Redesigning work in
an age of automation
Kevin McCullagh
Leaders of Change
05 December 2018
‘There certainly will be job
disruption. Because what’s
going to happen is robots
will be able to do everything
better than us. ... I mean
all of us’.
Elon Musk, National Governors Association, 16 July 2017
Automation
anxiety
‘Consider thou what the
invention could do to my
poor subjects. It would
assuredly bring them to
ruin by depriving them of
employment, thus making
them beggars’
Elizabeth I, on refusing to patent a knitting
machine invented by William Lee
There many
reasons to be
cheerful about
Automation
1
There many
reasons to be
cheerful about
Automation
Most jobs are
best tackled with
a mix of human
and machine
strengths
1 2
There many
reasons to be
cheerful about
Automation
Most jobs are
best tackled with
a mix of human
and machine
strengths
Most jobs will
be redesigned to
take advantage
of automation...
1 2 3
There many
reasons to be
cheerful about
Automation
Most jobs are
best tackled with
a mix of human
and machine
strengths
Most jobs will
be redesigned to
take advantage
of automation...
including design
1 2 3
Three
reasons
to be cheerful
Automation tends
to raise prosperity
and employment
1
Automation Productivity Prosperity
GDP per capita in England since 1270
Adjusted for inflation and measured in British Pounds in 2013 prices (000s)
1270 1400 1500 1600 1700 1800 1900 2016
Source: GDP in England (using BoE 2017), OurWorldInData.org/economic-growth
30
25
20
15
10
5
0
‘Productivity isn't everything
– but in the long run
it's almost everything’
Paul Krugman, Nobel prize winning Economist
Dismally low productivity growth
-2%
0%
2%
4%
6%
8% World
War I
World
War II
Great
Depression
Great
Recession
McKinsey Global Institute: Solving the productivity puzzle; Brookings Institution
United States
Europe
Great
Recession
Annual productivity growth
1870 1880 1890 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010 2018
Source: McKinsey Global Institute analysis 2017
Source: McKinsey Global Institute analysis 2017
98%
Automation tends to
eliminate tasks
Automation tends to
eliminate tasks and
create more jobs
Labour market is in
constant state of churn
US Census Bureau’s Dynamics Statistics, 2015
Annual job creation and destruction rates (US)
1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012
10%
12%
14%
16%
18%
20%
Automation often
makes work
more rewarding
2
Bank tellers vs. ATM machines
Fulltime-equivalent bank tellers
and installed ATM machines in the US
Tellers/ATMs(1000s)
500
400
300
200
100
0
1970 1980 1990 2000 2010
Source: James Bessen, How computer automation affects occupations:
Technology, jobs, and skills’, 22 September 2016, Vox
ATMs
Bank tellers vs. ATM machines
Fulltime-equivalent bank tellers
and installed ATM machines in the US
Tellers/ATMs(1000s)
500
400
300
200
100
0
1970 1980 1990 2000 2010
Source: James Bessen, How computer automation affects occupations:
Technology, jobs, and skills’, 22 September 2016, Vox
Fulltime
equivalent
workers
ATMs
New technology generally
reshapes jobs, rather than
replaces them.
New technology generally
reshapes jobs, rather than
replaces them. It takes
on the mundane tasks,
New technology generally
reshapes jobs, rather than
replaces them. It takes
on the mundane tasks,
as humans tend to move
onto more complex – and
often more meaningful –
work.
Pessimists
overestimate
machines, and
underestimate
humans
3
Less than 8%
of Toyota’s
production line
is automated
‘Machines are good
for repetitive things,
but they can’t
improve their own
efficiency or the
quality of their work.
Only people can.’
President of Toyota Manufacturing Plant,
Kentucky
Automation
– is expensive
– is highly inflexible
– creates quality
problems
Gorlech and Wessel
‘By 2029,
computers
will have
human-level
intelligence.’
Raymond Kurzweil, SXSW interview 2017
The Singularity
‘By 2029,
computers
will have
human-level
intelligence.’
Raymond Kurzweil, SXSW interview 2017
Narrow
Artificial
Intelligence
General
Artificial
Intelligence
Narrow
Artificial
Intelligence
General
Artificial
Intelligence
Moravec’s paradox
Hard easy
Moravec’s paradox
Easy hard
‘We can
know more
than we
can tell...’
Michael Polanyi, 1966
Human
intelligence
Artificial
intelligence≠
[I aim to make]
‘machines
slightly more
intelligent —
or slightly
less dumb.’
John Giannandrea, Head of AI, Apple
‘The real danger ... is not
machines that are more
intelligent than we are ...
The real danger is basically
clueless machines being
ceded authority far beyond
their competence.’
Daniel Dennett, ‘The Singularity—an
Urban Legend’, Edge
Overtrust
noun
The growing
tendency of
humans to place
unwarranted
trust in, and defer
to, automated
technology
J. C. R. Licklider
‘[people] will set the goals,
formulate the hypotheses,
determine the criteria, and
perform the evaluations.
‘Men will set the goals, formulate
the hypotheses, determine
the criteria, and perform the
evaluations.
‘Computing machines will do
the routinizable work that must
be done to prepare the way for
insights and decisions. . .
‘The symbiotic partnership will
perform intellectual operations
much more effectively than man
alone can perform them…’
J. C. R. Licklider, ‘Man-computer symbiosis,’ 1960
Most
work is
made
up of...
Machine
strengths
High data
situations
Human
strengths
Low data
situations
Most
work is
made
up of...
Machine
strengths
Following rules
Human
strengths
Judgement
Most
work is
made
up of...
Machine
strengths
Following rules
Analysis
Human
strengths
Judgement
Empathy
Most
work is
made
up of...
Machine
strengths
Following rules
Analysis
Speed
Human
strengths
Judgement
Empathy
Creativity
Most
work is
made
up of...
Machine
strengths
Following rules
Analysis
Speed
Accuracy
Human
strengths
Judgement
Empathy
Creativity
Improvisation
Most
work is
made
up of...
Machine
strengths
Following rules
Analysis
Speed
Accuracy
Repetition
Human
strengths
Judgement
Empathy
Creativity
Improvisation
Leadership
Most
work is
made
up of...
Machine
strengths
Following rules
Analysis
Speed
Accuracy
Repetition
Always on
Human
strengths
Judgement
Empathy
Creativity
Improvisation
Leadership
Mode shifting
Headline 30pt
Human
Machine
Interlace
5 types of
collaboration
Assigned
Assigned
– Certain tasks in a
human workflow
are outsourced
to a machine.
– The machine
completes
the task unaided,
with varying
levels
of instruction.
5 types of
collaboration
Assigned Supervised
Supervised
– Decision making
processes are
automated, but
under a human
eye.
– This mode
requires the
machine to be
aware
of and
communicate
risks and
unknowns
to human users.
5 types of
collaboration
Assigned Supervised Coexistent
Coexistent
– We will
increasingly
live and work
alongside
intelligent
machines,
sharing the
same spaces,
but focusing on
separate task-
flows.
– Machines in
these scenarios
must be able
to effectively
negotiate shared
space and
anticipate human
intent.
5 types of
collaboration
Assigned Supervised Coexistent Assistive
Source: Jaguar Land Rover Bike Sense. Seat shoulder taps the  rings a bicycle bell if it senses a cyclist near the car
and Door handles ‘buzz’ to prevent doors being opened into the path of bikes
Assistive
– Machines that
will help
us perform tasks
faster
and better.
– They support
particular
tasks in human
workflows, and
will excel in
discerning
human goals
and learning their
preferences.
Assigned Supervised Coexistent Assistive Symbiotic
5 types of
collaboration
Symbiotic
– This emerging
mode of
collaboration is a
highly interactive
and reciprocal.
– People input
strategic
hypotheses and
the machine
suggests tactical
options.
Job
Destruction
Humans vs machines
as rivals
Job
Re-design
Humans + machines
as allies
Reimagining
work
New
Human + Machine
capabilities
Machine
strengths
Following rules
Analysis
Speed
Accuracy
Repetition
Always on
Human
strengths
Judgement
Empathy
Creativity
Improvisation
Leadership
Mode shifting
Facilitating
automation
– Training
– Explaining
– Sustaining
Machine
strengths
Following rules
Analysis
Speed
Accuracy
Repetition
Always on
Human
strengths
Judgement
Empathy
Creativity
Improvisation
Leadership
Mode shifting
New
Human + Machine
capabilities
Facilitating
automation
– Training
– Explaining
– Sustaining
Human
augmentation
– Amplifying
– Interacting
– Embodying
Machine
strengths
Following rules
Analysis
Speed
Accuracy
Repetition
Always on
Human
strengths
Judgement
Empathy
Creativity
Improvisation
Leadership
Mode shifting
New
Human + Machine
capabilities
Facilitating
automation
Laying the groundwork
for effective automation
– Training
– Explaining
– Sustaining
Facilitating automation
Training
Teaching
machines how
to perform
tasks or act
more human
Job titles
– Automation design
anthropologist
– Data hygienist
– Empathy trainer
– Personality trainer
– Worldview trainer
– Interaction modeller
Activities
– Identifying relevant
data
– Cleaning data
– Tagging data
– Having machine
observe decision
making
– Improving machine
language
– Defining and
developing brand
AI personalityMellisa Cefkin, AV design anthropologist, Nissan
Facilitating automation
Explaining
Untangling
machine
decision
making and
translating to
stake-holders
Job titles
– Algorithm forensics
analyst
– Transparency
analyst
– Explainability
strategist
Activities
– Test, observe and
explain algorithms
– Make sense of
machine outputs
– Explain outputs to
stakeholders
IBM AI OpenScale
Facilitating automation
Sustaining
Ensuring
proper use of
AI, overcoming
setbacks,
stakeholder
management,
maintaining
momentum
Job titles
– Context designers
– AI safety engineers
– Ethics compliance
managers
– Automation ethicists
– Robot maintenance
technicians
Activities
– Commercial, ethical
and legal policy maker
– Ensuring data
and output quality
– Thinking critically
and holistically about
AI performance
and impact
‘Saftey green’ cobots for GM
Human
augmentation
Machines giving people
superpowers
– Amplifying
– Interacting
– Embodying
Human augmentation
Amplifying
AI enhances
the
effectiveness
of human
activities and
decision
making
Activities
– Automate repetitive
and low-level tasks
– Prioritise options
– Identify anomalies
and trends
Human augmentation
Interacting
AI agents with
advanced
voice-driven
interfaces
facilitate
interactions
between
people at
scale
Activities
– Answer customer
support FAQs,
and hand-on hard
questions to humans
– Accelerate customer
understanding based
on context
– Enable natural
language querying
SEB Aida chatbot
Human augmentation
Embodying
AI combines
with
sensors and
actuators to
allow robots
to safely and
effectively
physically
augment
human
workers
Activities
– Navigate around
humans
– Extend sight,
hearing and
touch
– Assist with
precise, repetitive
and physically
arduous work
Cobots at BMW
Human augmentation
Embodying
BMW researchers
found that human-robot
interactions in their car
plants were 85% more
productive than either
humans or robots on
their own.
Cobots at BMW
Redesigning
design
Design and tech careers are
forecast to be among the winners
McKinsey, 2018
Skills
Hours worked in
2016 (billions)
Change in hours
worked by 2030 (%)
Change in hours
worked by 2030 (%)
Hours worked in
2016 (billions)
Physical and manual
Basic cognitive
Higher cognitive
Social and emotional
Technological
90
53
62
52
31
113
62
78
67
90
-11 -16
-14 -17
+09 +07
+26 +22
+60 +52
Design
Machine
strengths
Following rules
Analysis
Speed
Accuracy
Repetition
Always on
Human
strengths
Judgement
Empathy
Creativity
Improvisation
Leadership
Mode shifting
Some
Design
Machine
strengths
Following rules
Analysis
Speed
Accuracy
Repetition
Always on
Human
strengths
Judgement
Empathy
Creativity
Improvisation
Leadership
Mode shifting
Augmented
creativity
Machine
strengths
Following rules
Analysis
Speed
Accuracy
Repetition
Always on
Human
strengths
Judgement
Empathy
Creativity
Improvisation
Leadership
Mode shifting
Redesigning design
with AI
Level of sophistication
2Empathise
AI uncovers new insights
from existing consumer or
user insight data
7Optimise
AI optimises parameters
1Discover
AI identifies new data
patterns and connections
6Test
AI lowers the analysis load
3Generate
AI created design options
within predefined constraints
8Customise
AI enables new levels of
personalisation
4Prototype
AI accelerates and
democratises prototyping
9Collaborate
AI facilitates more effective
collaboration
5Refine
AI accelerates iteration
and unlocks new creative
possibilities
10Hire
AI streamlines hiring process
1Discover
AI identifies new data
patterns and connections
Yossarian
2Empathise
AI uncovers new insights
from existing consumer or
user insight data
Crimson hexagon
3Generate
AI created design options
within predefined constraints
Autodesk Dreamcatcher
4Prototype
AI accelerates and
democratises prototyping
Aimybox
5Refine
AI accelerates iteration
and unlocks new creative
possibilities
Adobe Sensei
6Test
AI lowers the analysis load
Descript
7Optimise
AI optimises parameters
Nike Superfly elite spikes
8Customise
AI enables new levels of
personalisation
Stitch Fix
9Collaborate
AI facilitates more effective
collaboration
Marcel
10Hire
AI streamlines hiring process
Arya
The Human-Machine
Interlace should make
work more human
Champion
human strengths
in an age of
automation
We join the dots
www.plan.london
@kevinmccull

Redesigning work in an age of automation