Comprehensive deck overviewing artificial intelligence and machine learning and how companies can integrate these technologies. Created by Kadeem Clarke, founder of Clarke Global and former VC investor
2. Agenda
22
I. What is Artificial Intelligence?
II. What is Machine Learning?
III. Adopting Machine Learning and
Artificial Intelligence Technology
IV. Risks of Artificial Intelligence
V. Costs and Benefits of AI
Overview: AI will change the future
business climate, see how it can
bring businesses to the future.
3. What is Artificial Technology?
“The theory and development of computer systems
able to perform tasks normally requiring human
intelligence, such as visual perception, speech
recognition, decision-making, and translation
between languages.”
(Source: Forbes)
3
4. Debunking Artificial Intelligence 4
Top 5myths about artificial intelligence:
AI is going to
replace all
jobs
Only
low-skilled and
manual workers
will be
replaced by AI
Computers
will become
better than
humans at all
tasks
Artificial
intelligence will
quickly exceed
human
intelligence
AI will lead to
the
destruction of
human race
by robots
1 2 3 4 5
5. Artificial Intelligence is NOT Intelligence 55
STEP 1
Define the
problem
Generate the
objective
assessment
criteria
Choose the best
solution
Evaluate the
success of the
chosen
alternative
Implement the
chosen decision
STEP 2 STEP 3 STEP 4 STEP 5 STEP 6
AI assisted
Generate all
possible solutions
AI does not bring us intelligence, it brings prediction
... prediction is essential to the decision making process
6. Prediction is Used to Fill in Missing Information 66
Information you
have (called
”data”)
Data analysis
completed by AI
tools
+
Human
judgement and
decision making
+
Unknown information
turned into predicted
information
→
7. Data is Essential to Quality Predictions 77
Machine learning utilizes three types of data
Training data for
training the AI
Input data for
predicting
Feedback data for
improving the prediction
accuracy
8. What is Machine Learning? 8
“The concept that a computer program can learn
and adapt to new data without human
interference. Machine learning is a field of
artificial intelligence (AI) that keeps a computer’s
built-in algorithms current regardless of changes
in the worldwide economy.”
(Source: Investopedia)
9. ML Is The Most Powerful AI Technology 9
Benefits of Machine Learning
Systems using
this technique
learn and
improve over
time
Enhanced
prediction enables
ML machines to
perform tasks that
were originally
exclusive to
humans
Value of
human
judgment will
increase as
predictions
are better,
faster, and
cheaper
ML can
accommodate
complex
models with
many more
interactions
between
variables
Less need to
specify in
advance what
goes into the
model
1 2 3 4 5
10. Automated decision making will be first
used in applications where…
● Other elements are already automated
except for prediction (e.g. mining)
● Benefits of prediction are high in
relation to speed of action and reduced
waiting time
3
2
1
Machine Learning is an Ongoing Process 1010
● ML will be used in many
business functions, such
as inventory and demand
forecasting
● When the tools become
powerful enough, they
may motivate changing
the strategy itself
Current Strategy
Decision Making
Implementation
● Some companies can simply drop an AI tool into
their work flow and realize immediate benefit
● For others, deriving a real benefit from
implementing an AI tool requires rethinking the
entire work flow
11. C-Suite Leadership Must Not Delegate ML
Strategy
1111
The trade-off is influenced
by uncertainty (e.g. higher
sales from ship-then-shop
are outweighed by higher
costs from returned items)
An ML tool that reduces
uncertainty tips the scales
of the trade-off so that the
optimal strategy changes
from one side of the trade
to the other
There is a core trade-off
in the business model
(e.g. shop-then-ship
versus ship-then-shop)
1 2 3
ML can lead to strategic change if three factors are present
12. The AI Canvas Can Streamline Decision Making 1212
What do you
need to know to
make the
decision?
Use it to think through how AI could help with business decisions
PREDICTION JUDGEMENT ACTION OUTCOME
How do you
value different
outcomes and
errors?
What are you
trying to do?
What are your
metrics for task
success?
INPUT TRAINING FEEDBACK
What data do
you need to run
the predictive
algorithm?
What data do
you need to train
the predictive
algorithm?
How can you use
the outcomes to
improve the
algorithm?
Source: Harvard Business Review
13. The AI Canvas Can Be Used For Any Business 1313
Predict whether
an alarm is caused
by an unknown
person vs.
something else
(i.e., true vs. false)
An Example Using AI to Improve Home Security
PREDICTION JUDGEMENT ACTION OUTCOME
Compare the cost
of responding to a
false alarm to the
cost of not
responding to a
true alarm
Dispatch a security
response or not
when an alarm is
triggered
Observe whether
the action taken
in response to the
triggered alarm
was correct.
INPUT TRAINING FEEDBACK
Sensor inputs from
movement, heat, camera, and
contextual data at each point
in time when the alarm is on;
these data are used to operate
the AI.
Historical sensor data
matched with historical
outcome data (actual
intruder vs. false alarm);
these data are used to train
the AI before it is deployed.
Sensor data matched with data
collected from outcomes (verified
intruders vs. verified false alarms); these
data are used to update the model,
continuously improving the AI while it
is operating.
Source: Harvard Business Review
14. Implementation of ML Tools Will Impact Jobs 1414
Implementation of ML tools generate four implications for jobs
ML tools may lead to
reconstitution of jobs
with some tasks added
and others taken away
ML tools may contract
jobs, as in fulfillment
centers
ML tools may augment
jobs, what spreadsheets
did to bookkeepers
ML tools may shift the
emphasis on the specific
skills required for a
particular job
15. ML Carries Many Types of Risk 1515
Six most relevant risks to ML
Predictions from
AIs can lead to
discrimination
Incorrect input
data can fool the
machines
AIs are ineffective
when data is sparse
Trade-off
individual- and
system-level
outcomes
Feedback can be
manipulated to fool
the machine
Prediction
machines can lead
to IP theft
16. AI Can Be Used to Save Banks Money 1616
Middle Office
$217 billion of this will
be attributed to
compliance, KYC/AML,
and other forms
$350
billion
savings
Back Office
$31 billion of this will
be attributed to
underwriting and
collection systems
$200
billion
savings
Front Office
$199 billion of this will
come from reductions
in the scale of branch
networks
$490
billion
savings
Artificial Intelligence and the Banking Industry’s $1 Trillion Opportunity
Source: The Financial Brand