has to be designed with people in mind. It has to bring people with it. 

Secondly, how is this possible? Well, we can realistically look forward to a rapid increase in scale and continued rapid decrease in costs thanks largely to digital technology. 

Third point is really a caveat insofar as this involves AI – and I’m not talking about Hollywood disaster movie here – but we do have, somehow, to regulate its use so that it’s used right, so that it’s used for the good of humanity and doesn’t take over.

Lucy Yu: Anything else?

Giles Whittell: I think that’s quite a lot to be going on with! “Anything else? B minus!” 

No, I think there is something. I think the basic point that Adam made, which is that there’s no reason not to be very optimistic about a very fast energy transition, is really inspirational. Because it’s so easy to take on board all this information about accelerating climate change, and a very, very slow response by humanity so far, and think that we are – to use a technical term – screwed. 

But if he’s right, that you can have rapid transformation in energy, then we really could be looking forward to transformation at the speed and scale that we need.

***

Lucy Yu: You’ve been listening to Inside the Energy Transition, a podcast from Tortoise Media, Centre for Net Zero, and Octopus Energy. A big thanks to our guests this week: Greg Jackson, Adam Dorr, and Priya Donti. Find out more about us and what we do on centrefornetzero.org. 

Giles Whittell: And tune in next time to find out whether it’s really possible to run a whole country’s energy grid on renewables. Spoiler alert: it will take enormous advancements in flexibility, interconnectivity, and storage. 

Lucy Yu: This episode was produced by the lovely Phil Sansom with support from Izzy Woolgar, Gurjinder Dhaliwal, and Clizia Sala. The executive producer is Ceri Thomas.
Giles Whittell: Thanks very much for listening, and if you liked the episode, please do leave a review or recommend us to a friend. We hope you’ll listen in as our series continues.