The most interesting thing in tech: when can algorithms protect against bias and misuse better than humans can? There's a wonderful study about Uber drivers and NYC taxi drivers that show that the latter are more likely to take you on longer, less-direct routes, particularly if you're from out of town. Why? It's not that taxi drivers are less ethical. It's that Uber/Lyft and other services have more transparency and feedback. You can see on the map if the driver is taking a direct route and you can rate them. This doesn't always mean that algorithms are better, obviously. An Uber with a full monopoly and bad intentions could be vastly worse. But it's an interesting insight into how different systems—transparency, feedback loops, competition—can help us.
I rode in a taxi in New York today. I looked at Google maps and it said 64 minutes to get to my destination. The taxi driver went off the route shown by Google maps and I wondered how much extra time it would take me to get to my destination after he made the deviation. I go to my destination in 54 minutes and cut 10 minutes off my trip duration compared to the map. The taxi never looked at what a map said, he just knew the fastest way to go.
Nicholas Thompson Love this! Funny ‘taxi’ story… I moved to Paris in June 1989. I’d never been to Europe. This was my ‘year abroad’ (and was about to become my home for the next 22 years). I landed on a Sunday morning at around 6 am and immediately took a taxi from the airport. I gave the driver a slip of paper that had the address of the hostel I had lined up for my first days, and proceeded to look out the window and take in everything around me. We crossed into Paris within about 15 minutes (zero traffic) - but from there, it took another 90 minutes to reach my destination (and, no, not traffic). We drove by Sacré Coeur, Parc Monceau, the Arc de Triomphe, the Eiffel Tower (twice), and the Louvre. Yes - I gotten ‘taken’ for a ride. But it was the best damn 90 minutes of my life!
My write up on technology, dataism and algorithms. Click the link below to read my in-depth analysis: https://www.dailyparliamenttimes.com/2024/06/26/era-of-hacking-humans/
I traveled on business in the USA a great deal for several years, in the days prior to iPhone. There were drivers who would take you on an indirect route because they were trying to inflate the fare, and then there was the occasional driver who would take an indirect route because it was faster. Without an intimate knowledge of the city (and obviously in the days before smart digital maps) it was usually difficult to tell them apart. One night after landing at National Airport, I had a driver in D.C. take a route that I hadn't been down, before. Suddenly we were in traffic moving slowly. I looked out the window and realized he'd taken me through some sort of red light district, and hookers were all up and down the sidewalks and in the street, trying to get the attention of people in the cars. I have no idea why he thought that was a good idea. I didn't ask. At the time I assumed he had arrangements to get a kickback. I had a bit of a "rock star" look (long pony tail, biker boots, leather jacket) in those days, though, and maybe he thought I fit some sort of customer profile. 🤣
I’m not a New Yorker, but I visit New York almost every year and on my last trip, I decided to ride MTA buses. But when I used to take the taxi, most of them used to ride on purpose in crowded avenues when they were able to simply avoid them as per the maps instructions. It used to happen most of the time until I used to interfere and ask the driver ‘use the 8th avenue not the 6th etc.’ to show them that I am aware of the city’s transportation infrastructure.
I think this study as you describe it ignores pricing models? A ride share app has an algorithm that is designed to figure out the maximum amount you’ll pay for that airport ride and charge you that, which isn’t particularly transparent. For example if you make the same trip at the same time every day a taxi will charge you roughly the same amount; an Uber will increase the price every day until you stop riding.
Algorithms in environments where transparency is even possible or legal, seem to me to be the odd case. Ex: medical issues. Usually the imbalance of information makes the temptation to cheat a very low risk gamble. That is why we have such tight regs on aircraft design or medical DME. There is a trail of bodies leading to regulation. There are cases where your anecdotes apply, but that does not say it applies everywhere. We know nothing of most of the algorithms we encounter. And nothing of the AI used on us. All we know for certain is that there are vast areas in current AI and algorithms open to programming to take advantage of us, often with no liability.
CEO @ The Atlantic | Co-Founder, Keynote Speaker
2wHere's the study: https://pubsonline.informs.org/doi/10.1287/mnsc.2020.3721