Unpleasant Robotaxis, AI Boyfriends, Warped Legal Incentives, and Martian Dogs
Off-Ramps for 6 March 2025
Welcome to Off-Ramps! Today I’ll highlight five interesting pieces that I think you will enjoy reading. Please enjoy these on your morning commute, or save them for your weekend.
You may know that I am a Fellow of the Roots of Progress Institute. Today, I’d like to highlight recent work by other Fellows. These pieces cover a wide array of topics, but are united by curiosity about the future, and how to build one with abundance and prosperity for all.
Unpleasant Robotaxis
of is interested in cities, and what robotaxis might be able to do for them. Jeff is also a former employee of Lyft, both as a driver and a customer-service rep. This means he is intimately familiar with how the public abuses ridehail vehicles today… vehicles that have a human driver, who often also owns the vehicle, in them at all times. Given how badly people behave in these circumstances, how will they conduct themselves in driverless robotaxis? 
Jeff thinks the answer is ‘badly’. He includes a photo of ketchup smeared all over a Waymo’s interface screen, and notes dryly that this is “not the worst thing I’ve ever seen slathered all over the inside of a car”.
There are steps that can be taken to mitigate the effects of all this, of course. Inward-facing cameras are one; welcome to the panopticon. The fact that they have a name and a credit card on file is another. And form factor will help quite a bit too; Jeff quotes one expert (who is, ahem, me) on how design will trend toward the utilitarian, for ease of cleaning: chrome and vinyl everywhere.
One thing Jeff doesn’t imagine, but I think is a likely possibility, is tiered membership: if you pay more, or have taken enough incident-free rides to be flagged as ‘trustworthy’, you would get access to more comfortable, better-maintained vehicles. Different classes of ticket are endemic in air and rail travel, and I expect the same will happen here.
Your AI Boyfriend Doesn’t Love You, But You Will Love Him Anyway
of is interested in the phenomenon of AI assistants that not only help us carry out our tasks, but also help us feel cherished or sexy. The key words there are “help us feel”; there are no reciprocal feelings on the other side of the transaction. That may seem obvious to you. But, as Kevin shows, for many people, it is not obvious.
Machine intelligence is so new, and so alien, it is easy to hold a variety of opinions about it, and some are of the view, drawing on Turing and others, that successful mimicry of emotion is emotion. Kevin does a good job of suggesting why these views are mistaken, and an AI cannot actually love us.
But he also seems to suggest that while we should build our public policy on that foundation, our emotional lives can be conducted on other bases, if we so choose. Much of our emotional lives happen entirely in our own heads, so a really good imitation of interest and care will, for some purposes, be just the same.
To misquote Arthur C. Clarke: any sufficiently advanced performance of affection is indistinguishable from love.
That may warm your heart or chill it. If you’re not sure, best decide, as an AI future is coming in either case.
Heritage Designations for Buildings Are Bad
of cares enough about Austin that he volunteers as a member of its Zoning and Planning Commission. He does the work because he loves cities, their built form, and their history. This makes it interesting that he has opposed recent attempts to designate some Austin properties as heritage structures.
He comes by his opposition honestly, based on his previous involvement in the renovation of a 1924 movie theater in Brooklyn into a Montessori school. The project foundered on significant pressure to budget and timelines, thanks to the building’s designation as a heritage property. That status required the team to restore features they hadn't budgeted for, while simultaneously adding features to meet accessibility requirements. The project was finished, with millions of dollars of cost overrun, and delays in opening, depriving the community of a needed daycare. The terracotta mask of Comedy on the building’s façade smirked at these follies all the while. It’s this experience that informs Ryan’s concern about a trend he sees in Austin of applying heritage designations to dilapidated structures… in some cases, to prevent their demolition, in defiance of the wishes of both the owners and the neighbourhood.
Why prevent it, then? Ryan sees the best in people, I think. He argues that it has to do with people’s genuine desire to respect the past, and the history of the marginalized communities who lived in them, even if that is at the expense of the marginalized communities who live in them today.
Perhaps I’m more cynical than Ryan. I would say it is another expression of the desire—founded on misreadings of Jane Jacobs, our nostalgia for our childhoods, and the fear of our own inevitable senescence and death—that nothing in the built environment change. Whatever else happens, the city’s façade should remain as it is. That way it’s easier to pretend that we will remain as we are.
The problem with catering to this view is that it imposes high costs on others, and a high opportunity cost to the future. Ryan sourly observes that crumbling frames that might become housing today instead are left alone, to crumble further, as the owners are not permitted to demolish them, nor pay to rehabilitate them… meaning that attempts to preserve the neighbourhood simply make it worse.
At some point in the future, local rents will rise high enough to justify renovation, but to offset those costs, the block won’t get affordable housing; it will get “a Lululemon or artisanal juice bar. You know, something that reflects the historic character of the neighborhood…
“In all our efforts to preserve the past, we never seem to learn anything from it.”
Liability Laws Affect Everything
of set out to explore how liability should work for AI systems. He got there eventually, but first took a deep dive into the uniquely expansive liability system of the United States. As he puts it, this is a system that “shapes your life in ways you almost certainly have never realized”, and that came into being “without even a single law passing, without a single vote by any electorate or legislature”.
Dean begins by considering the 2014 case of Dewayne Johnson, a groundskeeper who was accidentally sprayed with the herbicide Roundup. He later developed cancer, and sued, and a jury awarded him millions of dollars in compensation… despite the fact that the EPA (and, for that matter, the WHO and the European Commission) classify it as an unlikely carcinogen.
Why should a judge and a jury think themselves capable of overruling the scientific and regulatory consensus on this matter? More importantly, why are they permitted to do so?
Dean traces the matter through the history of twentieth-century jurisprudence, which recognized, correctly, that in an age of multi-national corporations and complex products, the ‘little guy’ would have little scope to negotiate with, or prove negligence on the part of, the counterparties to all his transactions. Thanks to a few pioneering legal scholars, a new standard was set. “Suddenly, the tables were turned: so long as a consumer could prove that a harm occurred and that it was caused by the product in question, the manufacturer was to blame.”
The reformers envisioned a ‘race to the top’ for product safety, and they got their wish, but with a host of unintended consequences. They took complex questions of causation and gave them wholly over to judges and juries. They imposed strict liability without regard to whether the manufacturer intended it, or could have reasonably prevented it, or even if a regulator had certified a product as safe. They warped the insurance industry to the point where many things were no longer, practically, insurable, which is why so many municipalities lack pools and cities lack daycares.
Dean has gone on to investigate how liability might be best applied to AI. He has inspired me to do the same for automated driving, a subject I’ve engaged before and wish to revisit. But it’s worth understanding how this legal regime undergirds everything; as my wife, trained as a lawyer, likes to observe, once you become aware of the concept of liability, you see it everywhere.
The First Dog on Mars
The first four entries in today’s issue of Off-Ramps have been, perhaps, a bit sour, so let me clear your palate with something sweet.
of has gifted us with a delightful piece of science fiction. Taking its inspiration from “Lena” (a chilling story I have discussed before), Rob has prepared something charming, a Wikipedia entry from the future that tells the story of Maverick, an American beagle who becomes the first dog to live on Mars.
The story weaves together the politics of interplanetary colonization (competing activist groups like the “Intergalactic Kennel Club” and “Friends of Mars”), the practical challenges of space travel (designing dog-friendly habitats and handling launch procedures), and the psychological benefits dogs might bring to colonists suffering from “dead world syndrome.” Rob even includes charming details like Maverick's signature look deriving from native-to-Mars titanium crown replacements for cracked canine teeth.
I would like to see more work like this, which is entertaining, but also helps to train our imaginations to envision positive outcomes, and in concrete terms. By writing from the perspective of someone looking back on success rather than forward at obstacles, Rob shifts our thinking from “how could this possibly work?” to “how did they make this work?” I’m a fan of this kind of optimistic yet detail-oriented imagining… it helps makes audacious futures feel attainable.
(A parting gift: if you liked Rob’s piece, you may enjoy this essay by
founder .)