Learning Code
There is a buzz around artificial intelligence and machine learning and how these can be used as powerful tools
As driverless cars and robot teaching assistants are becoming a reality, Elizabeth Croft, Dean of Engineering, Monash University, Australia, which is known for its strong academic science education, decodes the uses of machine learning and artificial intelligence.
Know the difference
Machine learning (ML) is using large data set to optimise the relationship between inputs and outputs. Mathematical concepts find connections between the input and the output and for that, it uses automation and statistics and takes advantage of the computation power of graphics processing unit. If you look at the human brain, we have synapses that optimise the relationship between inputs or what we see and the output, which are our reactions. We have an input and an output and thus, there are billions of cells in our brain making those relationships. Machine learning uses the same optimisation in a similar way. Artificial Intelligence (AI), on the other hand, is about developing logical relationships between things. So, if you want to understand what cups are and the relationship between a cup, how you hold it and what goes into it, you will then extrapolate that logical relationship about what happens when you use a cup with what happens with a bottle. Although they are of different shapes and can be held differently, there are relationships between what you can put into both. Artificial intelligence can extrapolate knowledge about one thing and use it in another situation. Humans are very good at knowing how to operate in one situation, learn that, go to an entirely different environment, and adapt what they have learnt. While Machine learning can drive cars automatically but it fails when there are things that are out of context.
Opportunities for artificial intelligence
Algorithms are being developed fast so those who can step back from using one particular tool and come up with creative ways on how to use them in various contexts will have the best opportunities. We want people to be creative designers. We want them to be creators and not just coders; those people who can learn machine learning in new ways and come up with thoughtful and ethical ways around using it.
Skills to learn machine learning and artificial intelligence
A strong grounding in mathematics is essential and helpful to understand logic. People worry when they come to engineering, saying I haven’t done coding yet. But it’s about learning the rules, just as you would while learning a new language. If you can write a flow chart, you can code. People who are good designers are able to think about how to use it in another context, make it versatile and robust to different things and deal with exceptions. Logic and design are the skills needed.
Key areas of application
There are big opportunities for AI and ML in social sciences because there is so much data in libraries and policies. We can extract that data and learn from it in ways we haven’t done before. In healthcare, bioinfomatics is a burgeoning area and people with a background in microbiology and genetics will have great opportunities. Similarly, there is immense scope in material sciences. For example, in physiotherapy, while working with people who have had a stroke, engineers have been doing work in using ML to recognise how well someone is doing with an exercise by tracking their movement, how they are recovering, or changing exercise automatically to improve their recovery.
Artificial intelligence and India
AI has huge opportunities for India because of the scale. If you want to optimise city design, the transportation network or banking system which has massive scale, you need to have huge processing power. But the biggest challenge is to make sure that while designing these systems, we do not forget the rules of engagement and ethics.