Slides from preso at GovInnovate 2017-10-11

1. Headline Verdana Bold
The future of work
AI, robots, and creating good jobs
Peter Evans-Greenwood, 2017-10-11
Who are C4tE
• An explorer, as every team needs an explorer
• ‘now-ists’ rather than ‘futurists’
We’ve been looking into the future of work, as have many pundits, due to a combination of:
• existential angst, as automation seems to be replacing good jobs with bad jobs (sharing economy, growing
promotion of part time & casual work…)
• the somewhat recent emergence of AI, threatening many white collar jobs that were previously thought safe
Reactions have typically fallen into two camps:
• Doom: what few new jobs will be created will be highly technical (making the machines) and many (possibly a
majority) of people won’t be able to find work
• Can we retrain bus drivers to create autonomous busses?
• Radical interventions will be required – lie UBI – to keep society whole
• Utopia: new jobs will be created for a range of capabilities, we’ll all be gainfully employed
• This is historical norms reasserting themselves
• Though we can’t see where these jobs will come from, so it is a leap of faith
Both of these opinions are founded on conceiving work as collections of task
• it's tasks that are automated, not jobs, as astute pundits are pointing out
• consequently, the focus for much analysis is to determine
• based on knowledge and skills that are unique to humans (creativity etc)
• what tasks can be expected to require these knowledge and skills
• predictions are then based on what are thought to be unique tasks
However, we’d like to challenge this, as in our research we discovered that:
• there is no knowledge of skill unique to humans, none
• More to the point, if we can define the task then we can automate it (given cost-benefit)
Every time a line is drawn in the sand, AI steps over it:
• Chess, Jeapody, Go
• Financial advice
• Developing naval strategy & tactics
Even now AI is being developed that is ‘creative’
• started with writing news or other semi-formal reports
• now composing music, poetry
Sure, it’s a narrow definition of creativity but, as always, what was dismissed as impossible is becoming
commonplace
What we also discovered in our research is that framing work in terms of tasks an processes that is the problem:
• AI, after all, isn’t a ‘task performing’ technology like many in the past
• AI is better of as automating a behaviour – a response to a change in the environment – and must be wrapped in
other technologies before it can perform a task
It is possible that the most effective use of AI is not simply as a means to automate more tasks, but as an enabler
to achieve higher-level goals, to create more value
The advent of AI makes it possible—indeed, desirable—to reconceptualise work, not as a set of discrete tasks laid
end to end in a predefined process, but as a collaborative problem-solving effort where humans define the
problems, machines help find the solutions, and humans verify the acceptability of those solutions.
So what I’d like to propose today is that we choose this third option.

2. 2
Power loom weaving. Public domain.
https://commons.wikimedia.org/wiki/File:Power_loom_weaving._Wellcome_L0011293.jpg
A set of power looms:
• A weaver responsible for ~2-5 looms with specialised helpers
We forget that our current approach to work – task specialisation – is a fairly recent invention. It only really took off
with the industrial revolution, like above.
• make the watch spring rather than the entire watch
Indeed, if I was to pick one idea that I consider most essential to the industrial revolution, then this would be it.
Task specialisation:
• makes if worthwhile for the worker to discover superior techniques
• provides the standardised environment required for mechanisation
➡ mechanisation improves precision (less waste, cost out)
• which a precondition for automation
➡ automation improves capacity (productivity up)
One important affect of mechanisation and automation is that the remaining (manual) tasks become more
important as they take a greater proportion of the worker’s time
• created something of a virtuous cycle where the workers would improve techniques and identify opportunities for
mechanisation (and then automation)
• indeed the majority of productivity gains came from this ‘learning by doing’
• 2.5x invention vs 20x learning by doing, for the power loom
Nor did this result in fewer weavers
• improved productivity resulted in lower prices
• lower prices stimulated demand
• improved demand stimulated production
with the population of weaves only peaking in the 70s
We forget that prior to the power loom the majority of the population had few clothes, typically just the clothes they
were wearing, as clothes were expensive. It wasn’t uncommon for many family to pawn their winter cloths in
summer, and their summer cloths in winter. Productivity improvements due to the power loom is a significant
contributor to the fact that we now all have all the clothes we need (or want). As the economists tell us, its
productivity improvements via innovation like these that improve our quality of life.
Today, though, this virtuous cycle seems to have ground to a halt. Productivity grown appears to have reverted to
preindustrial levels, something concerning the economists as its productivity growth through innovation that
improves our quality of life.
Many explanations have been proposed:
• measurement problems
• the exhaustion of ‘one time’ technologies
None is entirely satisfying though
Something we haven’t considered though, is if the task-based approach to constructing work has run out of steam
• the bad jobs we’re seeing might be the result of firms trying to schedule in ever narrower slices of time -> The
Good Jobs Strategy
• the lack of productivity growth due to our inability to capitalise on these newer ‘AI’ technologies
We’re creating jobs that a good for neither human nor machine.

3. 3David Lapetina
David Lapetina: https://commons.wikimedia.org/wiki/File:Chess-king.JPG
AI came to many people’s attention when Deep Blue defeated Garry Kasperov in 1996-1997
• One magazine called it “the brain’s last stand”
Eight years later, in response, the chess community created freestyle chess, a team sport with teams containing
both people and computers
Everyone assumed that it would be the most skilful player with the most sophisticated computer that would win as
chess is a game of knowledge and skill
• they were all wrong, the winners were a couple of competent players with a couple of ok chess computers
The secret – as it turns out is that
• it’s not how skilful the individual are that matters, but how they work together
This is something we saw again recently when doing some background research on a report on ‘should everyone
learn how to code’
• solutions created by a people and computers tend to be superior to those created by people or computers alone
• this appears to be general trend across domains
Working with AI to solve a problem forces us to create, to externalise, a model of the problem to be solved
• we build the model incrementally, tweaking it as we discover more about the problem
• AI (digital) behaviours respond to changes in the model, making suggestions, correcting mistakes, searching for
options…
• Human behaviours enrich the model, integrating new observations, defining/refining terms, evaluating options,
discovering new connections
So, if we’re to draw a line between humans and machine then it should be in terms of attitudes and behaviours,
rather than knowledge and skills
• humans are the repository of the social behaviours that enable us to explore the world around us, notice the new
and unusual, and create new knowledge
• machines are the repository of instrumental behaviours: identifying known patterns, enumerating options, and
applying knowledge
It’s this social ability to create new knowledge that separates us from the machines.
• knowledge is a social construct
Which brings us to the distinction between task-based and behaviour based work
• work built on tasks is designed to find correct or optimal solutions to well-specified problems
• work built on behaviours is designed to be effective in a complex world, in accordance with possibly many
objectives and constraints, making good use of limited resources to produce a timely and useful, rather than
optimal (but potentially late), outcome.
In our current model of using tasks and processes to define what the final outcome will be we essentially limit the
possibilities and the value created.
A good analogy is:
• behaviour-based work is a team of workers standing around a shared whiteboard, each holding a marker,
responding to new stimuli, carrying out their action, and writing (or drawing) their result on the same board.
whereas
• task-based work is a bucket brigade where the workers stand in a line and the ‘work’ is passed from worker to
worker, with each worker carrying out their action as the work passes in front of them.
The question then is: how does this translate to the world of work?

4. 4
The image is released free of copyrights under Creative Commons CC0.
https://pxhere.com/en/photo/623875
Consider ‘a happy retirement’.
Superfunds and the like might think that they sell financial products, though what they really sell is the promise of a
happy retirement. There’s a disconnect between the
‘product’ and the ‘problem’ which means that many folk are disengaged. It’s all to hard and quality financial advice
is expensive, so many of us don’t both and remain on the default option.
It is hard to come up with a solid definition of ‘happy retirement’, other than the recursive ‘one in which one is
happy’. We need to go from wanting a happy retirement through:
• what will actually make me happy, as opposed to what I think will make me happy
• what are reasonable expectations
• how can I change my behaviour now to have the future I want
before we reach quantifiable data like income streams and desired requirement income
Robo advice – which asks a bunch of questions trying to elicit the quantifiable data – can’t do this
The problem is that the individual doesn’t know what their happy retirement is. While they might have preferences,
these need to be grounded. New knowledge needs to be created.
What we need is something like a Socratic dialogue
• someone who can prod, poke and understand us to help us understand ourselves
to the point that we’ve teased out the details
• income streams etc
At which point roboadvisor can take over
Ideally the human advisor would start capturing these details in a model at first conversation, while digital (AI)
behaviours respond to the details and present options
• applying actuarial models
• applying different investment strategies
• …
enabling advisor and the client to play what-if and explore options and they find a solution
Similar to freestyle chess, we can hope that the solution created would be superior to that created by human or
machine alone
Similar to the power loom, we can expect that the automation of these simple behaviours while improve
productivity, reduce costs and extend high-quality financial advice to more people, making it more equitable
So:
• if the industrial revolution as characterised by products and progressive definition and automation of tasks, then
• the next revolution will characterised by problems and progressive definition and automation of behaviours
The challenge then is to reconstruct jobs along these new principles

5. 5Patrick Despoix
CityMobil2 véhicule expérimental sans chauffeur en situation de test à La Rochelle Charente-Maritime France
Patrick Despoix: https://commons.wikimedia.org/wiki/File:020_-_CityMobyl2_-_La_Rochelle.jpg
So what does it mean to build work on behaviours, rather than tasks?
Consider autonomous busses
• Predictions are that autonomous vehicles are going to be a disaster for professional drivers
• This, however, ignores the fact that ‘driving’ is only a part of what a bus driver does
Bus drivers also deal with:
• challenging weather; heavy rain or extreme glare, when image recognition, LIDAR and RADAR as insufficient
• unexpected events – accidents, road work, or an emergency – that require a human’s judgement to determine
which road rule to break
• routes might need to be adjusted, anything from teaching the bus where a temporarily moved stop is through to
modifying routes due to an incident or roadwork.
• a human presence might be legally required, from monitoring underage children through representing the vehicle
at an accident.
Rather than replace the driver let’s accept that automation will replace the ‘simple’ behaviours:
• lane following, separation maintenance, route following, etc.
• adhering to a schedule, or, if frequent enough, the collection of busses might behave as a flock
As with the power loom, this breaks the requirement for a bus driver to be constantly present. Rather than drive
one bus they can drive a collection of busses:
• These busses could all be on the same route. A mobile driver (on a motor scooter) might be responsible for 4-5
sequential buses on the route, following along zipping between them as needed, managing accidents and other
events, or dealing with customer complaints (or disagreements between customers).
• The driver might be responsible for all busses in geographic area, on multiple routes, dealing with problems over
a few blocks.
• We might split the work, creating a desk-bound ‘driver’ responsible for remote operation of a larger number of
busses, while mobile and stationary drivers restrict themselves to incidents that require a physical presence.
School or community busses, for example, might have remote video monitoring while in transit, complimented by
a human presence at stops to help passengers embark and disembark.
We should note that these jobs do not require training in AI or software development. It’s a shift from driving
busses to shepherding busses.
The question then is: what to do with the productivity benefit?
• take the saving and make a currently subsidised form of public transport profitable
• make an unreliable form of public transport reliable by increasing frequency, improving quality
• transform public transport and eliminate public transport shadows, making it more equitable
As with the power loom, its possible for the right choice to result in increased patronage, a higher quality of life,
and a more equitable system that creates more jobs, not destroy them.
The choice, though is ours.

6. Issue 20 | 2017
Complimentary article reprint
Cognitive
collaboration
Why humans and computers
think better together
About Deloitte
Deloitte refers to one or more of Deloitte Touche Tohmatsu Limited, a UK private company limited by guarantee, and its network of member firms, each of
which is a legally separate and independent entity. Please see http://www/deloitte.com/about for a detailed description of the legal structure of Deloitte Touche
Tohmatsu Limited and its member firms. Please see http://www.deloitte.com/us/about for a detailed description of the legal structure of the US member firms
of Deloitte Touche Tohmatsu Limited and their respective subsidiaries. Certain services may not be available to attest clients under the rules and regulations of
public accounting.
Deloitte provides audit, tax, consulting, and financial advisory services to public and private clients spanning multiple industries. With a globally connected
network of member firms in more than 150 countries and territories, Deloitte brings world-class capabilities and high-quality service to clients, delivering the
insights they need to address their most complex business challenges. Deloitte’s more than 200,000 professionals are committed to becoming the standard of
excellence.
This communication contains general information only, and none of Deloitte Touche Tohmatsu Limited, its member firms, or their related entities (collectively,
the “Deloitte Network”) is, by means of this communication, rendering professional advice or services. No entity in the Deloitte network shall be responsible for
any loss whatsoever sustained by any person who relies on this communication.
Copyright © 2017. Deloitte Development LLC. All rights reserved.
By James Guszcza, Harvey Lewis, and
Peter Evans-Greenwood
Illustration by Josie Portillo
6
To code or not to code,
is that the question?
2017
Issue 21 | July 2017
Complimentary article reprint
Reconstructing
work
Automation, artificial
intelligence, and the essential
role of humans
About Deloitte
Deloitte refers to one or more of Deloitte Touche Tohmatsu Limited, a UK private company limited by guarantee (“DTTL”), its network of member firms, and their
related entities. DTTL and each of its member firms are legally separate and independent entities. DTTL (also referred to as “Deloitte Global”) does not provide ser-
vices to clients. In the United States, Deloitte refers to one or more of the US member firms of DTTL, their related entities that operate using the “Deloitte” name in
the United States and their respective affiliates. Certain services may not be available to attest clients under the rules and regulations of public accounting. Please
see www.deloitte.com/about to learn more about our global network of member firms.
Deloitte provides audit, tax, consulting, and financial advisory services to public and private clients spanning multiple industries. With a globally connected
network of member firms in more than 150 countries and territories, Deloitte brings world-class capabilities and high-quality service to clients, delivering the
insights they need to address their most complex business challenges. Deloitte’s more than 200,000 professionals are committed to becoming the standard of
excellence.
This communication contains general information only, and none of Deloitte Touche Tohmatsu Limited, its member firms, or their related entities (collectively,
the “Deloitte Network”) is, by means of this communication, rendering professional advice or services. No entity in the Deloitte network shall be responsible for
any loss whatsoever sustained by any person who relies on this communication.
Copyright 2017. Deloitte Development LLC. All rights reserved.
By Peter Evans-Greenwood,
Harvey Lewis, and James Guszcza
Illustration by Doug Chayka
Most of this was drawn from the following reports, along with an additional report that we’re preparing for
publication now.
• Feel free to contact me or one of the team if you would like a copy.
• Or download them from the web site.

7. This publication contains general information only, and none of Deloitte Touche Tohmatsu
Limited, its member firms, or their related entities (collectively the “Deloitte Network”) is, by
means of this publication, rendering professional advice or services.
Before making any decision or taking any action that may affect your finances or your
business, you should consult a qualified professional adviser. No entity in the Deloitte Network
shall be responsible for any loss whatsoever sustained by any person who relies on this
publication.
© 2017 Deloitte Touche Tohmatsu. Peter Evans-Greenwood