Autonomous Vehicles viewed through a lens of Antifragility
AntiFragility is not the same as Resilience
I recently came across Nassim Nicholas Taleb’s concept of Antifragility. His book is about 5 years old and I have not yet read it earnestly. I encountered Taleb’s Antifragility in ongoing attempts to learn more about Daniel Kahneman’s ideas about Prospect Theory and new thinking (via Michael Lewis’ The Undoing project) about how people make decisions in uncertain conditions.
Kahneman’s collaboration with Amos Twersky led him to receive the 2002 Nobel Prize in economics.
Taleb’s Antifragility is a more systems level approach to designing within complexity where the consequences of error can be catastrophic.
The video below is a fascinating cognitive choreograph where Taleb introduces his book and ideas alongside Kahneman.
This post is an attempt to view Autonomous vehicles through Taleb’s Antifragility frame while also borrowing some of Kahneman’s ideas about how people make decisions under uncertainty.
An Introduction to Antifragility
In the NYPL video, Taleb introduces AntiFragility by describing a glass of water on the table in front of him.
He explains that we all perceive this glass to be fragile while the table it is placed on relatively, less fragile.
Fragility here is a state of the glass and table when exposed to some shock (say, an earthquake) where the glass is less likely to be a glass while the table is more likely to stay a table in the event of an earthquake and the glass and table are both unlikely to benefit from this inertial uncertainty of the earthquake.
Benefit and Uncertainty are operative terms here and elegantly explain the presence of Taleb and Kahneman in the same room.
Taleb has uniquely and financially benefitted from multiple stock market crashes (a most empirical outcome of systemic financial uncertainty) leading to his award winning book “Black Swan” and “Fooled by Randomness” while Kahneman has benefitted from his study of human decision-making under uncertainty. (The Nobel Prize)
Taleb goes on illustrate how we lack an adequate epistemology or simply a category to describe systems or objects that inherently benefit from a lack of order and goes on to prescribes how we can design systems (and our personal lives) to benefit or even thrive from disorder.
It is argued that the building blocks of Life and the Cambrian explosion is a product of AntiFragility caused by the tectonic disorder in the Cryogenian period.
“The appearance of complex, multicellular animals is inextricably linked to a boom in algae enabled by the same destructive forces that made the Cryogenian seem so hellish.”
Organ Transplants and Autonomous vehicles
This Slate article is revealing in that it shows how a small section of the healthcare system (Organ Transplants) benefits from disorder in a non-autonomous transportation system.
In other words, the inherent uncertainty of human driving has benefits to the organ transplant system. Removing this uncertainty creates some harm where up to 20% of immediately available organs are lost in a world where traffic deaths approach zero.
The Aviation industry and Children of the Magenta
Taleb further talks about how the Aviation Industry as a whole is a fine example of AntiFragility at work. Consider that the outcome of any plane crash leads to some direct improvement of the future aviation industry. In some way or the other, we all owe our current trust and safety in air transportation to these past human tragedies. At the risk of sounding crude, transporting people safely in the air (and space) has directly benefited from aviation accidents.
However, automation has also led to some disturbing fragilities. The paradox of automation as described in the 2-part episode of 99% invisible is that automation leads us to being…
“locked into a cycle in which automation begets the erosion of skills or the lack of skills in the first place and this then begets more automation.”
This erosion of pilot skill was what alleged caused Air France flight 447 to perish.
How Uber can lead to an AntiFragile Tech sector.
Uber’s missteps this past decade can be perceived as a signal on AntiFragility. From its partnership (or lack thereof) with cities around autonomous vehicles to its widely criticized leadership culture — if we treat these events as equivalent to memorable plane crashes and in response, piece together a better model for public-private partnerships and leadership culture, we could find ourselves in an AntiFragile Tech ecosystem.
However, if we continue to reward toxic behavior either socially or financially — the public loses disproportionately while bad actors get away with a slap on the wrist. This basically sums up Taleb’s views on the financial sector post Great-recession.
So what does this mean for Autonomous Futures?
Cars, Computer companies, and increasingly Cities are currently racing towards realizing any and all forms of autonomous futures that include robots in the classroom to self-driving cars on our streets.
The dominant narratives of these futures include a culture that legitimately needs and worships Machine Learning, Artificial Intelligence, and Data Science talent.
Most folks who design these systems, approach them from a strong bias of aggressively trying to remove all uncertainty. In doing so, they unconsciously create systemic fragilities that we are yet to identify.
It is becoming increasingly more important to take stock of what is lost by removing uncertainty from our streets and cities and how this affects policy making around autonomous futures.
This piece from Google Research is a refreshing take on how many large Machine Learning systems are actually quite fragile in terms of an increasing amount of Technical Debt they take on to maintain over time in an environment of monocultural certainty.
It is this approach to systems design that leads to Google Photos misidentifying people for Gorillas. What can we expect next? The list of AI sourced errors will only increase.
The key question is how do we, the public and the institutions we fund, react to these mistakes? It’s ultimately up to us on how we collectively decide to take AI mega-corps to task.
Should the Consumer Finance Protection Bureau consider expanding to Consumer Bias Protection Bureau?
Will future felonies include tampering or recreating unauthorized reproductions of street signs or pavement markings just because they screw up how AVs work?
As we transition towards digitizing street networks and pushing to automate urban transportation, what do the Flash Crash equivalents of our Autonomous futures look like? and how do we design AntiFragility to avoid them?
Towards the end of the NYPL talk, a question was asked to both Taleb and Kahneman about how Hurricane Sandy caused gas shortages in New York City and how could we re-design our cities to be Antifragile in the face of climate change.
In their responses lay something crucial about our collective misunderstandings.
Taleb, the progenitor of the Antifragility concept chose to talk about how maintaining increased fuel stockpiles is Antifragile while Kahneman countered by saying that that true Antifragility comes from preserving a more diverse form of transportation.