Shaping the AI algorithm + innovative finance for local infrastructure projects

It's Friday, February 20th. I'm Brian Walsh, and welcome to this Week in Impact, where I talk with Impact Alpha journalists and editors about some of the most interesting Impact investing stories we featured this week. I'm back with editor David Bank. We'll discuss Shaping the Algorithm, investing across the tech stack to orchestrate good AI. Plus, building regional engines for affordable housing in California, and finally, local guarantees for local investors in infrastructure projects in Africa and Asia. That's all coming up on this Week in Impact. Hi, David. Welcome back to the podcast.

0:01

Thanks, Brian. I had the week off last week, but I'm here and ready to go.

0:38

All right, let's get into it. It was a big week this week. Good AI, David. Now, you and the team have been building toward this one for a while, and Impact Alpha just launched a new beat about good AI called Shaping the Algorithm in partnership with the Siegel family Endowment, and it kicked off this week with a call to start us off. What do we mean by good AI and why is this issue so important right now?

0:42

Well, the why is it so important? I think, is everybody's kind of becoming aware of the kind of disruption that we're already in with AI. We're trying to carve out a lane that's a little different than the discourse, which is either kind of doomer on the one hand or maybe kind of blithely utopian on the other. And it's around building good AI, as you say. And the distinction is there's a lot of applications. People are saying, oh, you could use AI to, you know, solve all kinds of healthcare challenges or even climate challenges. All that's to the. To. To the good. But there's a layer of actually building AI itself and not just sort of leaving that to the mega corporations that are the leading providers of AI now, but actually kind of getting into the weeds in, in the infrastructure itself. And the thing that we keyed on in the call and we've been thinking a lot about is the sort of AI stack and you can think of it goes all the way from kind of the energy, maybe even the mining of the minerals to supply the energy and the data centers. That's at the base level, obviously, the chips Everybody knows about Nvidia GPUs and the big push on to sort of get as many of those as you can and then up through the large language models. And above that, there's an interesting layer that's called orchestra, and it's kind of the logic, in a sense, or the command post for bringing it all together. That has not gotten as much attention. And then on top of that are these applications, as I said, in healthcare, in engineering and environment. And so we're going to be looking up and down that whole stack and providing opportunities for folks to get involved, for investors to get involved, and policymakers and others to build an AI infrastructure that actually could deliver these outcomes that we're all looking for in terms of sustainability, inclusion, democracy, as well as, you know, business, business imperatives like, you know, accountability, traceability, auditability. So this is going to be a huge industry and, and a huge beat for us. Here's Chris Jurgens from Omidyar Network, who was one of the guests on the call.

1:04

We start with the need to invest in the core AI stack to make the core AI stack itself safe, secure, unbiased and aligned to the outcomes we want for humanity. And if we don't have that core AI stack safe and secure, then all the things we want to do for AI for good on top of it aren't going to work. We're going to lead to harm.

3:15

So good AI refers to not just what the AI produces, but how it produces it.

3:40

Yeah. Paul Fellinger from Project Liberty Institute, that's a very interesting nonprofit and investor all around this open source, transparent, traceable protocols in the tech stack. He makes the distinction of investing with AI, which everybody's going to be doing pretty soon here, and investing in AI itself. Here's Paul.

3:46

We need technological solutions to ensure agency by design, explainability by design. So there's a lot of need for innovation at the infrastructure level, the protocol level of the AI economy. And then there's a second bucket which is the application layers. So this goes a lot to what is an investable market? How can you make a lot of money, but also how can you innovate with better business models that are not only great for profits, but that are also have net positive effects for the societies we live in and the economies we live in. So this is the application layer. So there's the infrastructure layer, the application layer, and then of course there's the assurance layer, because you need the technologies to reign in, to govern, to sort of control AI systems that humans don't understand anymore.

4:11

All right, so what's the takeaway for an investor here? So if you're an institutional allocator listening right now, a pension fund and endowment, what's the practical entry point?

5:05

Well, we've been hearing that they're not all that satisfied with what's on offer right now. Like they're thinking in very. You fiduciary responsibility terms and with the kind of due caution that investors tend to have. And they want more governance and risk management and ways of auditing trails and whatnot. So there are a number of leading institutional investors that have been sort of outspoken about this. Norges, which is the Norwegian sovereign wealth fund, the Canadian investor Manulife Calstrs out here in California, the big Teachers pension fund, all writing AI into their investment policies and other documents to kind of send a signal that there's a demand from these big institutional investors for a more kind of enterprise ready or even you might say, society ready AI. That's not just the Wild west race for kind of what's the fastest and most powerful model you can build. And then on the GP side there's a number of funds, fund managers that are seeing that there's a big opportunity. Obviously that Wild west creates a kind of backlash or rebound effect for the kinds of companies that can supply the assurance and guardrails that are going to be much in demand. So we had Mohammed Nanabe of Mozilla Ventures, People might know the Firefox web browser and there's the Mozilla foundation, and they've got a fund started out as a $35 million venture investment fund to make investments in precisely these kind of startups. One of their companies, Fiddler AI, raised $30 million last month to kind of manage the agents that corporations are going to be increasingly deploying AI agents. We've been very keen on this whole agentic AI. Of course, the listeners on this call and the readers of Impact Alpha are all agents of impact. Now they're being joined by AI agents of impact as well. TruePik is another one. There's folks probably know if they've scrolled through their social media. A big challenge in distinguishing between AI generated images and AI generated content and authentic content. And so technology systems that can quickly distinguish that and sort of certify that will become very important for content creators, for copyright holders, obviously, as well as for users. And then similarly, there's something called Musubi which came out of the founders actually had worked on dating sites, and dating sites have the same problem that AI does, which is like, how do you know, like who the other person at the end, at the end of that app is? And are they real and are they who they say they were? And so they've developed a kind of content moderation system using AI to be able to detect fakes in postings, to be able to moderate chats, get rid of, you know, explicit content. If it's not appropriate for that channel. See, whether you're being, you know, lured into some kind of, you know, get rich quick crypto scheme or the like. So these are the kinds of things you can imagine there's going to be big demand for. And there's now a veritable investment ecosystem of startups and entrepreneurs and then fund managers looking for those solutions and then LPs investing into those funds. And so it's becoming another impact investing vertical.

5:13

Now, is there a tension between the impact investing community's kind of urgency around AI and the reality that a lot of this capital deployment is going to flow regardless, with or without impact investors at the table?

8:44

Well, that's exactly the point. I mean, the level of capital investment is huge, something like $660 billion last year just by the hyperscalers. So the flood has already been unleashed. The question is going to be, is there also an investment ecosystem around these more socially positive, responsible guardrails? But also just what are we optimizing for with all of this AI? There's been a very vibrant responsible investing community and now a growing responsible AI community. And I think the challenge is bringing those two things together, really. It's kind of a question of narrative. Here's Katie Knight from the Siegel family endowment. And then Chris follows on the question of narrative, which came up quite a bit yesterday.

8:56

Part of it is that we need a mindset shift so that we're thinking more about technology as tools and not as our masters. And I think there's a very strong imperative from big tech to make us believe that the tech itself is the end, that the tech that we serve the tech and not the other way around. But the reality is that we have both the agency to choose what we adopt and what we don't adopt, and ideally the power in sort of collective action to say, hey, this is what we believe is beneficial and this is what we believe is not worth it.

9:44

And right. There's these two narratives that, you know, AI is in. Tech is too complicated. If you're not an expert, you can't opine on what it does and doesn't do that is wrong. People are the protagonist. We should be determining through democratic processes in public policy and through within firms with workers engaged in management of what we want the tech to do and what we don't want the tech to do. So this notion of shared agency and how do we bring local communities into conversations about data centers, workers into how it is and isn't deployed in their companies, parents and kids groups on how AI isn't and is and isn't used in those contexts. I think we too easily default to the the narrative coming from industry that, you know, it's scaling the technology wherever and whenever it's profitable. But for technology as powerful as this, yes, we need to pursue profitable pathways, but we as society need to decide to instill the guardrails of what we don't want it to do.

10:19

All right, David, sounds like a lot to unpack from that call and a lot to discuss moving forward in this new beat. Looking forward to shaping the algorithm with you and the rest of the Impact Alpha team.

11:18

Moving forward, it's game on, Brian.

11:28

All right, you can read it right up on the call and see a full replay of the video on impactalpha.com there's also a link in our show notes. Later on we'll discuss how the Private Infrastructure Development Group provides local guarantees for local investors in infrastructure projects in Africa and Asia. But for now, we'll take a quick break. When we come back, building regional engines for affordable housing in California.

11:31

Sign up for Impact Alpha's free weekly newsletter Impact Alpha Open a crisp, expert curated investment briefing that lifts up people, data, tools, trends and job opportunities to help you stay ahead of the curve. Find the Alpha in Impact. Visit impactalpha.com open to learn more.

11:55

Next up, housing and not the usual story about lack of Supply or NIMBYs. This week impact Alpha published a guest essay arguing that the real problem is structural. Were financing affordable housing with tools from the 1980s. David the authors are actually the people building something new in California. The leaders of the regional housing financing agencies from the San Francisco Bay Area, Los Angeles and San Diego. So David, what's the case they're making for us?

12:22

Well, Brian, I've been covering housing in one way or another out here in California since about the 80s. So it actually is interesting to see and hear something. California, as you may know, is sort of the biggest in everything in housing. It's got the highest cost, it's got the largest number of people experiencing homelessness. It's got the most severe shortage of affordable homes. And so the innovation of these regional housing finance agencies, RHFAs, if people want another acronym, is very interesting. What LA in particular has done, I think it's the most advanced. They get a slice of some sales tax revenues every year and they can use part of that to basically anchor bonds and also anchor Impact funds. So it's basically some public money that then mobilizes private money. We love that. That's kind of catalytic Capital. And they're using that to then backstop kind of regional affordable housing initiatives and really just kind of ramp up the supply of affordable housing and it kind of gets the developers out of this cycle with the low income housing tax credit that's been kind of the main federal vehicle. It takes a long time. There's not enough of it. A lot of projects get sort of stalled in the pipeline. These regional housing finance agencies are able to kind of bring a whole nother set of public private money into the mix and possibly really change the game on the level of supply of affordable housing.

12:49

Right.

14:20

And I think this regional lens is really important because they are at the national level. They're not at the local level, but they're local enough to understand the market realities, but they're large enough to raise and deploy capital effectively. So one thing that Pete pointed out is that they're almost, these regional housing finance authorities are almost systems engineers for housing, not just lenders.

14:20

Yeah, that's right. The guest post we had this week from leaders of those three agencies talked about it being sort of the Goldilocks solution, just right. And as you say, folks do kind of live in regions. They might cross city boundaries or even county boundaries to get to work from where they live. Transportation and other things are regional challenges. And so you have to sort of not have these cities, you know, kind of competing with each other for federal money and things, but working together to solve it at a regional level. Chan Zuckerberg Initiative seeded this idea and seeded some of the planning money for these regional housing finance agencies. They've recently pivoted away from housing. They're just much more into sort of biomedical and science research now. But this is one of the legacies of that project. And in la, it's really starting to work. They get some of these sales tax revenues. They take a portion of that to anchor a social bond essentially, which they're going to market with. And they can do this essentially every year. So there's kind of going to be a dedicated public funding stream that anchors a series of social bonds that provides a kind of steady stream of capital for affordable housing. And it kind of turns affordable housing into kind of a public infrastructure, not just a kind of one off private development. We hope, we hope a lot of it gets built, but something that actually gets planned for and financed in the same way that transportation projects do and other kinds of things.

14:38

Yeah, I was struck by the international comparisons. In places like Paris or Vienna or Singapore, they affordably house 25 to 75% of residents using public financing mechanisms versus the 1 to 2% in California currently.

16:13

Yeah, they call it social housing. And it's just part of the way the society works because having a safe and secure place to live that's affordable makes the whole community work better. I mean, makes families more secure and schools and businesses. And it's just seen as one of the ways that makes society function. And in the US it's kind of been an afterthought and again, like an aspiration that we hope you can afford somewhere to live. Good luck.

16:27

All right. Well, you can read the guest essay on building regional engines for housing affordability on impactalpha.com there's also a link in our show notes. But for now we'll take a quick break. Coming up after the break, the same playbook, California is trying using guarantees and public structures to move private capital but apply to infrastructure across Africa and Asia. The results might surprise you. Finally this week, a story that connects to some themes we've been tracking for a while around local capital mobilization in emerging markets. Arloo Sangiki spoke with the CEO of the London based Private Infrastructure Development Group, which provides long term debt for private infrastructure investments across Africa and Asia. Now, David, the CEO of the Private Infrastructure Development Group has a striking claim. His loss rates are less than one tenth of what rating agencies predicted. If that's true, why isn't capital rushing into these markets?

16:55

Well, it's a classic misperceived risk and one person's risky bet is another person's opportunity. What they have done, as you say, they have a fund of their own to back infrastructure projects. These are things like solar mini grids in Nigeria and mobile telecom in Senegal and electric motorbike production in Vietnam. But then they quickly realized that it wasn't just direct investments that they could make themselves, but actually they could guarantee projects and thus bring in other investors. And that might be a higher leverage way to operate. So they created a guarantee, they called it guarantco to do that. And then they went sort of another interesting step, which is, okay, that's a kind of global guarantee facility. But wouldn't it be better to create local guarantee facilities in each country and bring in local institutional investors? As you said, that's been a major priority and a major theme throughout the year, particularly as kind of USAID has disappeared and foreign aid more broadly has kind of gone in retreat. So bringing in local pension funds, local insurance funds, local family offices and wealthy individuals. So starting eight or nine years ago, they set up one of these infra credit in Nigeria, Infra Zameen in Pakistan and Damana in Kenya. And they're going into several other countries. And these local guarantee funds become entities of their own. In fact, the Nigeria Guarantee Company went public on a stock market in Nigeria last year. And then PIDG was able to get some of its money out from the public market. So they've really gone sort of down a track to localize the guarantees as catalytic capital to draw in private investors for local infrastructure. Really quite an interesting journey.

17:49

So back to this idea of misperceived risk or mispriced risk. It's a phrase that we've heard a lot over the years in emerging markets. Is PIDG actually moving the needle on that perception or are they just finding the investors who were already willing to go into these markets?

19:51

Well, that's a good question. They did find that they initially had to guarantee 100% of the investments, which make it completely risk free. And they're finding, as the track record gets established, that that number can go down. So there is some learning that goes on. There is this misperception. I think what you mentioned about the loss rates is a guarantee is really a liability against your balance sheet at some level, but it doesn't actually cash out the door unless the loan goes bad. And you have to make good on your guarantees. So it can be a very capital efficient way to mobilize additional capital because you're not really putting anything out. You just have to have a bank account, essentially a balance sheet that you can obligate against possible losses. If those losses are less than anticipated or less than the ratings agencies or the investors think they're going to be, then you actually make some money providing guarantees, which you do get paid for.

20:05

All right, well, you can read Lucy's story on the private infrastructure development group on impactalpha.com there's also a link in our show notes. But for now, that's going to do it for this episode of this week in Impact. Thank you, David.

21:02

Thanks, Brian.

21:15

We really covered some ground this week.

21:16

Yeah, we really did. All the way from AI to affordable housing in California to emerging market infrastructure.

21:17

All in a week's work here at Impact Alpha.

21:22

Indeed.

21:24

All right, thank you all for listening. This week in Impact is a production of Impact Alpha and part of the Impact Alpha podcast network. Smart conversations by and for Impact investing professionals. For exclusive stories and insights about the world of impact investing and sustainable finance, join thousands of other agents of Impact by subscribing. Just go to impactalpha.com subscribe for full access. Thanks as always to our producer extraordinaire, Isaac Silk. For David Bank, I'm Brian Walsh, managing director of the impact advisory for firm Human Nature. We'll see you next week for more impact.

21:25