“Post-Covid-19, AI will promote 24×7 operations in factories”
Dr Sriparna Saha, Associate Professor, Department of Computer Science and Engineering, IIT Patna, feels that proliferation of artificial intelligence (AI) in the future, will help move humans out of monotonous jobs, but they will have to learn to handle robots.
TPCI: In your view, how will the onset of Covid-19 and the emphasis on social distancing impact the application of AI and machine learning in industry?
Dr Sriparna Saha: Precisely, AI and ML can surely help industries operation by automating the processes. In lieu of human labours, if bots of varying nature could be deployed with tasks or steps could be automated, there is a possibility of negligible or no effect of Covid-19 (social distancing). However, because of manual intervention in industries, the impact of social distancing is huge.
Usage of robots in industry can reduce the impact of Covid-19.The current crisis has motivated the development of new technologies across all aspects of life, from e-commerce to remote-working and e-learning tools, including Alibaba’s DingTalk, WeChat Work, and Tencent Meeting. Thus, the focus of many industries is now the introduction of “contactless delivery” scheme and implementation of this requires significant usage of AI/ML based techniques.
TPCI: Please cite some use cases where AI can take over some work roles in major manufacturing/services industries in the immediate future.
Dr Sriparna Saha: Services industries can surely benefit from the use of AI. Chat-bots have a great impact when it comes to automating several sectors of service industries like customer-care, complaint-mining etc. just to name a few. Chat bots are popularly used in several different domains like healthcare, banking, insurance companies, education sector, online purchases, real estate and many more.
For e.g., Florence is a popular chat-bot in the healthcare domain. It can help patients to remember about taking their medicines, checking symptoms, providing more information about a particular disease, and help in finding a doctor as well.
COIN is a chat-bot used by JP Morgan Chase. This has been developed to manage the back office operations. The bot is efficiently analyzing complex contracts (much faster than human lawyers).
TPCI: What are the current barriers to the adoption of these applications of AI in the industry?
Dr Sriparna Saha: The automation systems are not fully accurate, therefore intensive tasks cannot be automated. Only those tasks, which follow some algorithms/predefined steps can be automated. Moreover, any automated system can fail anytime; so a backup plan must be in place to handle this situation.
From a technological perspective, one of the greatest challenges to AI is the lack of good data. Despite their large and growing stores of Big Data, many tech leaders say that they don’t have enough of the kind of data they need to support their AI efforts. Also, creating and training models requires huge amounts of data, as well as very fast systems. High-performance computing systems are very expensive, which drives up the costs of deploying AI. Thus, lack of IT infrastructure also adds to the list of challenges faced in the adoption of AI.
TPCI: In what ways can AI help medical teams in prediction, detection and cure of the Covid-19 disease? What are the insights emerging in this area at present?
Dr Sriparna Saha: AI is heavily used in handling Covid-19 situations. Deep learning-based techniques can be utilized in differentiating the chest images of Covid-19 patients from normal pneumonia patients. From the chest images, AI-based techniques will be able to predict the stage of Covid-19.
AI-based techniques can be utilized in designing appropriate drugs for COVID-19 without compromising on the quality. Natural Language processing (machine learning applied to text) techniques can be applied in building biomedical knowledge graphs, which illustrate the relationship between different biological entities (such as drugs and proteins) generated after processing several scientific articles. This knowledge graph can be applied on COVID-19 to determine the connection between the virus and the potential drug candidate Baricitinib.
AI-based forecasting techniques can be applied to forecast the spread of COVID-19 throughout the world. These forecasts may vary from short-term predictions to long-term predictions. The short-term predictions include next day estimates of COVID-19 cases like the number of infections, fatalities, and recoveries in many infected regions. The long-term prediction is more focused on the estimation of the case fatality and case recovery ratios after one or two months. We can also try to predict the recurrence of COVID-19 spread. Mathematical modeling techniques can be used to predict the optimal lock-down period.
There are a few researches that attempt to discover novel compounds for usage in targeting SARSCov-2. In this scenario, we can contribute in facilitating the detection and analysis of COVID-19 in the molecular scale by utilizing AI techniques. Here, the molecular scale includes the drug discovery-related research, predictions of structure and foldings of proteins associated with SARS-CoV-2, the virus that causes COVID-19. In this regard, we can analyze COVID-19 at the molecular level by utilizing various ML techniques along with various DL techniques.
TPCI: How is the possibility of increasing emphasis on AI in the post-COVID world expected to transform the employment landscape in the coming years in your opinion?
Dr Sriparna Saha: Usage of AI in the post-Covid world will surely help employees to avoid the problem of monotonicity. Advanced robots capable of recognizing objects and handling tasks where previously manual intervention was required will promote the operation of factories and other facilities 24/7, in more locations and with little added cost. The consumption patterns of different locations can be easily detected with the use of machine learning based techniques that can help in determining locations of the industries. Moreover, personalized services to online customers by different service industries can only be implemented with the use of AI/ML. The feedback received from different customers can be easily incorporated in the AI models, which in turn can help in increasing user satisfaction.
Thus, more jobs will be created for maintaining these AI-based tools. So, while human expertise will no longer be required for performing monotonous jobs, humans have to gain more expertise in building these automated robots.
TPCI: What will be the major focus areas of investment and innovation with regard to industrial applications of AI for businesses in the coming five years? What key transformations can we expect?
Dr Sriparna Saha: Healthcare systems should be supported heavily by the usage of AI and ML-based tools. The prediction and forecasting systems developed on the basis of AI and ML can help the doctors in better diagnosis of different diseases. The prognosis systems built using DL-based techniques can help in predicting the number of survival years of a cancer patient. This can help medical practitioners to decide whether complex treatments are required to be given or not. Evidence-based medicine is another important concept in the field of AI-based medication. Moreover, chat-bots can be developed as counselors for the depressed people.
The banking industry could be highly benefited with the use of chat-bots. Personalized banking can be implemented easily with the help of chat-bots, this can in turn help in improving customer satisfaction and engagement. Frequent banking queries like account balance, bank statements, transfer funds, creating a deposit, saving and investment advice, and so on can be answered by the chat-bot.
Education sector can also be improved with the usage of chat-bots. Chatbots can help students in learning new languages, giving feedback, professor assessment, essay scoring, acquainting a student with school culture and for administrative formalities. Thus, usage of chat-bot in the education sector will help build a personalized education system.
Dr. Sriparna Saha received her M.Tech and Ph.D. degrees in computer science from Indian Statistical Institute Kolkata, India, in 2005 and 2011, respectively. She is currently an Associate Professor in the Department of Computer Science and Engineering, Indian Institute of Technology Patna, India, and is also serving as Associate Dean Research and Development of IIT Patna. She has authored or coauthored more than 200 research papers. Her current research interests include machine learning, text mining, multiobjective optimization, and bioinformatics. Her h-index is 26 and the total citation count of her papers is 3745 (according to Google scholar). More details are available at :www.iitp.ac.in/~sriparna