Showing posts with label Antifragility. Show all posts
Showing posts with label Antifragility. Show all posts

26 April 2026

(AI) Technology doesn't democratize efficiency. It benefits early adopters

Technology doesn't democratize things. Initially, it benefits early adopters. Those with the resources and education to adopt and adapt fast. 


The new tech buzzword is: AI agents make everybody more efficient.

Yes, it does. But not equally.


It's a good punchline and buzzword, saying that technology democratizes things - information, knowledge, education, transportation, resources, communication, or, with contemporary AI agents: work efficiency.


But this is rarely true.

I don't remember any technological innovation who really democratized things - at least at first. 

Technology tends to benefit first the rich and the educated - those who already have the capacity to invest, absorb and adapt faster. 


Yes, technology trickles down, eventually. But not immediately. 


It was the same with automobiles - first adopters were rich people; later beneficiaries were industrialists who sold (Ford Model T) affordable cars to the working class. 

Same with writing, printing, books, and newspapers: for hundreds of years, the beneficiaries were still the upper and middle classes, who were already literate, or had the resources to educate their children. For hundreds of years after Gutenberg, the majority of the population remained illiterate and didn't really benefit. Intergenerational mobility was significantly lower at the time, when education was a luxury, and most children had to start working VERY early in factories or agriculture. 


Same with the internet and email: first few years, it was reserved for universities, researchers, top tech entrepreneurs, academia, financial magnates - not everybody. 


So no, AI is not for everybody, it doesn't democratize first, and it doesn't benefit everybody equally. We have to move and adapt fast.


New relevant study (from Marius Comper): 

https://www.ft.com/content/0873e3cb-cb02-4b47-941f-14da74149670?fbclid=IwdGRjcARa7tVjbGNrBFruyGV4dG4DYWVtAjExAHNydGMGYXBwX2lkDDM1MDY4NTUzMTcyOAABHtvpmg9E9NL4vg8-W7XJcFqyz5I2YLwqWuTkFra4mwAO2yOhKZZzgCDBPoPh_aem_WAnPyG78hQSzOTJtuOIDVw


See also discussion 

https://www.facebook.com/share/p/18YjsXCRPB/


02 June 2021

Covid recovery and resilience plans, and eurocratic slang - a pedantic linguistic approach

The EU's Recovery and Resilience Facility was approved in February with a budget of €672.5 billion. It targets the post-covid return to normal, repairing the economy, and making Europe more prepared for future similar crises. 

Most European countries already submitted to the European Commission their national recovery and resilience plans. Everybody talks about them, but most European citizens probably consider the words yet another eurocratic gibberish. So here we go.

Recovery is the short-term action of returning a system to a previous state.

Resilience is used by most people as a buzzword - similarly to sustainability. So let's take it a step higher, from vulnerability.

Vulnerability management is a relatively new niche domain, that deals with negative (external) events, analyses their impact, and the system's capability to cope with them. It is closely related to management, and particularly to risk. Marle and Vidal proposed an interesting analysis and management framework.

There are several definitions; I prefer these:

  • Resistance is a static characteristic of a system, that refers to its capacity to withstand instantaneous damage incurred by external negative events.
  • Resilience is a dynamic characteristic of a system, that refers to its capacity to recover in time to a previous state. 

So here is the first distinction between recovery and resilience. Recovery is a short-term action; resilience is the characteristic, the capacity of a system to recover. In a way, the meta-recovery.
But even academics use the words interchangeably. The domain is quite new, the concepts and terminology are still fuzzy. 

Interestingly, both resistance and resilience are passive characteristics, and they focus on negative aspects and events. Enters antifragility.
Antifragility is the capacity of a system to not only resist to, or recover from, adverse events; but also to improve because of adverse events. The concept was introduced by Taleb in 2012. 
A nice example of antifragility is physical exercise, cardio or weight training. We purposely exercise, i.e. stress our muscles outside their comfort mode, so that they become stronger. The muscles initially mobilize existing resources to cope with the stress. It uses local glucose, brings more oxygen, then produces and brings even more sugars. After some hours, you get muscle soreness - when muscles enter repair/recovery mode. And then the muscles grow even stronger, they improve.

Image source is freepik.com
Obviously, these terms and concepts are reasonably new, so their meaning and nuances can hardly be grasped from mainstream dictionaries.
As any serious political initiative, the EU's recovery and resilience plan already has a good touch of positive thinking. It supports not only recovery, but also resilience. It talks about sustainability, about preparing Europe for future challenges, as well as opportunities. Which is nicely aligned with modern risk management, as well as to my personal passion for positive complexity and the bright side of risk.

19 March 2021

Target fixation in risk management. Arguments for the bright side of risk

Always look on the bright side of life

Target fixation
is the tendency to hit an obstacle because of excessive attention to it. 

The source is practical experience of fighter-bomber pilots in WW2. It is popular with racing and motorcycle schools: "always look where you want to go, not where you are going - or else the bike won’t turn"; "never look directly at the obstacle, or else you hit it" . 

It is related to tunnel vision, which is a medical condition: the loss of peripheral vision, while retaining central vision. Or, in a generalized form: diminished awareness of additional objects because of fixating on a specific target.

Target fixation and tunnel vision are relevant metaphors for cognitive biases regarding the relation between objectives, expectations, and results; e.g. the Pygmalion effect - a psychological phenomenon wherein high expectations lead to improved performance.

Risk management is a negative concept

Risk management is a key knowledge area in project management. Its terminology has negative connotations. It typically deals with negative external events. Its mathematical apparatus uses negative numbers. 

Focusing on negativity violates key empirical rules in all sort of domains. A key rule in communication is: never start a sentence with a negative word. 

A brief history of risk and opportunity management 

  • “Risk management” appears in scientific and management literature since the 1920s. 
  • It became a formal science in the 1950s (Dionne, 2013), when articles and books with “risk management” in the title also appear in library searches. 
  • Most of research was related to finance and insurance.
  • Opportunities first appear in academic research or management books in the 1990’s. The first Project Management Body of Knowledge (PMBoK, 1987) doesn’t mention opportunities at all. 
  • Modern project management school does recognize the importance of opportunities, in theory. Opportunities have been included in project management literature since the 1990s (PMBoK, 1996) and became a significant part of project risk management in the years 2000s (Goodman, 2005), when articles titled “opportunity management” also begin to appear in library searches. 

Modern risk management theory deals with any type of external events, positive and negative. Positive risks are called opportunities. Similarly to risks, opportunities have specific mitigation strategies: exploit, share, enhance, ignore.

In practice, risks are considered “usually negative”. Risk-related research and practice focus significantly more on threats than on opportunities. 

Negative risk management is not constructive. It misses out on the important category of optimistic bright positive events. 

The lack of opportunities from risk management practice is actually fascinating. It is a widespread phenomenon. You can find practically no opportunities in any of the risk registers of any actual project. I witnessed this phenomenon throughout my career; it is also documented by research.

Other optimistic management approaches: antifragility, positive complexity

In vulnerability management, (Taleb, 2012) introduced the concept of antifragility as a new type of response. Antifragility is beyond the traditional resistance, resilience (recovery), or robustness: it is about systems improving as a result of adverse external events. It is a convex response to external stressors, leading to positive effects.

Complexity management proposes an optimistic approach in the form of the appropriate (requisite) and positive complexity (Benbya & McKelvey, 2006) (Morcov et al, 2020). The law of requisite variety (or complexity) says that organizations, or in general systems, must increase internal complexity to match (or even exceed) external complexities. Positive complexity is the one that creates beneficial results such as increased viability, innovation, or increased functionality: smarter phones or cars. 

Conclusion

Risk management is a negative process; it deals almost exclusively with negative events. This triggers cognitive biases and failure-prone behavior such as target fixation and tunnel vision. Projects and managers focus on obstacles and ignore opportunities.

Maybe it's time to make a leap of faith towards a positive optimistic approach to risk management; to finally jump to opportunity management. To start fixating on positive targets.

Or, as Monty Python would say: 

Always Look on the Bright Side of Risk.

The engineering view: AI is amazing, but not human. It's a tool. It's not perfect. It works

The engineering view: AI is amazing, but not human. It's a tool. It's not perfect. It works. This text by Andrew Ng is so good, that...