MIT SMR CONNECTIONS

M A N AG E R ’S G U I D E

Transform Your Workforce With

Skills for Machine Learning

ON BEHALF OF:
 MANAGER’S GUIDE — TRANSFORM YOUR WORKFORCE WITH SKILLS FOR MACHINE LEARNING

C O N T E N TS
Transform Your Workforce With Skills for Machine Learning............................................ 2

1. Assess current workforce skills and identify gaps.

2. Identify individuals with promise for data science roles.
3. Build data literacy across the organization.
4. Explore diverse options for training in machine learning skills.

Machine Learning Upskilling Checklist............................................................................ 5

Sponsor’s Viewpoint: Training and AI Services Can Help Close the Skills Gap............. 6

MIT SMR Connections develops content in collaboration with our sponsors.

It operates independently of the MIT Sloan Management Review editorial group.

Copyright © Massachusetts Institute of Technology, 2020. All rights reserved.

MIT SMR CONNECTIONS

 MANAGER’S GUIDE — TRANSFORM YOUR WORKFORCE WITH SKILLS FOR MACHINE LEARNING

Transform Your Workforce With

Skills for Machine Learning

T
he mandate to implement machine learning is growing hires, partnerships, or acquisitions. But they overlook a more
as CEOs and boards across many business sectors see obvious answer: upskilling their current workforce. That up-
the results it’s producing at the leading edge. Those in skilling requires action on two parallel paths:
the vanguard are using machine learning systems to generate
real-time insights for better decision-making, and to automate • Enterprises must identify, train, and elevate those with
rote processes so that employees can spend more time creating technology skills and talent to work directly on developing
value for customers. machine learning applications.
• Organizations must intentionally build data literacy across
But leaders seeking to implement machine learning in the the entire workforce so that those with business-domain
enterprise, and glean real business value from this area of arti- expertise can successfully collaborate with machine learning
ficial intelligence (AI), are confronting the reality of a skills gap: experts, and those in process/operational roles can work
There’s a shortage of qualified talent. According to an Accenture with AI-based systems in production.
survey on the future workforce, 67% of workers consider it
important to develop their own skills to work with intelligent This guide provides a framework for action on closing the machine
machines. Yet only 3% of business leaders are planning learning skills gap by developing capabilities and aptitudes already
significant increases in their training and re-skilling programs present in an organization’s workforce. Here’s where to begin.
to meet the skills challenge posed by AI.
1. Assess current workforce skills and identify gaps.
Creating Learning Pathways for Employees Performance reviews and employee screening can help busi-
Many businesses bridge the machine learning skills gap by nesses assess their current workforce’s skills. But as demand
acquiring expertise from outside the company — through new for AI experts grows, many businesses are deploying innovative
instruments to gain clarity around capability gaps and determine

“As demand for AI experts where training may be needed to provide current workers with
the skills required by a machine learning practice.
grows, many businesses
are deploying innovative One enterprise that has fielded such an instrument is PwC,
which uses its Digital Fitness app to give employees a person-
instruments to gain clarity alized assessment of their technology acumen and guide them

around capability gaps and to the customized educational content that will help them
close any knowledge gaps. By identifying employees’ areas of
determine where training strength and weakness, the app not only offers workers a tai-

may be needed.” lored learning path but also provides PwC with a baseline as-
sessment of its workforce’s digital IQ — a jumping-off point for
skills development and training programs.

MIT SMR CONNECTIONS 2

 MANAGER’S GUIDE — TRANSFORM YOUR WORKFORCE WITH SKILLS FOR MACHINE LEARNING

2. Identify individuals with promise for data science roles. Software engineers are also more likely to move seamlessly
A statistical background, strong math skills, and an ability to into machine learning roles and are well worth investing in.
code are key indicators of an aptitude for machine learning. Typically self-directed learners, Pierson says, “they are always
But businesses that only consider hard skills when seeking embedding themselves with machine learning engineers or
candidates risk overlooking some important attributes that machine learning scientists and actually trying to learn [ma-
may recommend an individual for additional training and de- chine learning] on their own.”
velopment. In addition, the emergence of modeling tools that
require less specialized knowledge may expand the pool of 3. Build data literacy across the organization.
people who could be considered for this work. Systematically boosting data literacy across an entire work-
force helps those with functional expertise and institutional
For instance, employees who exhibit creativity, curiosity, and knowledge to better collaborate with machine learning experts
perseverance are more likely to possess the interpersonal and — a critical success factor for moving beyond pilots and exper-
communication skills required to present data insights in a iments to successful enterprise implementation.
compelling and interactive way and articulate the value of their
work to C-level executives, says Ramona Pierson, managing di- “I struggle to think of a function within my organization that is
rector and head of data innovation and products at PwC. not going to implement some sort of machine learning appli-
cation,” says Jason Rhodes, chief technology officer at financial
People with academic or work experience in math, science, and services firm Morningstar. “At the end of the day, it’s going to
engineering may also be a fit for machine learning roles. That be everywhere.”
includes individuals with a background in the social sciences,
who typically bring experience with quantitative analysis that Many businesses are giving domain experts such as marketers,
is applicable to machine learning. They may also possess a skill HR executives, and financial analysts training in how machine
that not all computer scientists have: an ability to look beyond learning works, so they can apply their firsthand knowledge of
algorithms and consider human factors in machine learning, processes, business problems, and enterprise data to collabo-
such as the impact of culture and bias on data selection, as well rations with developers. This helps ensure that machine learn-
as an ability to ask questions about how humans will interact ing models produce results that are effective in production.
with systems. “The more that we can bring people with business knowledge
in with the machine learning scientists, the better the products
Most people don’t think about hiring people with a social sci- we’ll be able to develop,” says Pierson.
ence background, says Pierson, “but having that understand-
ing of human cognitive behavior or social behavior can really LexisNexis Legal & Professional, for instance, is training its
inform how your machine learning models behave,” she adds. lawyers to become subject matter experts who work with data

“The more that we can bring people with business knowledge

in with the machine learning scientists, the better the products
we’ll be able to develop.”
RAMONA PIERSON, PWC

MIT SMR CONNECTIONS 3

 MANAGER’S GUIDE — TRANSFORM YOUR WORKFORCE WITH SKILLS FOR MACHINE LEARNING

scientists to train algorithms to summarize cases and identify Build out internal offerings. At Accenture, which has trained
legal precedents. more than 300,000 employees in new technologies over the
past three years, its training program offers flexible, personal-
In order for systems based on machine learning to be deployed ized options by blending classroom-based training and digital
successfully, it is critical that employees whose jobs involve learning. “In 2019, our people completed more than 25 million
interaction with such systems understand how models work activities through our nearly 4,000 learning boards — on-de-
and what their constraints and limitations are. Upskilling these mand educational modules across a wide range of topics,” says
individuals should revolve around teaching them how to prop- Eva Sage-Gavin, senior managing director of talent and orga-
erly interpret the results and recommendations of machine nization consulting at Accenture.
learning models, and how to intervene and communicate with
machine learning experts when those results seem off. Morningstar harnesses budding interest in machine learn-
ing by engaging employees in the AWS DeepRacer program,
4. Explore diverse options for training in machine learning skills. where they program and race autonomous 1:18 scale race cars
A robust training and development plan for upskilling an enter- that have machine learning capabilities onboard. Employees
prise workforce will need to address the wide range of needs receive hands-on training and experience the practical appli-
described above, with in-depth technical training for those cation of machine learning while participating in enterprise-
being groomed for machine learning development roles, but wide competitions.
more general instruction in data literacy for others. It also
makes sense to build a portfolio of options that addresses dif- Formalize mentorship and apprenticeship programs. This is
ferent learning modalities: Short online course modules may among the most popular approaches to upskilling employees.
work best for some, while a college-level course may be more Those who have the right baseline skills and aptitude are em-
appropriate for another. Here are some approaches: bedded with machine learning teams and assigned an expert
mentor or sponsor to help guide their progress.
Identify good online offerings. Online training is typical-
ly among the most accessible and flexible options. Training The Last Word: Culture
platforms like Udacity and Coursera offer advanced machine Upskilling is essential to preparing a workforce for machine
learning curricula and scholarship challenges. Providers that learning’s continued march into the mainstream. But devel-
partner with universities, such as EdX, offer numerous free oping a plan to identify employees for training and drawing
courses. Specialist online training companies are also emerg- up a menu of options is only the beginning. To succeed in
ing in the machine learning space to meet this growing need. the long term, leadership must foster a culture that supports
continuous learning and demonstrate its willingness to invest
Partner with a local college or university. LexisNexis, for exam- in employees’ development — a factor that has been shown to
ple, created a machine learning curriculum with North Carolina aid in retention.
State University. Employees work on actual use cases, using Lex-
isNexis’s own data to solve problems through machine learning. Supporting and executing a robust strategy to build up the
machine learning skills your business needs can do more than
Tap technology partners. Vendors that supply machine learn- provide an advantage in deploying this key emerging technol-
ing tools may have training offerings that can help get employ- ogy. It can also supplant a fear of technological change with
ees in tech roles up to speed more quickly. Their professional enthusiasm about the individual opportunities for growth that
services staff can also be a resource. can come with machine learning implementations. l

MIT SMR CONNECTIONS 4

 MANAGER’S GUIDE — TRANSFORM YOUR WORKFORCE WITH SKILLS FOR MACHINE LEARNING

MACHINE LEARNING
UPSKILLING CHECKLIST
Discover the best practices that businesses should bear in mind for successfully
launching and maintaining a training program and retaining skilled staff:

[3] Gain support from C-suite executives for upskilling programs in every area of the organization.

[3] Take a citizen-led approach: Make each of your employees responsible for identifying an appropriate
training path, and hold them accountable for acting on it.

[3] Provide tools for ongoing assessments of your workforce’s talents and skills. These should serve
two purposes: They should help employees determine their training and development priorities, and
they should provide business leaders with an inventory of employee capabilities, both active and
latent, across the business.

[3] Evaluate learning options to ensure that they are relevant to company needs, build on prior
knowledge, and are engaging.

[3] Look for learning options that can be accessed in easily digestible modules; these will be more
likely to fit around employees’ other responsibilities.

[3] Design incentives to support upskilling, both for individual learners and for those who support
them in mentor or guide roles.

[3] Find a technology partner that has relevant training offerings and can help bring current staff
members up to speed.

[3] Celebrate success by regularly recognizing those who complete training, and by calling
out examples of employees who have moved into new roles or taken on additional responsibilities
to contribute to machine learning efforts.

MIT SMR CONNECTIONS 5

 MANAGER’S GUIDE — TRANSFORM YOUR WORKFORCECWITH
U S T SKILLS
O M R FOR
E S EMACHINE
A R C H RLEARNING
EPORT

S P O N S O R ’S V I E W P O I N T
Voluptas nem sus, occat. Lam simo dolesto quae nis non pro Eque nectum etur seque di blaborro tenia aut occum hillignate

Training and
venihita con rerem ut quaeperum eum ventias et, consequide
et ma quunt lam, volorei untio. Commodio es delibus aut ex
AI Services Can
voluptur?
Ribusanis debis dolestore elic tem ipsaerum qui temolliquas
Help
eum quiatur sa desci aut magnam eum raeprat Close the mod
utassint volup- Skills Gapipsaepu ditam, volupitae porunt, ut faccus
eum undelicil
tio. Et voluVident. Ehenitatis mo omni ut magnis sitiist, siti odis aut et la estibus totaspera quatem susanimi, id magnati stiasit
cone doles pore laborum et la corit dolupta turiam etur, am aci tet ad maximen iscitat verorruntus ex ex est facea conse-
About Amazon Implementing AI and machine learning throughout your organization requires a collab-
recta dolores endenimusam,
Web Services tem que latestiinterdisciplinary
orative, simillupti simpo-approach
quatiamong
andae data
id esed quuntium
scientists, exeruptios
machine autem
learning ut volent pere
specialists,
rempore seditAWS inisoffers
quam,the sim raturia.developers, and domain experts. As you build
nobitature nonse your AI and
verum as machine
dipsamuslearning
non plit,strategy
explam saest et
broadest and deepest within the organization, it’s critical to evaluate the skills your organization currently has hil-
set of machine learning utatus iuscimil expe ra si voloreium ut hario experuntum
and build a plan to close any gaps.
Samandnatius
Al services. On behalfoccabor eiumquunto dolorectium
sa quiaerovit, ibus.
of our customers, we
archillare
issitatur?
focusedAliquos andipsam eaAtpor
on solving renduci
Amazon, AI delent, sunt learning
and machine Aquid et
haveanda cusam
been nulparupart
an integral ptaturi to volupti
of the businessonsequia
for overconem
eum dus some
nitaof the toughest
quiatur, sit pa aditae veles20 years. AI and machine
pere, ommodisquis autlearning
quam enable virtually every
re, omnissum ea espart of our business,
acieniam, including am
voluptas dolorporias
challenges that hold back
modi machine
delenest learning
hiligenimped real-time personalization and recommendations on Amazon.com; inventory forecast-
from quuntiis simporp oraestius maxi- volendae dolutem. Nam quia vitiur reperchil maximus moditat
being in the hands of ing; and natural language understanding with Alexa. Our goal is to put machine learning
mus quo estiani hiciis si is restrumet aut. empedis cienis apereperero ipsandus, sant am hit optatasima
every developer. Tens of technology in the hands of every organization, including those that are still working to
thousands of customers nihici velescit aliquam quam et volor modite sam voloriatist,
build the relevant skills in the workforce.
are
Subhead already using AWS for offic te dolorrore nes aborianis duntio. In porporem undipsa-
their machine learning
Git asimenis
efforts. Youes doluptam
can choose Pretrained
is nit, volorero AWS
voluptas autAIaut
services
lan- provide
peremready-made
qui volores intelligence for business applications
sit et apis ant.
from fully managed Al and workflows. Fully managed services such as Amazon Personalize, Forecast, Kendra,
dam, omni rerspid quam ipsande rchitae volor rem dis sit plat Arum hicius autatem fugitaque voluptatibus aut aut ad ute
services for computer and Fraud Detector reduce machine learning complexity so you can focus on business
es estotaq uiatium duntem faccus eum si doluptiis essedi im
vision, language, priorities such as personalizingconse cum experiences,
customer invellabores quaepre
modernizingex enis
yourquam, et, sersperun-
contact center,
recommendations,
fuga. Dit omniantios reri delessequodi quia consequi ipieture
reducing online fraud, and increasing customer engagement. These services don’t re- maxi-
tur a vel elibus ma sequam into tem et, nos maior simus
forecasting, fraud
lignata dolo consequo
detection, and search;et landiostio
or quire deep
illuptas machine
exceptat quialearning
con- expertise, and developers
met lab idenda can easilyliquam
quiae. Aximossum get started.
net fugit quamet aut
Amazon SageMaker to
sequi ipieture lignata dolo consequo et landiost aliquiat. voluptat lit eictae pre dolupti nos plitempore, to moluptatem
quickly build, train, and Boot camps, hackathons, and learning events can help accelerate the learning in your
Ibusdae
deploy nos suntiislearning
machine se nullaute occaerf erchicat velenem fu- incia num quam se aspe pa volorem aditiasim inciandes molec-
organization. Learning tools like DeepRacer, DeepLens, and DeepComposer put machine
models at scale.
giaturit et et od qui oditia dolores et learning veliqui res
in remporitat
the hands of inci tatus is reremperibus
every developer, es natem to
allowing nonexperts cus inisciae
learn ped qui
and build on ut odis et
their
SageMaker Studio offers
ulpa the
est,first fully integrated
apedips ametustem eos etur?Da own. Amazon’s Machine
nobitis possed qua-Learning University
aliquid offers
itatur reicil a variety ofendit,
eumeturitas free courses
cum simi,forquo
business
cor as mos
development environ- decision makers, developers, and machine learning experts. Encourage your teams to
met es moment beatefor etmachine
estem nonsequiant voleseque mint, optat- ex et, enes volupta turibus.
leverage these great learning resources and build their own customized curricula. Iden-
ur? Um, learning.
imusandisYou can also
ernamust abo. Lorion cus vellis doluptas
build custom models tifying a few key AI and machine learning champions within your team and establishing a
nullesciis
withunto et fugiatia
support dis issum eat.
for all of center of excellence will help build Elendes toruptatem
a virtuous talent et quo
cycle minumqu
through atatis porpori
continuous learning.tatust
Obistheapedipsa
popular open-source
delesto doluptatiur? Quis consendae volupta et volo ommolen imenim et audaepu diciis dolum idi corpo-
frameworks. Our
spictacapabilities
ne ium discidu Amazon’s Machine
eaquiLearning
dips- Embark program can alsotentiasperis
help accelerate your machine
are builtntorestem
on nest, tem quo remped eum, consedic veruntio. Lor alicimi nven-
the most comprehensive learning journey. Embark pairs your team with Amazon experts to prepare data, build
aperibus rempore dis ent, ut laut aut est, sitas doluptati re sint tecese nulparu ntiaspi duciam fugiaepudam re omnisqu aturiti
cloud platform, optimized and train models, and put models into production. It combines hands-on educational
dolupietforproreic
machine temlearning
alitem. Et porporem non conse
workshops corro eos
with brainstorming simusant
sessions ullab idist, tempost
and advisory utectem
professional ea desto
services eritatis
help yourerferum
with high-performance
solorumquae niendis deror mod unt.identify machine learning solutions aceriatonon porrunt,
business conet evellaute
challenges and then et omnit, simenda nissimus
go step-by-step
computing and
Onsecte no dolent. Poressi alibus
compromises on maionthrough the process
et facestius of development.
di to duci ut dolentur? Quibust, utem. Qui audipsam, vellam, ut eicimus sol-
pro etsecurity and
laut arum analytics.
quam, ulliqui nis iur? orum qui aut as accabor ectibus ius esti at eos eos eiusand itat-
Learn more at aws.ai. Finally, our extensive network of AWS Machine Learning Competency Partners offers a
Et aceati ut pro cum dolora volorio venimod ellenimet, conem. ur aniscil ibusdae reheni cum dolest, aliciis min et periatur?
range of services and technologies to help you create intelligent solutions for your busi-
Caerunt offic te exeribeat a dolupicness.
temquost,
Leadingvenditas dollaworkPedigenia
companies with thesenos ad quetoseque
partners volenim
innovate with aut moluptas
machine sam sedios
learning
on AWS. sed
del inum ipidendanda ea arum iliquamendae These partners
quia cuptamecan help you eturiorae
millest start realizing business
ventiis value
qui quae while
dent eumyou build
exces skills sse-
doloria
within your organization.
nditat magniat uritatem fugitia simpor solum re as doluptate Learn more at aws.ai.
quis aliqui voleconsequiata volum quiaeru ntiisci to et eossum
etur? omnist laboreh

MIT SMR CONNECTIONS 6