This presentation outlines the drivers of an algorithmic economy, the people who work on it and its challenges. I pay special attention to the risks of algorithmic error and safety, and the human supervisors in charge of managing it.
1. Making an algorithmic
economy work
Creating its successes, and fixing its errors
Juan Mateos-Garcia
6th December 2017
2. Surviving a new techno-economic paradigm
Sources: Wikipedia, Google n-grams
● What are its
economic
drivers?
● What’s its
workforce?
● What are the
opportunities
and
challenges?
3. In an information rich society, attention becomes the scarce resource
"...in an information-rich world, the wealth of information means a
dearth of something else: a scarcity of whatever it is that information
consumes. What information consumes is rather obvious: it consumes
the attention of its recipients. Hence a wealth of information creates a
poverty of attention and a need to allocate that attention efficiently
among the overabundance of information sources that might
consume it" (Simon 1971, pp. 40–41)
Sources: CMU, New York Times
4. Algorithms are a technology to manage excess information
Sources: Edureka
Systems that learn
from examples
Transform an
information input
into a prediction
(and an action?)
5. Some economic characteristics of algorithms
Sources: Facebook, Google
Transferrable
The good
Scalable
The not-so-good
Fallible
Gamable
6. Implications for the workforce
Sources: Autor, Levy and Murnaane
Cognitive content
Routine Non-routine
Social
(physical)
content
Routine
Assembly
line
Copy editor
Non
routine
Hairdresser Scientist
What characteristics of jobs complement / compete with
automation?
Workers who create and use algorithms
Workers who supplement and supervise algorithms
7. Function: Develop and apply algorithms
High creativity, high productivity
Is the education system prepared to prepare this group?
Workers who create algorithms: ‘The sexiest occupation in the world’?
Sources: Nesta / RSS / UUK (2014)
8. Organisational implications
Sources: Nesta (2015)
% improvement in productivity for firms with higher than average levels
in a variable (with all controls). All statistically significant.
Big implications for the organisation of the workplace as
well
9. Move fast and break things?
Sources: XKCD
Increasing evidence of algorithmic error and gaming in
the financial sector, media and society...
10. Workers who clean up after the algorithms: The worst occupation in the internet?
Sources: The Guardian, Nesta (2017)
Function: Detecting and fixing algorithmic errors and
situations where the system is being gamed
Low creativity, low productivity
11. Organisational implications
Sources: Google, YouTube, James Bridle
Supervision makes sense in high
stakes domains. Also makes
algorithmic decision-making less
scalable
Outsourcing supervision to users
makes it cheaper but also has
costs
Human supervision as an early
warning sign against algorithmic
failure
12. Conclusions and challenges
We need algorithms to operate in an information rich
world, but they are bringing with them new divides
between:
● Superstars and supervisors
● Objects and subjects
○ Individual / community level
○ Company level
Our ability to manage these tensions will determine if
we harness these technologies for good or end in a
dystopian scenario. It’s still up to us!
13. Questions
● How does your organisation use
algorithms to manage the
information overload?
● Are you making the most of their
scalability and transferability?
● What are your defenses against
algorithmic error?