Onboarding AI & Machine Learning

Onboarding AI & Machine Learning
Brian Pichman
!

Evolve Project | Brian Pichman
2#CILDC
AI – Implications for Libraries !
What do you need to think about before bringing advanced
technology into your community, library or organization? How do you
introduce it to staff? Will they worry about being replaced or losing
their jobs? And how do you get machines to operate at optimal
efficiency? Machines need to learn to be effective, whether it’s Siri,
Alexa, or Watson. And people have to adapt to the machines. Join us
and learn more!
Today we are exploring…
Welcome

What is Artificial Intelligence
the theory and development of computer systems able to perform tasks that
normally require human intelligence, such as visual perception, speech
recognition, decision-making, and translation between languages.

What is Machine Learning
Machine Learning is a subset of Artificial Intelligence that uses algorithms and
statistical models to allow a computer system to make decisions around a
specific task without explicit instructions; relying on patterns and inference

5#CILDC
When talking about AI…
People will have one of two reactions:

6#CILDC
“The Agenda”
Topics for Today !
AI As A Tool
When introducing AI to the
staff/community, its best to
explain what it is…and what it
isn’t.
"
Impacts to Industries
How are we seeing AI Impact
different industries now?
#
Pulling it Together
How can libraries embrace AI
and move things forward?
$
Using AI
Many of us interact regularly
with some variant of Artificial
Intelligence.
!

7#CILDC
AI is becoming more available to the masses; being incorporated into our
smart phones, our connected homes, and simple robots we already use
today such as Roomba. With more accessible cloud-computing, open
source, and the making community, this field will rapidly expand.
More Widely Available
Using AI !
“The development of full artificial
intelligence could spell the end of the human
race….It would take off on its own, and re-
design itself at an ever increasing rate.
Humans, who are limited by slow biological
evolution, couldn't compete, and would be
superseded.”— Stephen Hawking

8#CILDC
!
What makes up an intelligent system?
AI Components "
#$%
Logic and Rules Based
Computer makes decisions
based on a decision tree, logic
rules, or a predefined process
with a calculated result.
Pattern Based
(Machine Learning)
Computer learns overtime by
using data and algorithms to
detect patterns.
Deep Learning
Deep Learning is a subset of
Machine Learning that
enables the computer to
make decisions on its own.
Neural Networks
A neural network allows an AI
to make its own conclusions,
where a simple pattern-only
based AI must rely solely on
data. A neural network
allows deep learning to
function.

Pattern Based Intelligence -> currently exists with self driving cars, language translations,
movie recommendations etc.
Strong Artificial Intelligence -> (doesn’t yet exist)
• computers think at a level that meets or passes people (abstract thinking)
Artificial Intelligence Exists

10#CILDC
Flash Light Examples
Understanding AI !
If an ML algorithm makes an
inaccurate prediction, then the
engineer needs to correct. In DL,
the algorithms can determine on
their own if a prediction is accurate
or not.
Deep Learning
Allow machines to make to their
own accurate decisions without
intervention from engineer
Neural Networks"
#
If detects {dark} turn on {light}
Logic Rules
$it’s performing a function with the
data given and gets progressively
better at that function
Machine Learning
Eventually, the system can turn
on the light with other queues
such as “I can’t see”
DL “Code”
Flashlight will turn on automatically
as it learns other words for “dark”
picking up on phrases that contains
the word
ML “Code”: %
♥
'

11#CILDC
create an algorithm that is able to teach itself without any external help
Pattern Recognition !
"
!
#
Deep Learning
Uses more complicated mathematical models to define
pictures content and speech
Self Learning
The advance machine learning
system makes decisions by
analyzing its own data and
making patterns
Learning on Examples
This method is used when a
machine learns through examples.
For instance, Google’s automatic
spam filtering learns as users
report spam.
Learning on Experience
The system learns from positive and negative experiences.

!
13
From Patterns to Automation
AI Models
The idea is that an algorithm will sift through the data, learn from it, and apply it to make a decision. This can be seen in any recommendation type
service. Machine Learning takes it a step farther by automating tasks; helping data security firms identify potential threats or finance looking for
favorable deals.
AI’s can be Transactional in which a question is asked and an answer is given, like a virtual assistant. AI’s can also be Automated in which routine
tasks such automatically taking trash out on garbage day.

14#CILDC
• When editing or using filters in photos (do X to eyes and Y to ears)
• Identification of license plates from an image in a toll violation
• Facebook’s ability to identify and recommend faces in photos
• iPhone users can have their phone categorize people by facial
patterns – in which you then define their name
• Google’s Image Recognition
Examples
How we see AI In Everyday Life
Image Recognition !
Think of how we can use facial imaging
to determine moods

16#CILDC
You probably see this everyday if you use Siri, Google Home, or an Echo
Product.
Overtime or with training, a system can tailored results based on
identifying the user asking. For example, Google Home will provide my
personal driving times to work if it hears me ask “how long will it take me
to get to work” versus a friend asking who it has no data on.
Examples
How we see AI In Everyday Life
Voice Recognition !
Think of how a system can respond
and remember a user based solely
on their voice

19#CILDC
Helpful – AI

20#CILDC
And How We Use It
Other Forms of AI !
Optical Character Recognition
Think of how a picture of your license plate allows
a machine to translate that to text and run a query
to determine who violated a toll.
Also see this in scanners that can take an image
and convert this to text.
Consider how you can take a photo of another
language and have it translate to yours
Advance User Preferences
This is the concept of an AI providing solutions
based on historic user’s preferences and
comparing it to similar users.
Compare how Amazon or Netflix makes
recommendations based on your purchases or
views – or even how Amazon guesses when you
might run out of a specific product.
Sensory Data Analysis
Your wearables that detect heart rate for instance
can determine without user intervention if you
are working out and even what kind of work out
such as jogging or bicycling.

Healthcare! Used in healthcare to identify and notice predictable
trends – such as having a machine look at charts to
recognize tumors sooner with more accuracy – or eyes to
determine stage of glaucoma

25#CILDC
Good Read:
https://www.businessinsider.com/healthcare-artificial-intelligence-pitfalls-2019-3

Smart Homes! See how a home can alert when it sees a person versus an
animal or know that its going to rain tomorrow so no
need to water the grass today

Service Industries! AI has the potential to replace many minimum wage jobs

https://www.bbc.com/news/av/technology-43292047/burger-flipping-robot-begins-first-shift

Autonomous Driving
Autonomous driving Level 4 describes vehicles that can operate without human interaction in most, but not all, conditions and locations and will likely operate in geofenced areas.
Autonomous driving Level 5 labels vehicles operating autonomously in all situations and conditions, and controlling all tasks.

33#CILDC
Inspiring AI’s !
AI: AlphaGo
AlphaGo is the first AI to beat a human
in arguably the most difficult game to
master. AlphaGo now teaches moves
to trainees.
" AI: ROSS
ROSS is an AI tool to make legal
research easier and faster
"

https://www.cbinsights.com/research-ai-100

36#CILDC
Run-away AI’s
Tay (Thinking About You) !
Released on March 23 2016 via Twitter, Tay (as TayTweets
on Twitter) was designed to mimic the interactions of 19 year
old girl through learned conversations on Twitter.
Users began tweeting pollitcally incorrect phrases to Tay, and
thus, Tay responded and answered with the learned
inappropriate behavior – as it was it was not taught what the
difference between Good Language and Bad Language was.
Microsoft Artificial Chatter Bot

37#CILDC
Run-away AI’s
Inspirobot.me !
I am an artificial intelligence dedicated to generating
unlimited amounts of unique inspirational quotes for endless
enrichment of pointless human existence.
-- From their website
Happy Accidents

38#CILDC
Logic and Rules Based
Challenges for AI
!" #
Training
Similar to having good data, an AI might
need to learn the correct response for
the correct situation or identify dangers
or inappropriate interactions
Precision
The idea of garbage data in
garbage data out. If you flood
an AI with bad data and don’t
set the proper syntax or
thresholds you will get
incoherent results
Context
AI’s can struggle with understanding context. For
example, asking Siri ”call me an ambulance” may yield “OK,
from now on, I will call you Ambulance”
$

39#CILDC
Things to Expand Your Knowledge
Cool Resources to Check Out !
IBM Watson
Watson was created as a question answering (QA) computing system that
IBM built to apply advanced natural language processing, information
retrieval, knowledge representation, automated reasoning, and machine
learning technologies to the field of open domain question answering. –
Wikipedia
Powered by the latest innovations in machine learning, Watson lets you learn more with
less data. You can integrate AI into your most important business processes, informed
by IBM’s rich industry expertise. You can build models from scratch, or leverage our
APIs and pre-trained business solutions. No matter how you use Watson, your data and
insights belong to you − and only you.
--IBM Watson

By Pgr94 - Own work based on diagram found at
http://www.aaai.org/Magazine/Watson/watson.php, CC0,
https://commons.wikimedia.org/w/index.php?curid=14575947

41#CILDC
Kaggle
Kaggle is an online community of data scientists and machine learners, owned
by Google, Inc. Kaggle allows users to find and publish data sets, explore and
build models in a web-based data-science environment, work with other data
scientists and machine learning engineers, and enter competitions to solve
data science challenges. Kaggle got its start by offering machine learning
competitions and now also offers a public data platform, a cloud-based
workbench for data science, and short form AI education. -- Wikipedia

43#CILDC
TensorFlow
TensorFlow is an open-source software library for dataflow programming across
a range of tasks. It is a symbolic math library, and is also used for machine
learning applica=ons such as neural networks. It is used for both research and
produc=on at Google. TensorFlow was developed by the Google Brain team for
internal Google use. It was released under the Apache 2.0 open-source license
on November 9, 2015. -- Wikipedia

Applying It In Your Library
Thoughts, Ideas, and More
!

How can you prepare people for these fields?
45
Skills For Learning !
Understanding data and how to
read data sets is valuable
https://dzone.com/articles/ten-machine-
learning-algorithms-you-should-know-to
Math and Algorithms
Statistics
Learning how inputs of code can
interact physical parts
Hardware + Software
Robotics
Learning to code at a basic level
with syntax and flow; then move
to Python (most common)
https://www.geeksforgeeks.org/top-5-best-
programming-languages-for-artificial-
intelligence-field/
Coding Languages
Coding
Learning this is a huge skill to
master, along with object
recognition
https://www.pyimagesearch.com/start-
here-learn-computer-vision-opencv/
How Do Computers See
Computer Vision

How can we onboard AI today?
46
Automations and Community !
Receptionist
Allow for an interaction that’s
quick and frees up time for staff
for more complex and human
needed interactions
“Where’s The Bathroom”
Industry Risks
If car automation takes hold,
what does that do to the shipping
and delivery industry?
Preparing The Future
People sometimes need help
finding information on your
website. A Chat bot that notices a
user on a page for a long time can
make recommendations or hand
off to staff
Online Support
Chat Bots Futures
What can we make with the
technology to make the world a
better place?
What Can We Make

48#CILDC
“The Agenda”
Bringing It All Together !
Impact On Staff
At some point, AI can very well
replace some functions and
roles of what we do. IE:
Receptionist/Simple Q & A
"
Improving AI
If building an AI solution, you
will want to be accurate with
the data you provide
!

49#CILDC
Making Your Job Better
Shift of Focus !
AI can replace many of our tasks that we do to make them:
* More Efficient =
* More Accurate = As long as the data provided to the AI
is “good”, then the function will be more accurate.
*Tedious Work may be “outsourced” to the AI
All in all, with a positive shift in focus, it can allow staff more
time to focus on other projects, more resources can be
available to the library.
Making The Work Day More Meaningful

50#CILDC
AI Struggles
Shift of Focus !
AI may not have emotional intelligence or show
empathy…(yet).
AI makes a decision based on facts and may not
account for emotional variables.
To assume AI can’t replace a job role because it
doesn’t have emotional intelligence would be a
failure point to any organization.
Empathy / Emotional Intelligence

51#CILDC
Maslow

52#CILDC
Feel free to reach out!
Questions / Contacts !
815-534-0403
"
www.evolveproject.org
bpichman@evolveproject.org
#
Twitter: @bpichman
linkedin.com/in/bpichman
slideshare.net/bpichman
$
CHECK OUT MY OTHER SESSION:
E205 l Trending @ School Libraries
4:30 p.m. – 5:15 p.m.
Our experienced and popular “Games
& Gadgets” guy explores the tools,
tech, gadgets and robotics that are
trending for school libraries.

Onboarding AI & Machine Learning

More Related Content

What's hot

Similar to Onboarding AI & Machine Learning

More from Brian Pichman

Recently uploaded

Onboarding AI & Machine Learning