“Robots may be helpful to industry, only if used selectively”

Anupam Basu, Director, National Institute of Technology, Durgapur and Professor (on lien), Department of Computer Science & Engineering, IIT Kharagpur feels that robots can play a huge role in reviving the industry post-Covid-19, particularly to ensure social distancing, perform mundane tasks with high efficiency and accuracy and also to replace humans in precarious environments.

TPCI: The workplace is expected to witness a number of changes in the Covid-19 era, particularly the emphasis on social distancing and minimum human intervention. How is this expected to propel the adoption of AI across industries?

Prof Anupam Basu: There are two aspects to this question, because Covid-19 itself will bring in a lot of unemployment. From that point of view, I think the use of AI should be made very judiciously. I would vote for deployment of AI without displacement of manpower. We can utilise the manpower and have much better services given and enhance productivity using AI and other technologies and have new revenue generation paths.

AI can help in numerous ways. First, acquiring and analysing data will be of primary importance in the days to come; not only the pandemic and virological data but also the data of the business, agriculture, infrastructure and human behaviour. We will see that once the restart button is pressed (the reset button has already been pressed), and we start again, the nature of the demand and market may also change. We cannot continue the way we have done in the past. AI will be important in identifying the recent market trends and hotspots that have the biggest opportunities.

We will need the help of technology, including AI, to understand what is the trend and what people are looking for – basically the common people. This is because the market now will be for the common people and I think that industries will have to tune to them. Till now, consumer demands were mostly generated as there was always an affluent market to sustain the demand. Now possibly, it will have to be bottom up and discovery of what is needed will be more important. As we will be going ahead, we will need to analyse logs of business transactions, people, tweets, social networks, articles, etc. To reach to a larger segment, the common mass, AI-based language tools will be important, and so will be data analytics.

TPCI: What are the industry segments where you see potential for immediate deployment of AI and robotics?

Prof Anupam Basu: If selectively used, robots will be helpful, since we have to maintain production, maintaining social distancing. Robots will also be helpful in carrying out routine jobs under precarious conditions, such as handling bio-hazardous material. I foresee that in the automobile sector there will be more demand for low cost electric vehicles, with more cyber control and safety measures, where AI will be of importance.

As social distancing is there to stay at least for a while, caregivers for the aged and the disabled must be assisted with special intelligent tools.

In healthcare, robots can be used to perform routine monitoring tasks including sensory data collection and health parameter check ups, when a regular doctor or paramedic is inaccessible.

Education sector will open up a lot of opportunities. Stable and affordable internet to every household, intelligent management of bandwidth sharing, multilingual content creation, machine translation, speech to text and text to speech technologies should gain more importance in near short term, in India.

AI should also play a role for creating an inclusive society where the need of every socio-cultural and socio-economic clusters can be properly identified and addressed.

One of the primary reasons why this has not been adopted aggressively in India, is because large-scale deployment of robots and AI will create lot of unemployment. However, I believe if it is judiciously used, additional revenues can be generated, which in turn must be circulated to welfare for the deprived.

However, I would like to raise an important cautionary flag to the use of AI. Presently, machine learning is at the core of the modern AI systems. Such algorithms are non-transparent as of date. Consequently, bias can be inbuilt, which raises ethical questions. It should be very strongly audited.

TPCI: How can AI help in controlling the spread and helping in treatment of the Covid-19 infection?

Prof Anupam Basu: The COVID-19 infection is new to us, but there is a company in Canada, Bluedot, which predicted the onset of this pandemic well in advance of WHO. They used machine learning and data analytics to identify a cluster of “unusual pneumonia cases” happening around a market in Wuhan, China and flagged it.

When it comes to data analytics and machine learning, it opens up a big opportunity in the Indian context – in the healthcare system. The Covid-19 pandemic has exposed the limitation of our health infrastructure. In our hospitals, we face shortage of both infrastructure and manpower. Addressing this deficit will take time. Artificial intelligence can play a major role in taking off the routine workload from doctors.

This is from the days of Good Old Fashioned AI days in the ‘80s, when the expert system MYCIN showed that the initial screening and treatment of bacterial infections could be handled by a paramedic with the help of a computer. Only the specific cases would be forwarded to the doctor, and the workload of the doctors would be reduced. But this has not been adopted in India so far. In a country where government hospitals are flooded with people, we can use a mix of AI and human intervention. I think this will be beneficial. The doctors often get tied up with routine work.

However, it is important to state that such systems should be developed with enough rigour. We carry out a lot of research and feel shy of the last mile translation as that is often denied credit. Industry must make long-term investments in these areas and make them reach the ground.

TPCI: You mentioned that AI must not impact the employment situation. What kind of unique roadmap for the deployment of AI in India?

Prof Anupam Basu: We have enough manpower, but there are a few shortcomings of human beings. Two major shortcomings are getting tired and possibility of what we call ‘human error’. Thirdly, there are some dangerous surroundings where we would not want human beings entering and getting affected.

For example, there could be a routine activity that is going on and on. Human beings may be doing the routine work to some extent, but beyond that we can take the help of robots. The onus should be on the person – when he wants to get himself involved and when he wants something to be done by robots. The primary objective being non-layoff, we can multiply the hours of production by a mix of machines and humans and generate more revenue.

It is like a teacher asking a student to write something for her, while she does something else. She then checks it later. The point is that it is not an activity that she could not do, but the student is assisting her and taking out her labour. However, she is controlling the process and outcome. The approach should be to use robots to assist people and not to replace them.

The second area is where we can make errors. For example, there could be an old doctor who is very experienced and knowledgeable, but his hand is no longer steady. He knows exactly what to do, and he can make the robot do it in an error free manner. These are the critical points where a robot should come in.

I would like to add another thing – providing information to the people. Another important aspect of AI is understanding people – their strengths , weaknesses, etc. For example in the field of online education, AI can analyse an answer provided by a student to see where he is lagging – spelling, calculation, etc and teachers can tune the content in accordance with the needs of the student. So far, in online education we are approaching the students as a community or a batch, where everyone is treated in a uniform manner, but the same size does not fit all.

TPCI: Where do you see Indian innovation in AI in the coming years, and what kind of transformations we can expect in the coming decade?

Prof Anupam Basu: There is an essential problem to be understood. A number of hackathons are held across the country. Even if 2% of them are useful, how many are actually taken forward for utilisation in the industry? There are very few that are taken forward to be made a product. My point is we have to be more cognisant of the needs of our country in the innovation process rather than following the West, and translate the innovations to affordable products.

In the midst of Covid-19, I can see a lot of Indian students and faculty attempting a lot of innovation. Some really fantastic products are coming up. We do not have an effective system in place that can help in translation to products and take it to the people even at this hour of need. But we never did it until our back was to the wall.

So I think that the Covid-19 episode is also an eye opener. India should now think of the real demands of the people. The demands are often generated by the industry that makes sophisticated feature rich systems and a segment of the population will consume them. But it can be bottom up as well, where we identify latent demands for a better life for all, and that can be solved by proper innovation and industrial participation.

Innovation should be promoted not by words but by deeds. Innovators should be connected to the industry and new products will come. That is very important in general. AI is of course a part to make the tools more adaptive, flexible and powerful. 

---

Anupam Basu is presently Director, National Institute of Technology, Durgapur and professor at Department of Computer Science Engineering in IIT Kharagpur, India. He has previously taught at University of California, Irvine and University of Guelph, Ontario, Canada as a visiting Professor and has been the Humboldt Fellow at TU Dortmund, Germany. He teaches Computer Architecture, Operating Systems and Embedded Systems and has also taken Artificial Intelligence, Intelligent Systems and Cognitive Science courses. His research interests are Embedded Systems, Natural Language Processing, Assistive Technology, Natural Interface Design, Human Computer Interaction and ICT4D.