Hyperlocal AI and Capital Absorption
Personal is the ultimate hyperlocal, and how I misinterpreted "investing"
Welcome to the latest issue of Decent Tuesdays, an experimental edition of The Main Street Journal that covers and connects the many aspects of decentralization.
Today’s Technology article, from Inc., notes that the current pursuit of all-consuming general AI overlooks the practical value of hyper-localized AI, which operates at the level where people live. For example, to optimize a store or restaurant’s operations, informed by city traffic patterns and the rhythm of its street on a Tuesday morning— and I would add: current news, weather, and sports, the stock market, space weather, and countless pattern-detected phenomena that humans would never recognize and probably lack the vocabulary to describe.
The article argues that this is the real future of AI: “distributed local intelligences, each tuned to its micro-environment but networked into a shared learning system” to focus on relevance, not achieving consciousness.
That sounds good, but I’d also want to take the localization even further, to the individual level, and flip it around. The AI apps that the article alludes to all work for businesses, so they direct AI to pursue goals like “answer questions from everywhere to maximize income by becoming as popular and indispensable as possible” or “monitor the security cameras in this jewelry store.” But what about individuals? That’s the ultimate hyperlocalization that we need even more in the age of AI.
As we interact through or with any globe-spanning AGI overlord, or even local business AIs, we’ll need a secure, private personal assistant AI app that looks out for our own interests— haggle for us, handle our communications, analyze legal contracts, keep us healthy; etc. These goals can be inherently adversarial to commercial AI’s goals, so our personal AIs need to run completely separately from them, and never share the data that they collect. Otherwise, goals will conflict within the same system, and our supposed assistant AI might collude against us. There has been lots of work on such personal AI agents, and I don’t know of any clear favorites yet, but the main thing is that they work for you and you alone, and don’t let any other AI see your poker hand.
Today’s Finance link, an article from Proximate, describes the Capital Absorption Framework, a successful approach to bringing funding into under-invested communities and ensuring that it’s used well. To avoid “helicopter money” dumping down and running off without being deployed effectively, framework users work with the communities to figure out their top priorities, identify projects that address them, and ensure that leaders and infrastructure exist to effectively use the money.
I’m impressed by all the great work that the framework has been applied to— developing affordable housing in Atlanta and the Coachella Valley; growing food systems, entrepreneurial support, and solar installations in Appalachia; and more. But when I read this article, I admit that I tripped on the phrase “capital absorption,” thinking that such phrasing would not appeal to investors, and I was also initially confused by how it used terms like “investor,” “deal pipeline,” etc. This led me to learn that in the community investing world, “investing” can refer to straight-up grants from philanthropic foundations as well as funding from banks, CDFIs, or other investors that expect repayment.
The language of investing is cool these days, so I understand why it’s used so broadly, and I likewise appreciate the teachable-moment possibilities when someone says, “Wait— that’s not investing!” But folks like me don’t generally use the term “invest” if the first number in a triple bottom line would be zero, and I want everything to be clear. The dynamics of { A) giving money to do something good, B) lending or investing to do something good while accepting the possibility of a sub-optimal financial return, and C) investing amorally to maximize financial return } are all different and create distinct roles and relationships, so I don’t want blurring A and B to confuse or put off people seeking to do either. I also feel bad saying this, because here I am just criticizing while these good folks have helped lots of communities, regardless of the superficial issue of what words they use to describe it.
News links below also list three other stories of interest:
To deal with new challenges and impending owner retirements, food and beverage companies are good candidates for becoming employee-owned, like Bob’s Red Mill and King Arthur Baking have done. Studies show that employee-owned companies outpace their peers, and unlike selling to outside investors, converting to employee ownership safeguards the company’s legacy, employee livelihoods, and connection to its community.
As federal programs end, states are forming coalitions to pursue shared priorities. For example, Connecticut and Maine, which both aim for 100% clean power by 2040, are combining their procurement processes by soliciting bids together and evaluating them via the same criteria, while choosing them separately. This comes after Maine approved one hydropower and four solar projects last month, and Connecticut solicited solar and onshore wind project proposals in September.
A “Kreisler” space in a busy part of Berlin’s Gropiusstadt neighborhood serves as a repair shop, tool lending library, learning center, and resilience hub, funded by sliding-scale membership fees. The article envisions the expansion of such Kreisler spaces into more neighborhoods, to form a network of affiliated Kreislers that adapt to local needs while sharing the same ethos and working across neighborhoods— for example, by Kreislers in richer districts helping to finance ones in less affluent areas.
A new book, The Pointillistic City, uses a painting metaphor to describe how, although we often think and plan at the level of neighborhoods, significant traits and inequities differ within neighborhoods at the scale of blocks and buildings. For example, in Boston, air pollution and stormwater flooding hazards vary from 70% to 85% within some of the same neighborhoods. But the explosion of urban informatics enables us to see, study, and plan for these microspatial variations.
Image Credit: Rijksmuseum, The Card Game on the Cradle: Allegory (detail)
NEWS
TECHNOLOGY
The Future of AI Is Hyperlocal Intelligence, Inc. (October 27)
FINANCE
Reimagining the Playbook for Communities Seeking Private Investment, Proximate (October 22)
FOOD
Want a Stronger Food System? Start With Who Owns It, Food Tank (October 20)
ENERGY
Connecticut and Maine Team Up to Fast-Track Renewables, Canary Media (October 29)
MAKING
Kreisler: Community-Driven Catalysts for Degrowth, Degrowth.info (October 24)
GOVERNANCE/POLICY
The Pointillistic City: Well-Being and Equity in Communities, Shareable (October 24)
To unsubscribe from Decent Tuesdays emails, click your profile at the top right, select Manage Subscription, and deselect Decent Tuesdays.