00:00:25
Welcome to this NPTEL online certification
course on artificial intelligence marketing
00:00:31
and now we are talking of module 8. So we
are discussing how to transform marketing
00:00:37
strategy using AI. So now this is part 2
of this and we are in module 8. In this
00:00:45
module we will explore the ideas framework and
understand 5 elements of technology landscape.
00:00:51
Then we will study the expertise, architecture
and strategy landscape of ideas framework in
00:00:56
detail. Thereafter we will understand human
and machine expertise and capabilities.
00:01:01
Then we will understand the phenomena of
human teaching machines and build machine
00:01:06
expertise. Thereafter we will understand IT
architecture as living system. Study digital
00:01:14
decoupling to transform legacy architecture
into living systems and to study technology
00:01:19
integrated strategy to explore the stages
of human-machine interaction and then study
00:01:24
the technology integrated strategy in the
age of ideas. So let us again look at the
00:01:30
5 elements of the technology landscape
that are these 5 things intelligence,
00:01:36
data, expertise, architecture and strategies
and combined they form this idea framework.
00:01:43
Now let us look at this we have talked about
the intelligence and data now we will talk
00:01:48
about expertise. Machine teaching can unleash the
often untapped expertise that exists throughout an
00:01:56
organization allowing a large number of people
to use AI in new and sophisticated ways. It is
00:02:03
customizable according to business situations
thus it opens the way to real innovation and
00:02:09
advantage. Now let us look at the human-machine
hybrid activities. So machine augment human how?
00:02:18
First amplifying our powers and in providing
otherwise unattainable data-driven insights.
00:02:27
Interacting with us through intelligent agents
and embodying us as with robots that extend our
00:02:34
physical capabilities. In turn human complement
machines how? By training them as in labeling data
00:02:41
for machine learning systems. Explaining them to
bridge the gap between technologists and business
00:02:47
leaders and sustaining them by ensuring that AI
systems are functioning properly, ethically and
00:02:54
in the service of the human rather than the
other way round. And in turn human complement
00:03:01
machines by teaching machines the by endowing them
with the experience of experts. In the new world
00:03:09
of human teaching machines the real difference
makers for business will be the domain expert.
00:03:15
So that will be the real difference makers
who are the experts. So machine teaching
00:03:31
is about the transfer of knowledge from the
human expert to the machine learning system.
00:03:36
Machine teaching includes three distinct areas
of human expertise that AI has long struggled to
00:03:42
incorporate. One is the professional experience,
second is collective social experience and the
00:03:48
third is personal experience. The innate and
acquired individual abilities of human being.
00:03:56
Making AI innovation business specific.
Developers or subject matter experts with
00:04:01
little AI expertise such as lawyers, accountants,
engineers, nurses or forklift operators can impart
00:04:10
important abstract concepts to an intelligent
system which then performs the machine learning
00:04:16
mechanics in the background. Someone who
understand the task at hand decompose the
00:04:22
problem into smaller parts and sets of rules and
criteria for how the autonomous devices should
00:04:27
operate. Then using simulation software the
experts provide a limited number of examples.
00:04:34
The equivalent of lesson plan that helps the
machine learning algorithm solve the problem.
00:04:39
If the device consistently makes the same
mistakes additional examples can be added
00:04:45
to the digital curriculum. Once the curriculum
is in place the system automates the process
00:04:50
of teaching and learning across hundreds and
thousands of simulations at the same time. It
00:04:56
is not randomly exploring it is exploring
in a way that is guided by the teacher. So
00:05:01
we are exploring in a way that is guided by
the expert the teacher. Collective expertise.
00:05:08
Teaching AI social context. Humans operate
often effortlessly in collective and social
00:05:15
context of immense complexity. These contexts
overlap and inter penetrate and are constantly
00:05:24
evolving on short and long time scales.
When we maneuver a car through an urban
00:05:29
environment we are negotiating a dense
web of social systems. We are processing
00:05:34
and anticipating the movements of other
vehicles and the intentions of their drivers.
00:05:41
We are following and may be bending the
formal rules of road and engaging in in
00:05:46
the informal ones embedded in our culture.
For instance in some cultures flashing your
00:05:54
headlights on and off means you are yielding to
another vehicle. In other cultures it means you
00:05:59
are coming through and the other vehicle
dam will better get give way. So these
00:06:05
are the two opposite courses of action.
While the course of action was the same.
00:06:21
So now let us look at the wisdom of crowds
plus machine. Forecasting real life events
00:06:27
of importance. Example geopolitical events
behavior of drivers on a highway etcetera is
00:06:32
notoriously difficult. Experts predictive
accuracy tracked over time has been shown
00:06:38
to be comparable to a random guess. One way
to improve forecast is to crowd source them.
00:06:46
Aggregating a large number of human forecast into
a single estimate of probability. This wisdom
00:06:52
of crowds holds that large groups of people
outperform a small allied groups of experts
00:06:57
at solving problems making wise decisions and
predicting the future. In other word collective
00:07:04
expertise in the broader sense can sometimes
be superior to highly specific individual
00:07:10
expertise. At the same time advances in machine
learning have led to models that produce fairly
00:07:15
reasonable forecasts for a number of tasks.
The SAGE that is synergistic anticipation
00:07:23
of geopolitical events projects combine the
power of crowdsourcing with advances in AI.
00:07:29
Hence the term synergistic in its name to
generate more accurate predictions that
00:07:34
either method could on its own. In a competition
held to test the accuracy of forecasting systems
00:07:40
SAGE was tested against two competing systems
given the same set of more than 400 forecasting
00:07:46
questions SAGE won hence highlighting the
importance of collective expertise. Now let
00:07:54
us look at the personal expertise inherent
human technology. For decades AI researchers
00:07:58
have struggled with how to imbue machines
with the basic building blocks of human
00:08:03
intelligence. But the human turn in intelligence
is not about recreating human consciousness.
00:08:09
Instead it is about solving problems by
mimicking the most powerful cognitive
00:08:13
characteristics of human and supplementing them
with the most powerful abilities of computers.
00:08:20
The radically human turn in personal expertise
is about directly leveraging not mimicking the
00:08:27
innate and acquired intelligence of human
to augment AI. This can be more subtle kind
00:08:32
of teaching. A kind where tested skills
some that the teacher may not even know
00:08:38
they possess are subtly transferred to a
learning system. The next comes expertise.
00:08:45
In supervised learning scenarios machine
teaching is particularly useful when little
00:08:51
or no labeled training data exist for the
machine learning algorithms as is often
00:08:57
does not because an industry or a company needs
are so specific. Now let us look at the example
00:09:03
of Etsy. For Etsy an online marketplace for
vintage and handmade goods classifying them
00:09:12
by style was particularly challenging. So
most of the product on its sites are one
00:09:17
of a kind creation and there are some
50 million items on offer at any given
00:09:22
time. So now you see there are 50 million
items and every item is a different item.
00:09:28
In the past style based recommendation system
produced unexplainable product suggestions for
00:09:34
the group of shoppers because the AI assumed that
two items would be similar in style if they are
00:09:40
frequently purchased together. So that was the
problem. In order to teach AI the subjective
00:09:53
notions of style Etsy merchandising experts
developed 42 style labels that captured buyers
00:09:59
traced across 15 categories from jewelry to toys
to crafts. The merchandiser produced a list of
00:10:07
130000 items distributed across these 42 styles.
Etsy technologies then turned to buyers who tend
00:10:14
to use style related terms in their searches
typing in things like art, deco, sideboard.
00:10:22
From just one month of such queries the
company was able to collect a labeled data
00:10:27
set of 3 million instances against which
to test their style classifications. So
00:10:33
that happened within one month. Etsy engineers
then trained a neural network to use textual and
00:10:39
visual cues to best distinguish between
those classification for each item. The
00:10:47
result was style predictions for all
50 million active items on Etsy.com.
00:10:51
Now let us look at the implications
of machine teaching by experts. One
00:10:57
is the increased relevance. The search engines
allowed buyers to find products that express
00:11:04
their sense of taste and style. The second
is increased sales. During COVID-19 sales of
00:11:11
masks tailored to the aesthetic sensibility
of customers went from virtually nothing
00:11:15
in April 2020 to some $740 million for the
next of the year for the rest of the year.
00:11:22
The third is the company's revenues more than
doubled during that time and its market value
00:11:28
rose to $22 billion. The next comes the A
on the ideas framework that is architecture.
00:11:36
So legacy architecture are tightly bounded
maintaining barriers between lines of businesses,
00:11:41
geographies, sales channels and functions.
The wide range of emerging information
00:11:47
technologies supports the development
of IT architecture as living system as
00:11:53
shown in the next slide. So let us look
at the IT architecture as living system.
00:11:59
One is that it is boundary less. It breaks down
barriers within the IT stack and also between
00:12:07
companies using cloud based platforms to harness
network effects. Another characteristic of this
00:12:14
living system is that they are adaptable.
Rapidly adjusted to business and technology
00:12:19
changes. Adaptable systems move legacy systems to
the cloud to reduce dependence between systems.
00:12:27
Increase speed and efficiency, capitalize on human
intelligence of your talent and meet the evolving
00:12:33
needs of the customers. The third is radically
human modeled on human brain and behavior. IT
00:12:41
architecture as living systems are radically
human in the ways they use agile methods,
00:12:48
complex human intelligence, nimble data strategies
to deliver insight and trusted experience. Such
00:12:56
systems connect people across organization
silos, bringing together business talent,
00:13:02
IT talent and ecosystem partners
to innovate and co-create. Digital
00:13:08
decoupling, the first component of that
is legacy architecture to living system.
00:13:15
For many companies the journey
towards living systems begin with
00:13:19
digital decoupling. So that is the
first step. Using new technologies,
00:13:31
data access methods and development methodologies
to build new system that execute alongside the
00:13:39
legacy systems. This includes open application
programming interfaces that is APIs, agile DevOps,
00:13:48
cloud migration factories and robotic process
automation that enables greater flexibility.
00:13:55
Using this and other approaches, organizations
can gradually decouple their core systems,
00:14:01
migrating critical customer facing functionality
and data to new service based platforms.
00:14:08
Instead of periodic large scale IT transformations
that rigid architectures require, this decoupled
00:14:15
approach provides a stable and constantly evolving
architecture capable of accommodating innovation
00:14:22
and scaling to respond quickly to changing market
conditions and the competitive landscape. LL Bean
00:14:29
is a 110 year old retailer with a heritage that
include classic clothing, rugged outdoor gears
00:14:37
and a deep commitment to customer satisfaction.
So these are the three things that they have.
00:14:43
In recent years, as the company reached out
to customers across multiple channels, print,
00:14:50
brick and mortar stores, computers and
mobile websites, email and social media,
00:14:55
it found itself hampered by a cumbersome
IT system. Different platforms only loosely
00:15:02
connected supported different customer
channels running on separate applications.
00:15:07
So providing a seamless customer experience across
all channels was next to impossible because there
00:15:14
were so many channels. Integrating all those
channels and a seamless movement across the
00:15:37
channes was next to impossible. And instead
of focusing on delivering customer value,
00:15:48
IT personnel had to spend time managing
the infrastructure. So they were spending
00:15:53
the time on infrastructure that was bad
instead of delivering on the customer
00:16:00
value. To compete successfully in the age of
amazing, LL Bean decoupled machine critical
00:16:07
applications from its legacy IT systems
and located them in the Google Cloud.
00:16:14
Now let us look at the implications of using
a cloud based architecture. First is increased
00:16:21
productivity. IT teams could integrate data from
multiple systems, handle peak website loads more
00:16:29
efficiently and deliver new customer features
faster. Second is the increased efficiency,
00:16:36
continuous optimization of backend cloud
based architecture, less time spent by
00:16:41
front end developers on managing it. And the
third is the flexible front end architecture.
00:16:48
The company can easily, quickly and cost
effectively scale up capacity in peak buying
00:16:55
periods and scale down during the LULs, the
low period. The next component of this ideas
00:17:08
framework is the S that is the strategy. So
technology integrated strategy. So far we have
00:17:14
seen examples of several companies that have
adopted new approaches to intelligence, data,
00:17:21
expertise and architecture and created
distinctive strategies as varied as the
00:17:26
industry in which the firm competes. Let us
now take a look at the fifth and the final
00:17:32
element of the ideas framework which talks
about human centered technological strategy.
00:17:39
These are the three stages of human machine
interaction. The response of companies to
00:17:44
intelligent technologies has unfolded in
three stages. Stage one the machine centric,
00:17:52
stage is collaborative and the three is human
centric. The first stage of the evolution was
00:18:00
machine centric. The dominant response
to the new technology was to re-engineer
00:18:05
AI and other emerging technologies were
used to automate the repetitive tasks.
00:18:11
Humans had to adopt machines and were often
replaced by them. Strategy and execution was
00:18:18
sequential, spread over steps like assess,
identify, design and implement. The second
00:18:26
stage of machine human machine interaction was
collaborative. Human and machine adopted to each
00:18:33
other. As a new generation of intelligent
technologies and techniques emerged,
00:18:38
companies sought to reimagine their
traditional business processes in
00:18:42
order to take advantage of collaborative
teams of humans working alongside machines.
00:18:48
Nevertheless, strategy and execution remained
separate. First a process was reimagined in
00:18:56
light of AI, machine learning and the like
and then tested in small experiments. If it
00:19:01
passed the test, then it was implemented
at scale across the enterprise. Again
00:19:08
that was a sequential approach that
separates strategy and execution.
00:19:13
The third stage underway now is the human
centric. Machines adopt to humans. The
00:19:19
agility and adaptability of radically human
technologies guided by human enables savvy
00:19:27
companies to interrelate technology,
strategy formulation and execution
00:19:32
in an organic whole. The three elements grow
and change synchronously often very rapidly.
00:19:40
The technology integrated strategies for the
age of ideas. So, three prominent strategies
00:19:46
illustrate the wide range of possibilities that
radically human technologies have opened up.
00:19:52
These three stages forever beta,
minimum variable ideas and the
00:19:59
third is collab create distinctive advantages
for companies and customers alike. Now let us
00:20:06
look at each one of them. So, forever beta
strategies offer software enabled products
00:20:11
and services that continually evolve and
improve after they have been purchased.
00:20:15
So that customers sees them grow in value
and utility over time rather than fade.
00:20:22
MVI strategies use one or more elements of the
ideas framework to preciously target weak links
00:20:30
in a traditional industry and provide a superior
customer experience that can be quickly scaled
00:20:35
to make rapid inroads in the market. Collab
strategies produce superior results in the
00:20:42
sciences or other knowledge intensive environments
through human guided machine driven discovery.
00:20:48
Now let us look at the evolution of technology. A
subset of examples of how technology is advancing.
00:20:56
First is this intelligence. The first
is this intelligence. So machine centric
00:21:02
that is industrial robots. Uses sensors
to guide pre-programmed actions behind
00:21:08
safety gates. So this machine centric
machine can do what its program to do.
00:21:14
Second is collaborative. Human trained
machines machine augment humans that is
00:21:21
deep learning. Employ neural networks to
learn from large data sets. The third is
00:21:29
human centric. Machines adapt to
humans and human teach machines.
00:21:34
So that is emotional AI. Response to
human emotions and increases relevance.
00:21:41
Then comes the data. Again the three stages
are machine centric, collaborative and human
00:21:47
centric. So in data machine centric the
machines can do what it is program to do.
00:21:53
So the data in machine centric means business
intelligence one. Produce reports from database
00:21:59
queries and batch processes. Collaborative
where human teach machines and machines
00:22:06
augment human data is big data. Uncovers
actionable patterns from extremely large
00:22:12
data sets. In human centric where machines
adapt to humans and human teach machines.
00:22:18
The data is synthetic data. Mimics original data
with strong privacy safeguards. Again in at the
00:22:29
third stage that is this expertise. Again there
are three stages human centric, collaborative
00:22:35
and human centric. So expertise in machine centric
is traditional programming uses computer codes to
00:22:42
instruct machines. Collaborative is data science,
extract insights from data to solve problems.
00:22:50
And in human centric the expertise is machine
teaching enables non-technical experts to
00:22:55
train AI systems. The next comes architecture
that is. Again the three stages human centric,
00:23:09
collaborative and human centric. So
architecture in human centric situation
00:23:15
is monolithic, works as a homogeneous static
integrated system. Collaborative is layered,
00:23:23
handles functions autonomously at separate levels.
00:23:28
And human centric are living systems, assembles
heterogeneous adoptive capabilities dynamically
00:23:35
like the Lego blocks. Then comes the last one
that is the strategy. Again the three stages
00:23:44
machine centric, collaborative and human centric.
So strategy in machine centric is re-engineer,
00:23:52
foster sequential change from analysis to
execution. In collaborative re-imagine,
00:23:58
rethink processes then sequentially
experiments, adopts and scales.
00:24:05
And then comes human centric that is interrelate,
synchronize human guided technologies,
00:24:11
strategy and execution. So in order to conclude we
have studied the ways in which human and machine
00:24:18
complement each other. We discussed the three
types of expertise which are the professional,
00:24:23
social and personal expertise. We discussed
the collaborative wisdom of crowds and AI.
00:24:31
Then we have discussed the IT architecture, how
IT architecture is now evolving as living systems.
00:24:39
Then we have discussed the three dimensions
of IT architecture as living systems are
00:24:43
being boundaryless, adaptable and radically
human. Then we have talked about human-machine
00:24:49
interaction and how it has evolved
from being machine centric to human
00:24:54
centric. And then we talked about the three
strategies. The first is beta, forever beta.
00:24:59
The second is minimum viable idea
MVI and the third is CoLab. And
00:25:06
how are they being adopted in the
age of idea. And these are the five
00:25:11
books from which the material for
this module was taken. Thank you.
![The Complete Web Development Roadmap [2024]](https://ruhnrawpgxrzdrgaohnf.supabase.co/storage/v1/object/public/images/video/GxmfcnU3feo/thumbnail.jpg)
