How can Artificial Intelligence (AI) help business ?

Having read lots about the benefits of AI and how AI would transform humanity, I was very inquisitive to know in spite of all the hype around AI, have we started using AI in a big way. 

In Dec 2018, I was attending a one day program on AI at the Indian Institute of Management, Bangalore which had an afternoon session with IBM engineers. Towards the end of their demo, I asked the engineers what were the actual applications of AI by IBM in the present day world, at least in India. The engineer was frank enough and admitted that it was just chatbot applications. 

AI had not matured beyond that at least for the industry. Even after two years, I find other than Google in their Maps, Gmail, search apps, not much AI is available in public domain for people to start using and experience the benefits.

While going through the article AI for the real world by Prof. Davenport and Ronanki in Feb '18, HBR (click here), I was able to understand what were the business benefits of AI and in general what were the AI challenges that global organisations face. 

Why are organistions skeptical of implementing AI when there is so much hype around it, is a question that would naturally come to any manager or engineer working in the industry. 

An anagram IEPPIO is apt for this occasion as per primary research conducted by the HBR paper authors on 250 executives in 152 AI projects, gave the real challenge faced by AI in the real world. 

I - ntegration

E - xpensive

P - oor understanding

P - eople

I - nnovative technology and 

O - verhyped technologies

The world is yet to fully understand the cognitive ability of humans, how humans distinguish one decision from another and in spite of the best intentions, fails completely in this area.

Of course, the technologies are too expensive, which takes it out of the realm of ordinary startups and medium sized organisations. 

We are yet to fully understand the nuances of AI and its applications

We do not yet have the real human resources to support full deployment of AI across organisations

Even though we extrapolate saying what all the technologies can do, we have immaturely developed technologies, ie. over hyping it.

The over hype or over selling of the technology has created a sort of glass palaces in the minds of the public and they expect AI to change the world radically overnight, which is not possible.

Once we know the challenges faced by AI, it makes sense to understand how we can overcome it. 

Understanding the Business benefits of AI

• I mprove
  
• O ptimise internal systems
• F ree workers from monotoinous tasks to do creative work
• B etter understandiong of p[roducts and processes
• C reate new products
• O ptimise external processes like Marketing etc
• P ursue new markets
• C apture new knowledge to venture into new products and processes
• R educe headcount at organisations
  

The potential business benefits of AI is great as how we understand it, the benefits may again increase over time as we progressively understand more features of the AI system. 

All these benefits if exploited well can result in great commercial benefits, besides reducing the boredom and monotony experienced by employees at their workplace.

George..

Role play on Artificial Intelligence in the class

We all know of the very many ways in which Artificial Intelligence (AI) is playing a very important role in our daily lives.  From the very simple search on Google, to replying to our emails on Gmail, to translating docs from one of the popular languages on Google Translate to another language, to the use of Personal Assistants from Amazon, Google, Microsoft or Apple, the uses of AI are varied and colourful. 

One of the biggest challenges for an AI system is to collect enough training data to prepare the AI system to learn and help make correct future decisions. If we start on poor quality training data and low volumes of training data, the output also would be equally poor, incorrect and incompetent.

The availability of large volumes of high quality training data is gives organisation the early mover advantage. 

The entry of Google in the search domain decades back gives it so much of training data and strong algorithms that can help it get the right search output for your search inputs. It will be very difficult for a new search engine incumbent to beat Google in search at least for the next fifty years. 

Maybe a new SEARCH ENGINE player with a 

• radically different approach to AI effectiveness or 
• strong faster processors working on quantum computing or
• ultra fast communication protocol for the results from servers to customer desktops or
  
• an efficient, innovative and quick algorithm may be able to thwart Google.

The role play being designed in the class for students is to help them understand the importance of high quality training data for a high quality AI output. 

What are the different ways in which an organisation can try to collect large volumes of high quality training data in a short time ? (click here for interesting AI based case studies)

Can it collect this training data from 

• different functionally related areas, 
• geographically different areas 
• different set of customers
• equivalent product markets ??

Let us leave it to the ingenuity of the students to tell us from where to collect the training data to help the AI systems take the right decisions .. 

Ref : 1. Ajay Agrawal,Joshua Gans, and Avi Goldfarb, How to win with Machine Learning, HBR, Oct '20.

George..

Types of AI

Artificial Intelligence is growing these days and is attractinmg attention of scientists and managers from across the globe.

Referring to the article 7 types of Artificial Intelligence, in Forbes of June 2019 where the seven different types of AI classifications are briefly mentioned.

The basic classification of Artificial Intelligence is of two types based on the likeness to the human mind and the ability to analyse.

According to the classification of AI based on it's likeliness to the human mind, Reactive, Limited Memory, Theory of Mind and Self Aware AI are the different classifications.

1. Reactive AI machines are 

• limited in their applications.
• do not have memory based functionality
• no ability to learn from previously gained experiences
• can only respond to a limited set or combination of inputs
• cannot improve their operations

An example is IBM's famous chess playing Super Computer Deep Blue. 

2. Limited Memory AI is what we observe in chatbots, Personal Assistants like Amazon echo, Google Home, Microsoft Cortana and Apple Siri, all self-driving vehicles.

In addition to the capabilities of Reactive Machines, these AI machines can learn from historical data to make use in decision making. Image recognition that makes use of Machine Learning and Deep Learning is an example of this. 

Image Courtesy, Forbes.

3. Theory of Mind AI is futuristic and exist only as a concept or work-in-progress

A Theory of mind AI will be able to better understand the entities by being able to discern the needs, emotions, beliefs and thought processes of the entities it is interacting with.  Theory of Mind AI tries to understand humans better.

4. Self Aware AI is akin to the human brain with all it's rationalising and emotional capabilities.

Though it may takes decades or centuries to fully realise and understand this type of AI, it can change humanity for the better or consign it for doom or disaster. The potential for self-preservation of these AI machines, can take over or spell the end of humanity, which can be a great threat to the world.

The other three types of AI are 

• Artificial Narrow Intelligence which we see in all our current applications with learning capabilities and ability to perform repetitive tasks autonomously using human like capabilities. 
• Artificial General Intelligence is futuristic and is the ability of machines to learn, perceive, understand and function like human beings. They develop competencies across multiple capabilities and domains with reduced learning time to carry out tasks efficiently and accurately.
• Artificial Super Intelligence is the ultimate in AI development. Because of their multi faceted intelligence these machines will be better capable at doing all things better than humans because of their faster memory, data processing, analysis and decision making capabilities. These machines can also spell the doom of humanity, thus being a great risk. 

How can humanity make effective use of these different types of AI to help in growth of humanity and not its doom is a serious question facing human civilization now. Can we, if not control, at least co-exist with AGI or ASi to help humanity for the next millennia ?

George.

Will AI enable a big revolution in the education scenario ?

Education is the process of facilitating learning, or the acquisition of knowledge, skills, values, morals, beliefs, and habits. Educational methods include teaching, training, storytelling, discussion and directed research. Wikipedia

The above definition of Education is very vast, sweeping and cuts across the breadth and width of human civilization.  So far we were depending on the human route to enable transfer of skills and knowledge from one generation to the next. With increase in population and the vastness of knowledge that needs to be transferred along with the multiple branches of knowledge that exists today, the correct streamlining of this knowledge, information and skills to the multude of people across the different continents of the world is very daunting. Can AI help to solve this crisis and how ?

I happened to go through couple of international publications in this regard. Ron Schmelzer in July '19, Forbes (click here) shares some ideas of applications of AI in education.  

Artificial Intelligence (AI) can help education in many striking ways. It can be used in the following ways.

1. Personalised learning - Instead of providing a single syllabus for all the students, learning will be personalised. What to learn personally, divide learning into small convenient modules for asynchronous learning and asynchronous testing, preparation of custom learning modules and learning material for students. This will also include online interface and feedback from teachers and Professors

2. making use of AI Personal Assistants like Amazon Alexa, Google Home, Microsoft Cortana, Apple Siri to provide conversation interaction to students that assists in collecting teaching material, educational assistance, setting reminders, alarms, to help answer small queues of students and to help in adaptive learning.

Adaptive learning, also known as adaptive teaching, is an educational method which uses computer algorithms to orchestrate the interaction with the learner and deliver customized resources and learning activities to address the unique needs of each learner. Wikipedia

3. aiding in administrative tasks like admission, examinations etc. that take faculty time away from students. This time can be reallocated to students increasing the faculty student interaction time. With AI Personal assistants taking the rote tasks of admission, institutional information seeking, arranging for field visits , placement activities etc more time is available for faculty to interact with students.

4. breaking barriers, cutting across the quality of the institution, teachers etc., AI systems can help ensure impartial decision making based on academic merit, improving the fairness and quality of the admission process.

An HBR article in Oct '19 by Lasse Rohinianen titled How AI and data would personalise higher education, (click here) talks of personalised learning that can be administered by AI to each student  for unique experiences helping Professors be more effective in teaching. This helps identify early student problems, avoiding students quitting and enabling successful completion. 

Chat bots that interacted with students in the University of Mursia in Spain helped improve the motivation levels of students by 91%. Understanding student interest and preferences, new courses and experiences were introduced to improve student learning, educational experiences, delivering the right knowledge and skills to improve later adult life professional accomplishments.

Chat bots can also help give mental health counseling to students. 

The challenges that were observed were

1. preparing students for tech based world

2. preparing students for digital transformation

3. AI cannot help in imparting creative, cognitive and emotional intelligence skills. This requires human intervention. 

Only Professors can help students by providing guidance, support, mentoring, understand the importance of what is being learnt and its practical applications. AI cannot help in this area.

With the invasion of AI in education, students need to understand how 

1. Algorithms use data to make decisions, 

2. how the data should remain private, 

3. how the data is being used and 

4. the design and development of AI systems.

A human-AI hybrid machine only can finally benefit humanity as a whole ..

George..

What is being customer centered like ?

Customer-centric (also known as client-centric) is a business strategy that's based on putting your customer first and at the core of your business in order to provide a positive experience and build long-term relationships. - superoffice.com 

Happened to read the article in HBR of Oct '18 on 6 ways to build a customer centric culture by Denis Lee Yvon. (click here) 

A company which gives the best product and service to customer will never ever fall short of customers as existing satisfied customers will help new customers to try the product. 

Customer centredness in other words means creating a positive experience for the customers. Getting the operations personnel in the organisation to understand the customer concerns and his/her pain points will help the organisation to understand the organisation better. 

This includes the following points 

1. maximise service / product offerings 

2. build relationships 

3. respect customer preferences 

4. keep cusomer satisfaction at the forefront of business growth 

5. engage with customers in innovative ways to understand customer preferences'

6. include customers in corporate decision making 

7. include customer in futire product/ service planning of the company 

8. involve customer to know his/her pain points in the interaction and try to solve them 

9. give more than what the customer wants or asks for 

10. given quality items and service offerings to customers to ensure repeat purchases 

11. meeting customers in person in fixed time intervals 

As per the above author, the organisation can build a customer centric culture as given below 

1. operationalise customer empathy 

2. hire for customer orientation 

3. decentralise customer insights 

4. facilitate interactions of employees with customers 

5. link employee culture to customer outcomes 

6. tieng employee compensation with customer service, giving employees skin in the game

7. adding strong reverse logistics network to handle customer returns unsatisfied products

 Amazon is said to be one of the best customer centred organisations in the world be it in the case of forward logistics or reverse logistics (returning unliked or poor quality items) without asking a single qiestion. The author heas been a beneficiary of this ASmazon India customer centred approach many a time and is highly appreciative of it.

The amazing growth of Amazon globally is indicative of the extreme customer centred approach it nourishes as part of the company culture.  

A University or a Learning Experience Center ?

When I was reading that Amazon is the most customer centered company in the world, I never expected it to be really obsessed with the customer.  Reading about Amazon and how in 1994, when I was teaching at the Goa govt. Engineering College, after PG in Industrial Engineering  and teaching Information systems to PG students, Bezos had already setup Amazon, after completing BSE from Princeton in 1986. I too completed my BTech in '86 from Govt. Engg college, Trivandrum CET, in India. The similarity is very striking. Bezos went ahead with an entrepreneurial zeal.

Taking a cue from Bezos, I was thinking loud, should we rename Universities as Customer Learning Experience Center. (CLEC) ..

Watch the video at 2012 re:invent meeting interaction .. (click here)

In the above video, Bezos explains the growth of Amazon Web Services, his personal experience with Lean systems, waste removal etc..

Passion and customer centricity are the two great qualities any entrepreneur should have ..

Between customers and competitors,  an entrepreneur should focus on customers..

 

George..

Is Uber a Disruptive or Strategic Innovation ?

The other day I was listening to an HBR podcast by Alexander MacKay, faculty member from Harvard Business School describing Uber's global strategy (click here). 

Uber the ride sharing app introduced by Travis Kalanick in 2009, has now changed to become the world's largest ridesharing app with 2020 revenues of $10 billion and offering 6.2 billion rides annually, ie. 17 million rides daily. It has a market cap of $60 billion presently.

Prof. Clayton Christensen, from Harvard Business School coined the term Disruptive Innovation around 1995 through his book The Innovator's Dilemma.

To qualify as a Disruptive Innovation,  an innovation should 

• start as a low quality product or service at the low end of the market and move slowly over time, disrupting established players at different levels by offering progressively higher quality and satisfaction to the customers
• be offered initially as a cheaper alternative to the existing products or service at the lower end of the market, gradually improving over time in quality and performance dislodging established players.

Uber is not a low quality product and it is not an initial cheaper alternative. Both these reasons disqualify the Uber phenomenon being classified as a disruptive innovation. (click here).

My earlier writing in this area has made the situation very clear. (click here) Taking the case of the Indian Space research programme. Starting in 1961, it started growing, slowly disrupting the launch of low earth orbit (LEO) satellite launch market dominated by ESA etc and then in 2007, through Chandrayaan 1 to the moon disrupted major players like ESA and China Aerospace, and then in 2013 with the launch of Mangalyaan to the planet Mars disrupted NASA in terms of cost and reliable technology. ISRO has not been able to do any major space disruptions of late, but it has come to be globally recognised as a space power to be reckoned by other space powers. The cost disruption initiated by ISRO at one tenth the cost of similar missions from NASA or ESA, in the planetary missions market is indeed laudable.

Tesla has been slowly disrupting the transportation scene through the launch of their electric cars. The Kochi international airport running on solar energy is slowly started disrupting the 10 million annual traffic airports, slowly building to 50 million annual traffic and then to 95 million, slowly disrupting the Jacksonville airport in US.

This article from medium (click here) explains the salient features of disruptive innovation, even though it says Uber is a disruptive innovation. Developments in AI, self driving cars, IoT applications are all going to disrupt the world in the coming days in ways humanity has not even dreamt before.

I have come up with a question set to decide whether an innovation is disruptive or strategic . Try to answer the following questions. 

1. Is there a large unmet, untapped market in need of the product or service because available goods and service are priced high for the uptrend markets ?
2. Is your product or service lesser priced than the market leaders ?
3. Is your product or service now able to meet the demand from the low quality segment of the market ?
4. Is there potential for the product or process to grow to satisfy higher end needs customers ?
5. Can the market leader drop or stoop down to your level and kick you out of the market by offering a similar low cost product or service ?

If the answers to the first four questions are YES and to the last question is NO, then your product or service is in line for a disruptive innovation. Finally we see, it is the reach, acceptance and the cost of the product or service that plays an important role to disrupt the leader and topple the apple cart. 

George..