Getting Real with AI - Columbus DAW - May 2024 - Nick Woo from AlignAI

Nick Woo from AlignAI presented at the May 2024 Columbus Data and Analytics Wednesday meetup. The topic was a practical, business-oriented approach to identifying use cases for AI.

1. Getting Real with AI
What use cases are actually practical and actionable?
www.getalignai.com
1
Nick Woo
AlignAI

2. Operational Partner
Helping enterprises
get value from AI
quickly and safely.
The content of this presentation is confidential. Do not share without explicit permission of AlignAI.

3. AlignAI Hub
Intelligently manage
use cases, employee
workflows and the
global view of
adoption & controls
in one place for
AI Solutions.
The content of this presentation is confidential. Do not share without explicit permission of AlignAI.

4. AI Projects
AI Standards
and Policies
Value
Governance Intelligence Engine
Adoption
The content of this presentation is confidential. Do not share without explicit permission of AlignAI.

5. AlignAI

What exactly is GenAI?
Artificial
Intelligence
Machine
Learning
Deep
Learning
GenAI
The content of this presentation is confidential. Do not share without explicit permission of AlignAI.

6. AlignAI

How did we get here?
The content of this presentation is confidential. Do not share without explicit permission of AlignAI.

7. What AI/ML Solutions Do
The content of this presentation is confidential. Do not share without explicit permission of AlignAI.
7
Machine Learning is an approach to learn
complex patterns from existing data to make
predictions on new data.
Existing
Data
Machine
Learning
Model
New
Data
Predict

8. What is the process
to do the work?
Why do the work?
How do I execute?
Who does the work?
Tools
Roles
Frameworks
Concepts
AlignAI

Core Elements of How We Work
The content of this presentation is confidential. Do not share without explicit permission of AlignAI.

9. AlignAI

Trawling for Use Cases = Less Effective
The content of this presentation is confidential. Do not share without explicit permission of AlignAI.

10. AlignAI

Feasibility
Value
Capabilities
Use Cases
Balancing Value with Feasibility
The content of this presentation is confidential. Do not share without explicit permission of AlignAI.

11. Anatomy of an AI Use Case:

1.Business Problem
2.Data
3.Training
4.Model
5.Accuracy Metrics
6.UX/UI
The content of this presentation is confidential. Do not share without explicit permission of AlignAI.
11
AlignAI

12. AlignAI

We must start with objectives
What do you want? Why do you want it?
The content of this presentation is confidential. Do not share without explicit permission of AlignAI.

13. AlignAI

Aligning AI strategy to company strategy
Operational & Organizational
Integration

ROI Realization
AI & Data Feasibility

Piloting
Use Cases

Prioritization
Company Strategy &
Priorities

ROI Potential
AI Adoption
1
23
4
KPIs
Metrics
Opportunities
The content of this presentation is confidential. Do not share without explicit permission of AlignAI.

14. AlignAI

Key questions to ask about use cases
●Are you making these decisions already today? How?
●How repeatable are the scenarios in which you make those
decisions?
●Are there are very specific edge cases that make a big
impact on the decision making process?
●What is the risk profile of wrong outputs?
●How available and complete is necessary data?
The content of this presentation is confidential. Do not share without explicit permission of AlignAI.

15. The content of this presentation is confidential. Do not share without explicit permission of AlignAI.
15
Garbage
Data
Garbage
Results
Powerful Machine
Learning Models
AlignAI

16. The content of this presentation is confidential. Do not share without explicit permission of AlignAI.
AlignAI

We did not capture data for machines to learn, we captured it for us
The content of this presentation is confidential. Do not share without explicit permission of AlignAI.

17. Common AI Use Cases
17
EnterpriseConsumer

Automated decision
making for internal
stakeholders to level up
the organization’s
abilities.
Embedded in products
to improve customer
satisfaction and
product benefits.
E.g. Compliance email
classification
E.g. Netflix
recommendation engine
AlignAI

The content of this presentation is confidential. Do not share without explicit permission of AlignAI.

18. Common GenAI Use Cases
18
Financial Commentary
and Presentations

Generating recurring
financial reports,
automatically importing
data into templates,
automatically generating
insights from data.
Gathering Market
Intelligence

Leveraging public data to
create market insights,
generating competitive
intelligence and customer
insights for specific
criteria/ regions/ personas.
AlignAI

The content of this presentation is confidential. Do not share without explicit permission of AlignAI.

19. 19
Managing
Docs/Contracts

Generating documents &
contracts, focusing on
non-standard terms,
identifying
revenue-related clauses,
and quickly searching
for relevant information.
Producing Insights
from Data

Analyzing mass amounts
of data in sources like
Excel / CRM / ERP.
Leveraging for decisions
about pricing /
performance issues &
critical workflows.
Detecting
Anomalies/Fraud

Detecting errors,
spotting fraud by
analyzing trends &
anomalies, improving
prevention through
ongoing transaction
monitoring.
Common GenAI Use Cases
AlignAI

The content of this presentation is confidential. Do not share without explicit permission of AlignAI.

20. 20
3 Main GenAI Capabilities Today
1.Making online interactions conversational
(e.g., conversational journeys, customer service automation,
knowledge access)

1.Making complex data intuitively accessible
(e.g., enterprise search, product discovery and recommendation,
business process automation)

1.Generating content at the click of a button
(e.g., creative, document generation, developer efficiency)
AlignAI

The content of this presentation is confidential. Do not share without explicit permission of AlignAI.

21. The content of this presentation is confidential. Do not share without explicit permission of AlignAI.
21
In general, AI is better suited for…
✔ Recommending content
✔ Predicting future events
✔ Personalizing experiences
✔ Recognizing and classifying content
✔ Understanding natural language
✔ Detecting infrequent occurrences (outliers)
AlignAI

22. When to Use AI
The content of this presentation is confidential. Do not share without explicit permission of AlignAI.
22
Necessary? Cost Effective?
AlignAI

23. Additional Characteristics
The content of this presentation is confidential. Do not share without explicit permission of AlignAI.
23
It’s repetitive
The cost of wrong
predictions is cheap
It’s at scale
AlignAI

24. When to (not) Use AI
The content of this presentation is confidential. Do not share without explicit permission of AlignAI.
24
Unethical
Simpler solutions
do the trick
Not cost effective
AlignAI

25. AlignAI

Where can AI/GenAI go wrong?
How are humans
wrong today?
What is the cost of
being wrong?
How can humans
be manipulated?
Errors Impact Security
The content of this presentation is confidential. Do not share without explicit permission of AlignAI.

26. AlignAI

When it is wrong, it may cause potential…
Reputational harm
Physical harm
Ethical violations
Environmental damage
Financial risks (regulatory fines, customer churn)
Manipulation risk (threat actors)
Data privacy & sensitivity exposure
The content of this presentation is confidential. Do not share without explicit permission of AlignAI.

27. AlignAI

Stakeholders - Cross Functional Collaboration
HR
Legal &
Proc.
LOB & Ops
CISO &
Risk
EDO &
EAO
PMOAICOE/AI Task Force
Product
The content of this presentation is confidential. Do not share without explicit permission of AlignAI.

28. AlignAI

Prioritization Criteria of Use Cases
Criteria High Medium Low
Value
•Significant impact on business goals or
KPIs
•Addresses a critical pain point or
creates a new opportunity
•High ROI potential
•Moderate impact on business goals or
KPIs
•Addresses a relevant pain point or
creates a potential opportunity
•Moderate ROI potential
•Limited impact on business goals or
KPIs
•Addresses a minor pain point or creates
a small opportunity
•Low ROI potential
Feasibility
Complexity of Solution
•Clear path to implementation with
existing resources and expertise
•Minimal technical challenges
•Low risk of failure
•Implementation requires some
additional resources or expertise
•Some technical challenges can be
overcome with moderate effort
•Moderate risk of failure
•Implementation requires significant
resources or expertise
•Major technical challenges present
•High risk of failure
Data
•High-quality reliable and secure data
readily available
•Technical infrastructure can support
data processing and storage needs
•Data quality or availability needs
improvement
•Technical infrastructure may require
upgrades or adjustments
•Poor data quality or availability
•Technical infrastructure inadequate for
the project
Security
•Data complies with relevant regulations
and privacy standards
•Some data security or compliance
concerns exist
•Major data security or compliance risks
present
Technical Solution
•Open-source solution readily available
and cost-effective
•Requires less technical expertise and
resources
•Faster time to market
•Bought solution is cost effective, but
open-source is feasible
•Provides higher performance and
scalability potential
•Requires more technical expertise and
resources
•Bought solution is the only viable
solution due to specific needs (approval
required)
•Requires significant technical expertise
and resources
•Longest time to market
The content of this presentation is confidential. Do not share without explicit permission of AlignAI.

29. Use Case Exercise
The content of this presentation is confidential. Do not share without explicit permission of AlignAI.
29
Work Task/AI Solution: Identify work tasks that could be supported by an AI solution
Underlying Data:
Risk & Compliance:
UX/UI:
Outline goals & data that an AI solution would need access to
Design how you would interact with the AI solution in your flow of work
Define potential risks, errors, impact & security considerations
Potential AI Use Case

30. AlignAI

Important Takeaways
Opportunity to augment
decisioning and create
efficiencies
Provide clear objectives,
clearly define risks, and
monitor outcomes
High quality data and domain
context will elevate AI’s
outcomes
The content of this presentation is confidential. Do not share without explicit permission of AlignAI.