Disruptive innovations, disruptive technologies, industry 4.0, artificial intelligence
1. Redefining the Role of Educator in Covid-19 Outbreak Era
e-Faculty Development Program
Gujarat Technological University,
Ahmedabad
&
Anand Pharmacy College,
Anand
by
on
May 11-16, 2020
2. Disruptive Innovations & Industry 4.0
A world driven/empowered by
Artificial Intelligence
Dr. Amit Gangwal
(M. Pharm., Ph. D.)
Professor & Principal
Shri Vithal Education & Research Institute's College of
Pharmacy, Pandharpur
May 11, 2020:
12:00PM to 01:00Pm)
3. I salute you all teachers
Because everyday you are dealing with 60
families in a lecture of 60 minutes; each time
you have to fine tune your overall connection
and bond with your students. This a rare
phenomenon across any of the industry.
1Dr. Amit Gangwal
11. 3D Organ Printing
(Three-dimensional printing of biological matters: Journal of Science: Advanced Materials and Devices
Volume 1, Issue 1, March 2016, Pages 1-17)
9Dr. Amit Gangwal
12. Hand-held devices & wearable devices
Apple sold nearly 10 million more watches than the entire Swiss
watch industry in 2019 - and it's a sign the wrist wear market has
entirely changed.
10Dr. Amit Gangwal
13. It is an innovative product for police as well, to
nab the violators; among many other
applications. 11Dr. Amit Gangwal
16. ⢠Any innovation may be called a disruptive innovation
if it
â Creates new product (s)
â Creates new service
â Shows new way to do business (generating sale, profit,
customers, making products or services or something
which literally affects or wipes out existing
businesses/players/products/services/concepts/likings/preco
nceived notions or other crucial component in entire
business or ecosystem for that matter)
Can you think some examples lately?
14Dr. Amit Gangwal
17. According to HBR
1. Disruptive innovation is a process by which a smaller company,
one with fewer resources, manages to overtake or successfully
compete with incumbent enterprises.
2. Smart disrupters improve their products and drive upmarket.
3. Some disruptive innovations succeed and some donât.
4. Disrupters often build business models that are very different from
those of incumbents.
5. Disruptive innovations donât catch on with mainstream customers
until quality catches up to their standards.
6. Disruptive innovations originate in low-end or new-market
footholds.
15Dr. Amit Gangwal
18. According to HBR continue----------
Disruptive innovations are made possible because they get started
in two types of markets that incumbents (existing player)
overlook:
â New-market footholds: Disrupters create a market where none
existed; they find a way to turn non consumers into consumers (e.
g. Netflix)
â Low-end footholds exist because incumbents typically try to
provide their most profitable and demanding customers with ever-
improving products and services, and they pay less attention to
less-demanding customers (e. g. Airbnb)
16Dr. Amit Gangwal
19. Artificial Intelligence. Blockchain.
Coding/Algorithms. Data Analytics
Internet of Things
Robotics
Hand-held devices for MRI, CT Scan
& what not
Much faster & well in advance
diagnosis of diseases
Driverless cars
Drones
Doctor-less clinics/Diagnostic centers
Banker-less Banks
Highly customized drug therapies
Faster drug discovery & development
Teacher-less institutes & many more.
ABCD has brought paradigm shift by its
very nature of disruption
17Dr. Amit Gangwal
20. Disruptive innovations How they affected existing or old technology/process/product
Personal computer Have almost replaced type writers.
Cell phones A replacement for fax, telegram and landline phones. They have also threatened industries
manufacturing personal diaries, alarms, MP3 players, calculators, cameras etc.
ATM Almost entirely ended the concept of visiting bank for various financial purposes.
Email Transformed the way we communicating, largely displacing letter-writing and disrupting the
postal and greeting card and related segments.
Laptop Replaced desktops. In very near future foldable phones may replace laptops as well;
indicating disruption is a continuous process.
Wearable devices,
portable devices
and other
techniques
ECG, step trackers and lot many health and fitness recording is easily available in wrist
watches, cell phones. Handheld devices are going to rule the healthcare domain, especially
diagnostic paradigms. Portable-foldable devices will be used by patients or their caretakers
for performing endoscopy, sonography, X ray, MRI or CT scan for that matter. Without any
incisions or piercing blood test for various pathogens or biomolecules will be possible in
next 10 to 15 years at highly affordable rates for masses.
Online/e pharmacy No need to visit nearby pharmacy, no need to set pills reminders/
Painless shots for
diabetics and
others
Painless insulin shots and other similar technologies are being worked upon by technology
giants and diagnostics organizations. Depending on the affordability it will decide the future
of traditional injections. 18Dr. Amit Gangwal
21. Sustaining innovations
(incremental improvement)
Disruptive innovations
Existing market New market.
Improves performance.
Cost is lower & incremental
change.
Innovation is dramatic & game
changing.
Market is predictable and
traditional business methods
are sufficient.
Market is unpredictable and
traditional business methods fail.
19Dr. Amit Gangwal
22. Disruptive innovations come with certain disadvantages
also, which are not visible in near future (post their advent in the
market or society). Few examples can be stated here.
Huge discussion going on, about increasing number of cancer victims
owing to mobile radiations etc., irreversible damage to environment
owing to use of single use plastic.
Years ago when plastic and its uses were propagated, at that time it
was not thought that such a wide spread use of plastic may damage
environment to a large extent.
It means certainly AI, ML, IoT may also have some any ill effects in
future. It requires visionary approach, ant not the myopic calculations
to gauze and check the mal effects of these innovations in future.
20Dr. Amit Gangwal
23. Current factory reset by the God has cleaned all the
rivers and air pollution like never before.
21Dr. Amit Gangwal
25. Revolution Meaning/Progress
Industry 1.0
(between
the late 1700s and early
1800s)
In this era manual labor performed by people and assisted by
animals and use of water and steam-powered engines were main
tools available for various purposes.
Industry 2.0 (In the early
part of the 20th century)
Electricity enabled manufacturing (concepts like the assembly
line to increase efficiency) was the main tool. This led to
enhanced productivity.
Industry 3.0 (Starting in the
late 1950s)
In this third industrial revolution electronic & computer
technology were at the forefront. Shifting was from analog and
mechanical technology to digital technology and automation
software.
Industry 4.0 (current era)
This is the era of disruptive innovations or disruptive
technologies. This is based on high speed internet connectivity
(5G), cloud computing, big data analytics, internet of things,
artificial intelligence, machine learning, 3D organ printing,
blockchain and related technological advancement like virtual
reality and augmented reality. Such is the application of these
innovations that none of the industries and none of the
department have untouched or immune to these disruptions. 23Dr. Amit Gangwal
26. Why so much disruption with a
speed of cheetah ?
24Dr. Amit Gangwal
27. Driving force behind disruptive innovations
(AI, ML, IoT, Cloud, High speed quantum systems and fast
internet connectivity)
ABCDEFG
25Dr. Amit Gangwal
28. AI vs ML vs DL
AI
Program with the ability to
learn & reason like humans
ML
Algorithms whose output
improves as they are exposed to
more and more data over the
time
DL
A subset of ML in which
multilayer neural
networks learn from vast
amount of data
26Dr. Amit Gangwal
29. What experts have to say about disruptive
technology/industry 4.0/Artificial Intelligence
âSuccess in creating AI would be the biggest
event in human history. Unfortunately, it might
also be the last, unless we learn how to avoid
the risks.â Stephen Hawking
âBy 2030, the global GDP could rise by 14
percent as a result of AI enabled activities. That
is equal to $15.7 trillion:. PWC
âArtificial intelligence (AI) will be
more transformative to humanity than
electricityâ. Sundar Pichai
âJapan should make AI a mandatory subject for
college entrance examsâ. Masayoshi Son
âExplainable AI wonât replace
human workers; rather, it will
complement and support people,
so they can make better, faster,
more accurate decisionsâ. âAI
technology can enhance business
productivity by up to 40%â.
Accenture
27Dr. Amit Gangwal
31. Three different domains of AI,
which almost cover all aspects or areas of AI
â˘This deals with the physical world, is able to directly
interact with humans. Robotics are in use to improve our
work in various ways viz. Fordâs exoskeleton or some of the
top class players in the field like
Boston Dynamics & MIT Labs.
Robotics
â˘A great example of this is chatbot. Chatbots are a very easy
to understand example. Here humans and machines work
together to accomplish a goal. A chatbot is a communication
interface e. g. the one which interacts with us on various
e commerce and other websites.
Cognitive
systems
â˘It deals with the information/data world. Machines use data
to learn. Machine learning, a important sub area of AI, aims
to get meaning from that data by using statistical methods to
enable machines to improve with experience.
A subset of it is deep learning.
Machine
learning
29Dr. Amit Gangwal
32. Various components of artificial intelligence
Applications Types of Models Soft/hard ware for
training & running
models
Programming
languages for
building models
Image recognition
Speech recognition
Chatbots
Deep learning
Machine learning
Natural networks
GPUs
Parallel processing
tools
Cloud data storage
Python
Java
C
TensorFlow
30Dr. Amit Gangwal
33. S.
No.
Applications of AI Examples
1 Language translation
applications
Google Translate
2 Word processors Microsoft Word, Grammarly,
U Dictionary etc.
3 Personal assistant OK Google, Siri, Cortana, and
Alexa etc.
Some very common applications of AI
31Dr. Amit Gangwal
34. Machine Learning
Data Training Model
Training data
Model (w, b) Prediction
Test &
update w, b
W: weights and b: biases
32Dr. Amit Gangwal
35. The prime goal is to permit the computers
learn automatically without any human
involvement or support and fine-tune
actions consequently.
33Dr. Amit Gangwal
36. According to Yufeng Guo, a Famous Developer and
Machine Learning Advocate @ Google Cloud;
âML is Using data to answer questionsâ.
HereâŚ.
Using data means training & answer questions means
prediction
âData is key to unlocking ML and ML is the key to
unlocking the insight hidden in dataâ. 34Dr. Amit Gangwal
37. Study from data, recognize patterns and make decisions with negligible
human intervention
Can understand images, sounds, and language.
Instead of writing a brand new algorithm, machine learning tools
empower systems to develop and refine algorithms, by finding some
arrangements in huge database.
Finding use from helping people with debilities to felicitating
businesses with decision-making and dynamic pricing, to identifying
skin cancer, to sorting vegetables, to detecting escalators in need of
repairs.
35Dr. Amit Gangwal
38. Whether a chemical structure is suitable
for diabetes treatment or heart problem ?
Letâs imagine that you have to create a system that answers
above question.
This question answering system that experts build is called a
âmodelâ, and this model is shaped via a process called
âtrainingâ.
The aim of training is to develop a precise model that
answers questions fittingly most of the time.
Finally to train a model, we require to collect data to train
on. 36Dr. Amit Gangwal
39. Due to complexity & size of data or pattern and their
association, these could have easily been overlooked
by human observations or insights.
37Dr. Amit Gangwal
40. Quickly we see definitions and
examples one by one
Regression & Classification
are the two types of supervised machine learning methods.
38Dr. Amit Gangwal
41. Classification
Learns a method for predicting the instances class from pre-
labelled (classified) instances.
It is a supervised machine learning and predictive method. It
is highly applicable in pattern recognition etc. e.g. decision
tree, Bayesian classifier.
The quality of a classification analysis is often assessed via
precision and recall which are popular metric procedures.
E.g. a classification algorithm would train a model to identify
whether a certain cell is malignant or benign.
39Dr. Amit Gangwal
42. Algorithms used in Classification
⢠Decision Tree
⢠Naïve Bayes
⢠Random Forest
⢠Logistic Regression
⢠KNN
40Dr. Amit Gangwal
43. Regression
An attempt to predict a continuous attribute.
It is a predictive method.
Supervised.
A regression problem has a real no as its output e. g.
we could use the data in the table below to show
relationship between weight and height of people.
41Dr. Amit Gangwal
46. Supervised learning
⢠When data is labelled; after training and testing
model is ready for real problem in hand.
⢠Data has labeled features that define the meaning
of data.
⢠A supervised machine learning algorithm is used
to solve classification or regression problems.
44Dr. Amit Gangwal
47. Here machine learning model knows the feature and
label associated with that feature.
45Dr. Amit Gangwal
49. Unsupervised learning
⢠When machine learning models , find some hidden
patterns in vast amount of data and they cluster or
classify the data set or information in a concrete
manner e. g.
Flagging or highlighting or removing or suspending
some account on social media sites like Twitter or
marking few emails as spam by Gmail
47Dr. Amit Gangwal
51. Name of cricketers Run scored Wickets taken
A 500 10
B 450 00
C 200 00
D 25 45
E 01 40
F 25 30
G 26 40
H 89 55
Batsman
A, B, & C
Bowlers
D, E, F, G
& H
49Dr. Amit Gangwal
52. Reinforced Learning
(model learns from rewards)
Working of self-driving cars &
Learning to navigate staring by robots.
50Dr. Amit Gangwal
55. Following seven steps are required to use ML
for a defined object to get a desired or
predicted output
Data collection
Data preparation
Choosing a model
Training
Evaluation
Parameter tuning
Prediction
53Dr. Amit Gangwal
61. In machine learning, there are many mâs as there may be many
features. The collection of these m values is typically formed into a
matrix, that we can show using W, for âweightsâ matrix.
Similarly for b, and together they will be called the biases.
The training process includes starting some random values for W and
b and attempting to predict the output with those values. At the outset
it may perform poorly. By continuous tweaking or adjusting of fine
tuning the values in W and b, we can accomplish correct prediction
matching with best output. Lot many frequent cycle of updating the
weights and biases are inevitable to get it. Each repetition or cycle is
called one training âstepâ.
59Dr. Amit Gangwal
62. A feature is one column of the data in your input set.
For example, if you are trying to predict the type of
pet someone will choose, your input features may
include family income, age, home region and others.
The label is the final choice, such as dog, fish etc.
Example
Once you are done with training the model, you will
expose it to set of new input having same features.
It will present the predicted label that is type of pet for
that particular person only.
60Dr. Amit Gangwal
64. AI
⢠IBM deep blue
chess program
⢠Electronic game
characters
ML
⢠IBM Watson
⢠Google search
algorithms
⢠Amazon
recommendations
⢠Email spam filters
DL
⢠AlphaGO
⢠Natural speech
recognition
⢠Self driving
vehicles
Some of the top class examples of AI, ML & DL
62Dr. Amit Gangwal
65. Artificial intelligence Machine learning
It is an independent concept. It is a subset of AI.
The purpose is to increase chance of success. The aim is to increase accuracy, but it does not focus on
success.
It works as a computer program that does smart work. Here concept is that machine takes data and learns from it.
The goal is to mimic natural intelligence to solve
complex problem the way human brains does.
The goal is to learn from data on certain task to maximize the
performance of machine on this task by self learning.
AI will go for finding the optimal solution. Solution will be there but does not guarantee for being optimal
one.
AI generates intelligence or wisdom. ML leads to knowledge.
Mainly focuses on automation by utilizing such field as
image processing, cognitive science, neural systems,
machine learning etc.
Influences userâs machine to gain from the external
environment like sensors, electronic segments, external storage
gadgets etc.
AI empowers framing of machines, frameworks and
different gadgets and letting them think.
It depends on user inputs or query. The framework checks if it
is available in its knowledge data base or not. If yes it will
show the output, if not machine will fetch user data will
enhance its knowledge base, to give a better value to the end
user next time.
Intelligence is the gaining of knowledge and the skill to
apply it
Learning is using past instances to make future decisions.
It focuses more on automating a task or a system, like
cars
It focuses on gaining and applying knowledge from the
external environment, like Cortana
AI enables the machines to think and perform routine
jobs that humans do, such as assembly line operations in
a factory
ML provides solutions on the basis of a constantly evolving
neural network
AI vs ML
63Dr. Amit Gangwal
66. But what about ethical issues
Your data
Storing
Sharing
Deleting
Selling & what not
Bias in AI driven or empowered tasks or innovations
Black or white people
Rich or poor
Arts or STEM
Claims on IPRs
Machines or Animals or Human Beings
64Dr. Amit Gangwal
67. Yuval Noah Harari, an Israeli historian & a
professor in the Department of History at the
Hebrew University of Jerusalem
Donât let your choices be known to âŚâŚ
New jobs and your up/re-skilling will compete with each other
to test your patience 65Dr. Amit Gangwal
69. An under trial self driving car of
Uber hit and killed a joyrider.
⢠Who will be held responsible?
⢠What if the algorithm has been designed to kill a
particular gender, community, group, race, class, citizens
etc.
⢠Or if a driver-less car can save the life of only one
pedestrian out of two, what will be its choice?
67Dr. Amit Gangwal
70. If we are biased then AI output will
also be biased, until we design a
neutral algorithms.
68Dr. Amit Gangwal
71. Here who will play major role to
make AI or Disruptive Innovations
Neutral
Socially responsible
Kind enough at least to its master, let alone other
Should not store information or indulge in any sort of
mal-practices
?
69Dr. Amit Gangwal
72. What next ?
⢠In coming years, you will be able to do your full
body scan, biochemical tests at home only suing
various hand-held and wearable devices which
will be managed by AI, ML, IoT, Blockchain etc.
⢠Depending on ethical considerations and
economical approach you may be allowed to print
your body organs using 3D printers.
⢠DIY diagnostic test machines will be a common
scene like ATMs, on which you can have print of
all your body parameters.
DIY
70Dr. Amit Gangwal
73. What for a nation builder that is for a
Teacher
?
71Dr. Amit Gangwal
74. Being Teachers
We should be the first to
Sense
Bring
Create
Develop
Incubate
Nurture
Patent & Implement
âDisruptive Innovationsâ 72Dr. Amit Gangwal
75. Why to wait for Corona sort of
tragedy ?
73Dr. Amit Gangwal
76. Re-skilling & Up-skilling are now more
important than core degree certificates;
like never before.
Are you implementing Train the Trainer
concept?
74Dr. Amit Gangwal
78. ⢠Are we ready ?
⢠Why few of us were waiting for Corona for coming out of comfort
zones to hug technology and other teaching âlearning tools?
⢠Are we preparing our students mentally & academically on this
front ?
⢠Are we undergoing re-skilling and up-skilling so often?
⢠Are we enabling out students to sit on a giant or we helping them to
become one?
⢠Are we challenging average so often?
⢠Are we exposing students to higher difficulty level frequently?
⢠Are we producing superior replicas?
⢠Are we visiting industries at least twice in a year ?
76Dr. Amit Gangwal
79. ⢠Do we share top quality learning tools and content from internet
with our students without any fear or complex?
⢠Are we reading same day latest news from newspapers directly in
classrooms?
⢠Are we inculcating /instilling/imbibing Sanskar in our students ?
⢠Are we only producing executives or some business tycoons and
some
Ratan Tatas
77Dr. Amit Gangwal
80. All teachers, regardless of their stream or expertise
should do some basic or advanced course on
⢠Artificial Intelligence
⢠Machine Learning
⢠Python
⢠Tensorflow
⢠Sustainable Development
⢠Innovator Leadership
⢠Design Thinking
⢠Emotional Intelligence 78Dr. Amit Gangwal
81. ⢠This courses are other than your core subjects
re/up-skilling; but equally important to survive.
79Dr. Amit Gangwal
82. There is no harm in becoming an
incessant learnerâŚâŚ
⢠Even if you are a civil engineering teacher, you
should know
â coding basics
â basics of algorithms
â working on excel
â General international affairs
â Data presentation skills/tools
â ERP, CRM, BI, Tableu etc.
â Most-in-demand skill sets required by head hunters in
industries.
80Dr. Amit Gangwal
83. In case if
you want to
âempowerâ
yourself or
your
students
81Dr. Amit Gangwal
84. Images and infographics are taken from original
resources, just to disseminate the information
and not for commercial purposes.
I am thankful to Boston Dynamics, MIT, Tel
Aviv University, Apple, Google and other
original copyright holders.
Major infographics from Simplilearn, with
thanks.
Dr. Amit Gangwal