Approximately 90% of data in the world has been generated within just the last two years. But what is the point of all this data if we cannot process it? “This the challenge for data scientists today,” said Prof. Nirmalie Wiratunga at the at the recently held Big Data Forum for Data Enthusiasts organized by Informatics Institute of Technology.
In her presentation at the event, Prof. Nirmalie explained that data scientists have to go through vast amounts of data and then learn how to process it. But it’s not just about figuring out what data is relevant or obsolete, but rather to figure out how to curate it.
In digital health, for example, a person can generate a Million GB of data throughout his/her lifetime. That’s the equivalent of 300 million books. The goal here is to map out a person’s health and see how one person’s data can help prolong the health of millions of other people.
The procedure is not yet perfect. The algorithms can go wrong. But that all depends on the data that is fed to the algorithm. This is especially true with facial recognition, which is becoming available almost everywhere. By focusing on small areas, we can make algorithms more computationally feasible and thus, improve the accuracy of the algorithm.
Speech recognition is another area that AI and Big Data scientists are working on. It’s not just about understanding the words spoken by a human to machine though. The machine would have to string the words together, understand the context of what they’re said in and then offer a meaningful and related reply. Chatbots and AI assistants such as Siri and Google Assistant are prime examples of this.
The goal of AI is to make our lives easier. This was the opening statement of Dr. Ruvan Weerasinghe In addition, some also believe that the goal of AI is to create intelligence at the level or surpassing that of humans. AI is not going away. It’s here to stay and it will change a lot of things.
To excel in any field, you need to have a healthy content for it. If not, you cannot excel in it. It is up to those who are in the fields of AI and big data to create content for it and carry it forward. We ourselves attended Sri Lanka’s first AI Summit and learned quite a lot about AI, machine learning and big data
What is AI?
Before the year 2000, this could have been explained as a computer doing something better than a human. After the year 2000, the definition of AI was changed to where AI is an intelligence that is on par or beyond that of humans. Samuel’s checkers’ program is a perfect example of this. Created by Arthur Samuel, this was a Checkers game that had no knowledge of Checkers. But over time, it learned and got better.
From there, people also idolized AI as the one-stop thing that could end all their problems. People are now focusing on using AI to understand what they say and to get machines to respond accordingly. They are also utilizing technologies such as image recognition so that a machine can understand the components and context of an image and make decisions accordingly. An example of this would be smart security cameras where it can detect shoplifters from the way they move.
In the postmodern era that we’re in now, learning has become a dominant underlying technology. It is applied across a number of various fields and genres such as personalized recommendation, sentiment analysis and web search. One of the main showcases of this is from IBM Watson when it played Jeopardy.
What drives AI?
Quite simply, it’s machine learning. Rather than encoding knowledge to solve problems, machine learning deals with techniques that enable a computer to learn by itself. With deep learning, for example, a machine can not only learn to predict but also learn what features it needs to predict.
But what if there’s no answer? What if there are clues instead? Well, that’s where reinforcement learning comes into play. It’s akin to learning with rewards. You can also learn from a task that has been already completed. This is called Transfer learning.
Where are AI and big data headed?
Now the focus is on general AI. The goal here is not only to solve a particular problem but to develop a general intelligence or artificial general intelligence. Everything we know about knowledge and intelligence will no longer hold with the advancements of AI and machine learning.
Take self-driving vehicles, for example. You wouldn’t want it to crash, would you? Achieving 100% is a major challenge though. In time to come, even our cities would become smarter. This would enable us to control or eradicate pollution. Imagine the next time you went to the doctor, a machine goes over your historical data and gives you the best cure? Well, that is possible with AI.
Does this also mean that Skynet and the Terminator are possible? Well, yes they are, explained Dr. Dilshan Silva – Head of Analytics at Nations Trust Bank. There are a few issues with AI. These range from protecting ourselves from AI and the biggest question of all: should we stay in control of AI?
Be it chatbots, to rankings on what you like, to even showing ads for items you were just browsing, AI and machine learning are aiming to make lives easier. But at the same, time, they are also affecting our privacy.
There is also a need for laws and ethics in AI to catch up to where we are. We need to think of these very carefully. But what should these new ethics be about? Well, that has to be taught by philosophy. So once again, it’s up to those studying in fields of AI and big data to pioneer and answers these questions.