34
Copyright © 2021, Cognilytica [email protected] www.cognilytica.com @Cognilytica Foundations of AI & ML for Executives & Decision-Makers Presented by Cognilytica Analysts: Kathleen Walch Ronald Schmelzer Laying the foundation for AI project success Cartoon illustrations by Timo Elliott (timoelliott.com)

Executives & Decision-Makers Foundations of AI & ML for

  • Upload
    others

  • View
    3

  • Download
    0

Embed Size (px)

Citation preview

Copyright © 2021, Cognilytica [email protected] @Cognilytica

Foundations of AI & ML for Executives & Decision-Makers

Presented by Cognilytica Analysts:

Kathleen WalchRonald Schmelzer

Laying the foundation for AI project success

Cartoon illustrations by Timo Elliott (timoelliott.com)

Copyright © 2021, Cognilytica [email protected] @Cognilytica

Key Takeaways/Learning Objectives

➢ AI terminology & AI Lexicon

➢ Problem areas where AI is best suited in the organization

➢ Seven Patterns of AI

➢ AI Use Cases/Applications of AI

➢ Applying Big Data know-how to AI

➢ How to approach AI projects

➢ Optimizing time and effort for AI

2

Copyright © 2021, Cognilytica [email protected] @Cognilytica

About Cognilytica● Cognilytica is an AI-focused analyst, advisory, and education firm

● Market research, advisory & guidance on AI, ML, & Cognitive Technology

● Provides role-specific education on AI, ML, and emerging technology

● Focused on enterprise and public sector adoption of AI, ML, and Cognitive Technology

● Kathleen Walch and Ron Schmelzer are Principal Analysts and Managing Partners of Cognilytica

● Produce the popular AI Today podcast and AI communities

● Contributing writers to Forbes and TechTarget (SearchEnterpriseAI)

3

Copyright © 2021, Cognilytica [email protected] @Cognilytica

The Dream of the Intelligent Machine

4

Copyright © 2021, Cognilytica [email protected] @Cognilytica

What can we do with an intelligent machine?● Achieving the greater levels of productivity● Improving accuracy● Improving reliability● Eliminating drudgery and dull tasks● Improving relevance● Achieving greater scale● Enabling continuous monitoring and auditing● Extracting more value from data● Providing assistance to those in need

The quest to understand intelligence

5

Copyright © 2021, Cognilytica [email protected] @Cognilytica

But what is Human Intelligence?!Generally, what we consider as what makes humans intelligent:● Perceive and Understand the Outside Environment● Learn from Experience● Adapt to New Situations

○ Use Reasoning to Solve New Problems○ Apply learning to different contexts

● Understand and Handle Abstract Concepts (Including “Emotions”)● Use that Knowledge to Manipulate One’s Environment● “Goal-directed Future”

○ In this light, you can be alive, but not intelligent○ Plants and bacteria vs. higher-order animals○ Higher-order animals vs. Humans

● Express Creativity and Imagination● Respond Quickly and Successfully Deal with Complex Situations● Discern what is Important from what is Not Important● Handle Ambiguous Situations and Incomplete Information

6

Copyright © 2021, Cognilytica [email protected] @Cognilytica

Framing a Common Understanding of AI

● Cognilytica’s definition:

Artificial Intelligence (AI) is machine behavior and function that

exhibits the intelligence and behavior of humans.

7

Copyright © 2021, Cognilytica [email protected] @Cognilytica

Narrow (Weak) AI vs. General (Strong) AI

Narrow AI★ Just using some of the capabilities of AI for a narrow function★ Avoid the term “weak AI”

General / “Strong” AIAchieving the vision of a generally

intelligent entity

★ Capable of adapting to any situation★ Able to respond as a human would ★ Artificial General Intelligence (AGI)★ Avoid the term “Strong AI”

8

Copyright © 2021, Cognilytica [email protected] @Cognilytica

Busting AI Myths1. AI is all about Superintelligent Machines2. AI is all about building Robots3. AI is all about building Autonomous Vehicles4. We don’t need AI, we can just use analytics and statistics5. This is just math - nothing special6. Automation and Intelligence are the Same Thing7. I can’t do AI without a Google-amount of Data8. AI, Machine Learning & Deep Learning are all the same thing9. I thought neural networks were dead

10. I can buy AI from a vendor

9

Copyright © 2021, Cognilytica [email protected] @Cognilytica

Why AI Now?

10

● Experience and infrastructure to deal with Big Data, that fuels AI● Rapid evolution of Machine Learning algorithms, Deep Learning in

particular ● Almost limitless, cost-efficient Compute Power● Mega-Tech Companies investing heavily in AI● Country-level strategic investment in AI

Copyright © 2021, Cognilytica [email protected] @Cognilytica

AI in the Context of Digital TransformationDigital Transformation is about changing the way organizations and societies do things by moving away from old, non-digital ways of doing things to new, digital ways.

But we’re hitting a Digital Transformation Log Jam…… what’s needed are intelligent systems

11

Copyright © 2021, Cognilytica [email protected] @Cognilytica

Who cares? Why does this matter?

We’re in the Always-On, On-Demand, Instant Access Era

● People are demanding greater quality of service● People are demanding greater reliability● People are demanding 24/7, instant access● People are demanding more relevance● People are demanding less friction with companies, governments,

electronic systems, devices, apps

People want companies, products, systems, and organizations to be more intelligent

12

Copyright © 2021, Cognilytica [email protected] @Cognilytica

What are AI & Cognitive Technologies Best Suited For? ● Classification & Identification

○ Object identification and clustering● Conversational Interfaces

○ Voice and text chatbots and interfaces● Big Data Predictive Analytics

○ Using trained data to make best-guesses● Pattern Discovery

○ Finding the hidden patterns in big data, when a simple algorithm doesn’t work● Autonomous Systems

○ Self-controlling systems without human intervention● Game & Scenario Playing (“Rules discovery”)

○ Letting systems play themselves and figure out the hidden rules● Hyper-Personalization & Recommendation

○ Summing up all the little pieces to make a bigger whole

If it’s probabilistic, use a learning model13

Copyright © 2021, Cognilytica [email protected] @Cognilytica

What are AI & Cognitive Technologies not Suited for?● Repetitive, Deterministic Automation Tasks

○ Just code it or record it and be done with it● Formulaic Analytics

○ This is what big data Business Intelligence (BI) platforms are for● Systems that require 100% accuracy

○ If it’s trained, then it can’t be always right● Situations with very little training data

○ If you can’t train it, then machine learning won’t work● Situations where hiring a person is easier, cheaper, and faster

○ Sometimes the human brain just wins out● A need to do “AI” without understanding what it is or what it’s for

○ There’s lots of value in AI & Cognitive Tech without being vague

If it’s deterministic, use a programming approach

14

Copyright © 2021, Cognilytica [email protected] @Cognilytica

The Seven Patterns of AI

15

Copyright © 2021, Cognilytica [email protected] @Cognilytica

Consumer Packaged Goods Applications● Understand customers through social media

○ CPG businesses can mine social media posts with AI to see what people are saying about their brand

● Optimize trade promotion○ NLU helps match trade promotion contracts to invoices, reducing errors and

overpayment● Personalize the customer journey

○ Digital shoppers can help personalize and simplify the experience resulting in increased customer loyalty and retention

● Real-time design variations○ AI can develop real-time website design variations for customers creating

personalized experiences to drive sales

16

https://www.cognilytica.com/2019/10/24/infographic-ai-in-consumer-packaged-goods-2/

Copyright © 2021, Cognilytica [email protected] @Cognilytica

Supply Chain Applications● Automate and streamline various processes● Improve shipping time and order fulfillment● Improve warehouse management● Improve supply chain visibility● Assist supply chain workers● Assist the sales team● Speed up decision making● Image recognition for inventory management

17

https://www.cognilytica.com/2019/01/24/infographic-ai-in-supply-chain/

Copyright © 2021, Cognilytica [email protected] @Cognilytica

Cybersecurity Applications

● Malware Identification● Proactive Response● Adapt to changing threats● Categorize attacks● Threat detection● Autonomous patch development● Better spam and phishing detection

18

Copyright © 2021, Cognilytica [email protected] @Cognilytica

Augmented Intelligence: Combining the Best of Both Worlds

HumansGreat At:

★ Intuition★ Emotional IQ★ Common Sense★ Creativity

Poor At:★ Probabilistic Thinking★ Dealing with large volumes

of info★ Bias

AIGreat At:

★ Probabilistic Thinking★ Dealing with large volumes of

info★ Being Trained

Poor At:★ Intuition★ Emotional IQ★ Common Sense★ Creativity★ Bias

+ =AugmentedIntelligence

19

Copyright © 2021, Cognilytica [email protected] @Cognilytica

Time to Positive ROI for AI Projects

20

Shortest time to ROI Longest time to ROI

Augmented Intelligence

Conversational SystemsAutonomous Systems

Copyright © 2021, Cognilytica [email protected] @Cognilytica

The Core Aspects of Intelligence

Perception

★ Sensing★ Understanding★ Processing lots of inputs

Prediction

★ Understanding possibilities★ Thinking Ahead

Planning

★ Adapting to new environments

★ Determining outcomes★ Self-learning

Feedback loop

21

Copyright © 2021, Cognilytica [email protected] @Cognilytica

Machine Learning: Cornerstone of AI● Learning is required to perceive, predict, and plan● Programming is not learning● Process flows are not learning● Rules are not learning● In each of the above it’s the human intelligence that makes

those things work● Imagine if the machine needs to come up with its own rules,

flow, logic, and processes. This is intelligence.● Intelligence is all about improving with experience● Intelligence is all about adapting and change.

How does the Human Brain Learn?We don’t know. So let’s try different approaches

22

Copyright © 2021, Cognilytica [email protected] @Cognilytica

The Three Kinds of Machine Learning● Learning is required to perceive, predict, and plan● Supervised Learning

○ Human helps feedback outputs back to inputs● Unsupervised Learning

○ Machine figures out correlations of data on its own● Reinforcement Learning

○ Use experiences, reward, and goals to iterate to the right outcome● Lots of ways of doing Machine Learning

Machine Learning is a Necessary Part of Artificial Intelligence, but it’s just a part of AI

23

Copyright © 2021, Cognilytica [email protected] @Cognilytica

Deep Learning: A Revolution in Machine Learning

Deep Learning is one form of Machine Learning, and only a piece of AI.

Will there be another AI algorithm revolution?

24

Copyright © 2021, Cognilytica [email protected] @Cognilytica

Relating AI, Machine Learning and Deep Learning

25

Copyright © 2021, Cognilytica [email protected] @Cognilytica

AI is Not About Application Development● After all these slides, does it take much convincing to say:

AI is not about application development! It’s about data

● Where is the code? ○ Implementing the AI algorithms○ Moving data around○ Building stuff “around” the model to feed it what it needs and process the results

● The code is a small part of making AI work. And not even the most important part

So, if you plan to run your AI Projects like you run your Application Development Projects you’re going to find out the hard way that it won’t

work.

26

Copyright © 2021, Cognilytica [email protected] @Cognilytica

What is Data Science?● Data Science is focused on

○ Extracting useful information from a sea of data○ Translating business and scientific informational needs into the

language of information and math● Data scientists need to be masters of statistics, probability, mathematics,

and algorithms that help to glean useful insights from huge piles of information.

● Programming is data-centric, not function-centric● Data Scientists often report into the Line of Business● Data Science is key to Machine Learning - determining which algorithms to

use and how to train models and sources of data

Data scientists can’t perform their jobs without access to large volumes of clean data, and need engineers to put Machine Learning models into operation

27

Copyright © 2021, Cognilytica [email protected] @Cognilytica

Big Data is Part of What is Powering this AI Wave● Need large quantities of data on which to train AI

systems

● Without data, you have nothing in AI

● We also now have the means to get more insight from all that data

● We have algorithms ready to act with data and the compute power available to chew on it

● We also have Experience, tools, and ability to deal with massive volumes of data

● All the pieces are in place...

Doing AI well means doing Big Data well which means doing Data Science well

28

Copyright © 2021, Cognilytica [email protected] @Cognilytica

80% of AI Projects are Data Engineering

29

Copyright © 2021, Cognilytica [email protected] @Cognilytica

Data Engineering● Data Engineers are focused on providing clean, efficient access

to data and operationalizing Machine Learning in practice● Data Engineers grew up in a Big Data world● Data Engineers are not focused on application development or

product development● Data Engineers often report into the IT organization

30

Copyright © 2021, Cognilytica [email protected] @Cognilytica

Project Management Approaches to AI are Challenged● “Waterfall” approaches definitely don’t work

○ Rigid and restrictive○ Too lengthy○ High risk of building the wrong model. ○ Requires teams to predict major obstacles○ Unable to quickly respond to changing technology,

requirements, needs, bias problems ● Agile approaches need some modifications

○ Continuously learn and iterate○ Opportunities to re-prioritize where necessary○ Deliver value faster and more efficiently○ Challenge: Simultaneous continuous model tweaking and

deployment

Update Agile Approaches with Data Science / Data Management Methodologies

31

Copyright © 2021, Cognilytica [email protected] @Cognilytica

CPMAI: Combining the Best for AI Project Success

● The data-centricity of CRISP-DM

● Specific details for AI / cognitive technology projects

● Leverages the Seven Patterns to provide guidance

● All steps are iterative with each other

32

Copyright © 2021, Cognilytica [email protected] @Cognilytica

Strategic Support for Strategic AI

33

Cognilytica AI Education Subscription

➔ Annual subscription to continuously updated role-specific AI education

➔ Self-paced and supported virtual courses on AI topics:◆ Fundamentals of AI◆ Thinking Like a Data Scientist◆ Emerging Trends in AI◆ Risks, Threats, and Challenges of AI◆ Diving Deeper into AI: Algorithms & Methods

➔ Gain understanding of key concepts in AI, machine learning, cognitive technologies, Robotic Process Automation (RPA), and Big Data

➔ Certifications and exercises included

➔ $3,000 per learner per year (12 months) for all-access

Copyright © 2021, Cognilytica [email protected] @Cognilytica

Thank YouPresenters:

Kathleen Walch & Ronald Schmelzer

Cognilytica - http://www.cognilytica.com

[email protected]

34