Open Forum - AI apps in Operations management

On 10 March 2023 in the Open Forum presentation, we discussed the scope and depth of AI applications in the field of Operations management. 

The basic question doing the rounds was as digital converts do we have the moral and ethical right to hand hold the digital natives, the younger generation. 

AI normally helps to enhance and / or automate Operations functions in organisation. 

Will AI through Machine Learning and reinforcement learning enrich and optimise Operations management functions or totally displace it with machine learning applications trained on large training data sets.

Role of AI in higher education ..

 

Higher education has had lot of challenges over the years the more important being that the way higher education has been delivered over the past two centuries. 

Benjamin Franklin had once said that "an investment in education is the best investment ever possible"

While educators have been thinking deeply of how to reduce the costs for education, improving the delivery, whether one to one peer education or one to many classroom style, the costs can be very prohibitive. It can play a role in the modern day as the old one-to-many model, the challenges have been many, especially with the onset of numerous technology tools like e-learning and so on. The advent of AI chatbot, ChatGPT it is claimed will change the face of education and teaching like never before.

There are several ways that AI can help improve higher education:

1. Personalized learning: AI can help tailor course content and recommendations to individual students based on their strengths, weaknesses, and interests, providing a more personalized and effective learning experience.
2. Grading and feedback: AI can help grade assignments and provide feedback to students, freeing up teachers to focus on more high-level tasks such as providing one-on-one support to students and designing new courses.
3. Improving efficiency: AI can help streamline administrative tasks, such as managing course schedules and enrolments, allowing educators to focus on more high-impact activities.
4. Enhancing research: AI can help researchers analyze large datasets and identify patterns and trends, leading to new insights and discoveries.
5. Improving accessibility: AI can help make education more accessible to students with disabilities, by providing tools such as text-to-speech and translation capabilities.
6. Course design: AI can be used to design and optimize courses, including selecting relevant materials and creating personalized study plans for students.
7. Tutoring and support: AI can be used to provide 24/7 support for students, including answering questions and offering guidance.
8. Adaptive testing: AI can create personalized tests for students, adjusting the difficulty of questions based on their responses.
9. Automated grading: AI can grade assignments and exams, providing instant feedback to students and freeing up teachers' time for other tasks. It is the fear of physical assessment that drives faculty to not go for quizzes ..
   
10. Virtual tutoring: AI can provide one-on-one tutoring to students through virtual assistants or chatbots.
11. Content creation: AI can be used to create personalized learning materials, such as customized lesson plans or learning modules, tailoring each course delivery suiting students needs and aspirations.
12. Plan and deliver effective quizzes / feedback systems : students are more committed and engaged if they get to know their progress in almost real-time. AI systems can help in that.
13. Predictive analysis: AI can be used to analyze student data and predict academic performance, which can help educators identify students who may be at risk of falling behind and intervene early and give challlenging
    
14. Create personalised learning experiences that could remove the challenge of disengagement that is happening with some of the students. (click here for HBR 2019 article)
15. Making education affordable : Many aspiring students have to discontinue higher studies due to the exhorbitant costs. AI can democratise and make economical higher education for the masses by making knowledge more available to the masses. The issue of EXCLUSIVITY fades into the future ...
16. Handle student queries more effectively and personally helping them make decisions to pursue higher education or join Universities quicker and efficient
17. Engagement and commitment : While in the University chatbots for academics and other extra curricular activities can enable students to be more enagaged and committed to their career 
18. More effective solutions : AI can help in coming with better possible solution than human minds can ever perceive or think, resulting in more ingenious applications
    

Overall, AI has the potential to personalise and revolutionize the way we think about and deliver education, making it more personalized, efficient, and effective.

For students sharing of personal data with different sites on the Internet can have a great liability into the future and can disincentivise students from sharing their data. Sites like mydata.org can help students to discern with which sites they can share or not. 

Recently, the AI based chatbot at the University of Murcia in Spain has been able to handle almost 39,000 queries from prospective admission seekers of the facilities and teaching learning experience for the students with almost 91% correctness.

While we may be aware of the benefits of AI in higher education in the modern context, there are challenges too that need to be handled with great caution. 

• high cost, 
• compromising privacy (fear of losing control over personal data), 
• lack of understanding for some, 
• fear of bias, 
• fear of unemployment due to automation, 
• issues with the seamless integration with existing educational practices and systems are challenges that  higher education systems across the world can face.

George.. (A big thanks to OpenAI, some of the inputs of this blog have been adapted from ChatGPT)

Artificial Intelligence and Amara's Law ..

 

In the continued experience of technology and the challenges, we have come across one more law, it is Amara’s Law (named after the scientist and futurist Roy Charles Amara (1925-2007)):

Amara's law sates that,

We tend to overestimate the effect of a technology in the short run and underestimate it's effect in the long run.

In the short run, humans tend to overestimate the impact of technologies, we tell great things about technologies, what it can accomplish and so on. The technology takes time to be assimilated among the population and the growth is really incremental in the initial stages. But in the long run, all the small deltas (increments) get added up and the impact is really great.

In the graph below, we plot time on the X axis and the impact of the modern technologies on the Y axis. (Image credit Shubham Vyas, IIT Guwahati)

As an example, when Internet got introduced in the commercial world in the late 90s, we predicted the great impact it would have on humanity but we saw the great dotcom bust and our short term dreams all went bust by 2004. But not losing heart, the Internet usage and applications got wider and wider acceptance and we see in just 25 years, it is penetrated all aspects of human life, from health, finance, commerce, education, news, banking , what not .. !!

Similarly we understand that the new technologies of AI, Virtual Reality and Autonomous Vehicles are benefiting humanity in increments presently, as it is in the evolutionary stage, but in the long run will provide revolutionary benefits. It will change the way we interact and deal with each other, with nature, our environment, our surroundings and with machines.

The impact of AI technologies on humanity will be profound in the long run ..

That day is not far at all, maybe in the next 20 years ..

George ..

Using AI in Alliance University Bangalore classrooms ..

 We have all the time been hearing, and indeed have a lingering fear in the back of our heads of how AI can replace humans and how it could be a threat to humanity in the long run. But carrying out an instance of an effective case of human-AI classroom collaboration, this impression is getting a changeover.

Human-AI collaboration is the study of how humans and artificial intelligence agents work together to accomplish a shared goal. AI systems can aid humans in everything from decision making tasks to art creation. -wikipedia

Human-AI symbiosis means interactions between humans and AI can make both parties smarter over time - H. Jarrahi

Here we explain a case of human-AI collaboration which is maturing into case of human AI symbiosis.

An AI depiction of Alliance University

A very practical case of human AI collaboration has happened in the MBA sem 3 Operations classroom in Alliance University Bangalore, India from August 2022- Dec 2022 when in the MBA third semester subject of Operations Strategy and Environmental Sustainability, in about 20 sessions, the 46 students had the benefit of taking 22 quizzes, ie. at the rate of 1.1 quiz for every classroom session, besides four internet based multimedia quizzes too. The benefits in terms of better understanding of concepts this has brought to the student community is beyond any explanation and is being assessed. 

 

In a 2020 study (click here) conducted at the Iowa state University in US it was found that the students who were quizzed at least once a week tend to do better in their understanding and grades than those who were not. 

 

The AI intervention in this case as written b7y the author in Alliance University Bangalore was not in the quiz administration but in the help offered by Google AI that helped the faculty member to prepare the high quality quizzes with the help of AI resources.

The multiple quizzes which were administered almost on a daily basis would not have been possible but for the big help offered by the Free and Open Source software MOODLE, the Learning Management System (LMS software) in Alliance University, Bangalore, to a very large extent Google AI and the Google cloud.

In quick feedback taken from students, they were positive at being able to assess for themselves their own understanding of the concepts, get almost daily and weekly feedback on their improved awareness of subject fundamentals to help improve their performance in the subject. Final exams are almost a fortnight away, student performance at that would highlight the real benefit of AI in aiding and improving student academic achievements, understanding and grades.

Google Decision Tree AI algorithms aided in preparing questions that  accessed the extremes of human ingenuity, understanding and comprehension in each subject area, which was otherwise impractical and out of reach of human intelligence and perseverance. Entire case studies were also prepared courtesy Google AI.

Looking forward eagerly to complete this Human-AI collaboration classroom exercise this semester and analyse the results in depth over the coming weeks and see this mature into a human AI symbiosis exercise soon. This would result in spanning out further  human-AI collaboration classroom exercises for other courses and get a comprehensive outcome, promising or otherwise, that could further push the boundaries of student understanding, grip and mastery over a subject domain and help the student community the world over.

George

Free and Open Source Software in AI field ..

 

Free and Open Source Software (FOSS), as the name implies, tells us that the source code of the software is open and anyone can see how it works and with specific knowledge can work to get it rid of virus or privacy violations. Offering free access, it means that anyone can work on the software to whatever depth they want, offerin g the whole software back to the community for public use and benefit.

In the world of Microsoft offerings, the Free and Open Sourec Software ofered by the Gnu/Linux group and Linux Foundation was a source of inspiration to kickstart the Internet revolution. Open Source software on the other hand, only the source code is open, the rights to edit the source code is with the organisation releasing the open source software, eg. Google and its Android OS.   

ChatGPT has this to say, 

Open-source software is software that is available to the public for use and modification. It is typically developed by a community of volunteers, who work together to improve the software and share their modifications with others.

One of the key features of open-source software is that the source code is made available to the public. This allows anyone to view, modify, and distribute the source code, as long as they follow the terms of the open-source license that the software is released under. This allows for a high level of collaboration and innovation, as developers can build upon and improve upon the work of others.

There are many advantages to using open-source software.  

• It can be freely used and modified by anyone, which can lead to a diverse range of applications and a large and active community of users and developers.  
• It can also be more secure and reliable, as the source code is available for anyone to review and identify potential vulnerabilities.  

Many of the most popular and widely-used software tools and platforms are open source, including the Linux operating system, the Apache web server, and the TensorFlow machine learning library.

Because of the  free use of the software libraries in AI, the growth of the software system instead of being linear is exponential. The field of Internet also saw the fast growth thanks to the use of Free and Open Source software like Apache webserver, the kernel for the Operating system put by Linus Torvalds etc. the development in the Internet area was exponential. Thanks to Google, Android was a free software based on Linux given free to the world. Almost 100% of the Fortune 500 companies work with Linux / FOSS as their backend.

Tensor Flow, the basic software library for Machine Learning and Artificial Intelligence was released as Free and Open Source by Google in 2015, meaning anyone from around the world can work on it and improvise it for the benefit of mankind.

TensorFlow is a free and open-source software library for machine learning and artificial intelligence. It can be used across a range of tasks but has a particular focus on training and inference of deep neural networks.

TensorFlow was developed by the Google Brain team for internal Google use, but it was released as an open-source library in 2015. It has since become one of the most widely-used machine learning libraries in the world, with a large and active community of users, contributors, and developers.

TensorFlow is designed to be flexible and efficient, with a focus on running machine learning models on a variety of platforms, from desktop and server environments to mobile devices. It can be used for a wide range of applications, including image and speech recognition, natural language processing, and predictive modeling.

Thanks to the Free and Open Source Community, the field of Artificial Intelligence is also bound to grow exponentiall in the coming years.

George. 

My first AI experiment with Diffusion AI

 Text to image generation by diffusion process in AI is an interesting area of Machine learning which builds models on vast amounts of data collected.

Deep fake AI

In Dream image generating AI app, I used the seeding words 'Image of a boy in a flowery meadow' and this is the result. (bottom left)

As different from deep faking apps, tech two years old, where people's portraits are inserted into existing images and videos, see my example, the latest technology is slightly interesting and advanced.

 

This worked using Generative Adversarial Networks (GAN)  having a generator and a discriminator. While the generator produces synthetic examples from random data, the discriminator component tries to distinguish between synthetic examples and real examples from a training dataset

 

Diffusion model on input Boy in a flowery meadow

AI has gone forward using Diffusion technology where using Contrastive Language Image Pretraining, diffusion system reconstructs data from noise, based on the word prompts. 

 

It is analogous to a master sculptor telling a novice where to chip a marble block to get a beautiful marble sculpture. 

 

On the right is an image generated using diffusion, where I have given the prompt words "George Easaw working in Alliance University Bangalore"

Diffusion model on prompt words of author at AU Bangalore

"At a high level, Diffusion models work by destroying training data by adding noise and then learn to recover the data by reversing this noising process. In Other words, Diffusion models can generate coherent images from noise. Diffusion models train by adding noise to images, which the model then learns how to remove" - scale.com

Diffusion model on prompt words of author at AU, Bangalore

The two high resolution images given to the left and right are generated from two different noise models. It is analogous to saying the two marble sculptures are taken from two different marble blocks excavated from two geographically different locations.

 

Click here for a practical guide to Diffusion Models.

Diffusion is applied these days in bio medicine to understand new treatments and to biochemistry to come out with new DNA sequences and molecules.

Diffusion models are generative models in the sense that they are trained to generate the same data output on which they are trained. How diffusion models work, click here ..

Sky is the limit, Diffusion is being used to compress images, generate videos, synthesise speech etc.

George