Artificial Intelligence and its implications for Higher Education:
There is a delightful story of how Albert Einstein as a Professor at the Institute of Advance Study at Princeton University one day gave his students a final exam that was a year old. In fact, Einstein had given his students the exact same final exam the year before. His teaching assistant noticed the “error” and timidly made the famous physicist aware of his mistake. Einstein looked closer at the newly distributed sheet and answered: “You’re right, these are the same questions as last year – but the answers have changed.”
To appreciate this anecdote we must know that the year in this story is the year 1905, annus memorabilis for Albert Einstein in which he published 4 path breaking papers in Annalen der Physik.
The first paper explained the photoelectric effect which was the only specific discovery mentioned in the citation awarding Einstein the Nobel Prixe in Physics. The second paper explained Brownian movement , which led reluctant physicists to accept the existence of atoms. The third paper introduced Einstein’s theory of special relativity. The fourth, a consequence of the theory of special relativity, developed the principle of mass-energy equivalence expressed in the famous equation E = mass x c squared.
These four papers, together with Einstein’s theory of general relativity and quantum mechanics, are the foundation of modern physics.
And this year 2021 marks the centenary of Einstein being awarded the Nobel Prize for one of those 4 papers, that on the photoelectric effect.
In education, especially higher education we are in a similar situation today.
The questions are still the same – what do we teach, how do we teach, who does the teaching, where do we teach, how do we assess what has been learnt and how relevant is what we teach to help solve the problems of our times……but the answers are not the same as they were in the past.
There are two main drivers of this change. One is the fourth Industrial revolution to which attention was drawn by Professor Klaus Schwab in January 2016 at the meeting of the World Economic Forum at Davos. The other is the fourth education revolution which is the title of a 2018 book by Anthony Seldon. And to top it we had the Covid pandemic in 2020 whose impact is expected to last for a decade. The World Economic Forum in its meeting in January 2020 estimated that about a billion people have to be re-skilled and has launched a Reskilling Revolution, a multistakeholder initiative aiming to provide better education, new skills and better work to a billion people around the world by 2030.
The drivers of change are emerging technologies of Artificial Intelligence, Machine Learning, Internet of Things, 3D printing, robots, drones, Blockchains. 5 G, self-driving cars…SAV Shared Autonomous vehicles, Augmented Reality, Virtual reality and Quantum Computing.
But the big challenge is that the time lag between the speed of change and the time it takes to produce a workforce with the desired educational qualifications and skills is becoming unacceptable. We therefore have to deploy these very technologies to achieve our goals.
The usefulness of useless knowledge :
In his classic essay The Usefulness of Useless Knowledge, ( https://www.ias.edu/sites/default/files/library/UsefulnessHarpers.pdf) Abraham Flexner, the founding Director of the Institute for Advanced Study in Princeton and the man who helped bring Albert Einstein to the United States, describes a great paradox of scientific research. The search for answers to deep questions, motivated solely by curiosity and without concern for applications, often leads not only to the greatest scientific discoveries but also to the most revolutionary technological breakthroughs. For example, no quantum mechanics, no computer chips.
There has however been a clamour in the last decades for education that is immediately useful and relevant, and Universities are often accused of not doing so. But just like the time gap between the frontiers of knowledge and their applications is reducing, so is the rapid obsolescence of present day knowledge. The skills of a lifetime become obsolete in an instant, and this is even more true as AI and Machine Learning automate more human skills. This leads to a rapid uselessness of useful knowledge. Several studies and reports suggest that professionals in software engineering, marketing, sales, manufacturing, law, finance and accounting must update their skills every 12 to 18 months.
When faced with the question of how to teach, we must consider the API
API: active pedagogical ingredient : Heutagogy . Alvin Toffler had emphasised learning, unlearning and re-learning. There has to be a shift to personalisation of learning instead of the present homogenisation. While implementing the pedagogical model, we must align it to the Directive Principles in our Constitution stated at Article 51A(h) to develop the scientific temper, humanism and the spirit of inquiry and reform.
We need to develop a course on introduction to Artificial Intelligence and related technologies that all University graduates should pursue. A course that we could describe as MSAI: making sense of Artificial Intelligence: an awareness course for all students…. like the first dose of the vaccination strategy, for the 4th Industrial Age.
The APJ Abdul Kalam Technical University led by Professors Vinay Pathak and Vineet Kansal have implemented this idea for all their students enrolled in Bachelor of Engineering across all branches / domain from Batch 2020. The course comprises Artificial Intelligence fundamentals with Natural Language Processing, Artificial neural Network, Robots, Speech recognition, and, exposure of Emerging technology including iot, virtual reality, 3D Printing & Drones, Cloud Computing. BlockChain.
The AKTU has also taken a lead by deploying a Chatbot to help students to get information on matters regarding the University. Arjun is the name of the Chatbot that is based on application of Artificial intelligence and gives basic information about university activities, persons, grievance, result etc.
With more experience, we may in future see dedicated Chatbots for VCs, Registrars, Deans and all faculty. This is both doable and desirable.
The attributes of an educated person has been changing over time. In the pre-computer age, it was largely about memorisation, and moving upwards in the six levels of Bloom’s Taxonomy from the simple recall or recognition of facts, as the lowest level, through increasingly more complex and abstract mental levels, to the highest order which is classified as evaluation. The intermediate levels are : comprehension, application, analysis and synthesis. In the age of computers, an educated person was expected to automate tasks that could be automated with the right algorithm and programming language. In this age of learning algorithms and machine learning, it is about being able to perform tasks using appropriately trained machine learning models. There was a big leap from the techniques used in IBM Deep Blue to defeat the chess champion Gary Kasparov in chess to that used in AlphaGo zero to defeat Lee See-dol.
The most recent breakthrough in artificial intelligence, Generative Pre-trained Transformer could prove just as disruptive and revolutionary to the education sector. So far-reaching are GPT-3’s implications for education, and educators ought to make themselves aware of this development. GPT-3 has 175 billion parameters. It was ‘trained’ on an unfathomable quantity of text data from the internet – to illustrate, all of Wikipedia’s six million English articles comprise just 0.6% of GPT-3’s 45TB training data.
A GPT-3-driven app called “ Learn from anyone” prompts the learner to enter the name of any well-known figure to be the ‘teacher’, and to answer queries. Students can thus ‘learn philosophy from Aristotle’ or ‘learn about rockets from Elon Musk’: the system churns out detailed, clear, accurate answers to questions in the style of the chosen teacher, and the student can ask follow-up questions or request alternatives if they’re unsatisfied with the first answer.
In January 2021, Google’s new trillion-parameter AI language model is almost 6 times bigger than GPT-3. It is a massive one trillion-parameter transformer system.
Because GPT-3 can “generate content which human evaluators have difficulty distinguishing from content created by humans,”GPT-3 has the “potential to advance both the beneficial and harmful applications of language models.” which include “misinformation,spam, phishing, abuse of legal processes, academic dishonesty and because of these dangers, there is need to do research and develop methodologies for risk mitigation, guide the development of ethical solutions that are free from bias arising from both the data and the training models.
The challenge for higher education today is to prepare graduates who can do more useful and valuable tasks than AI alone, a rather challenging task to say the least. It is the synergy of well educated humans and the best machine learning models that will address the really wicked challenges the world faces today, such as climate change, health, education, greater equity and better governance.