Sophia is a citizen of Saudi Arabia. She is the first ever robot to receive a full citizenship of a Nation. She is a ROBOT. She is what all the futuristic Si-Fi movies show as a robot. She can create sixty-two different facial expressions and can interact with people as other people do. Sophia was made by Hanson Robotics, a Hong Kong-based engineering and robotics company. Sophia is just an example to illustrate the progress the world has made in the field of Artificial Intelligence. Apple Launched Siri a long time ago and ever since that people have started talking with there own phone. While the world is making such progress in the field of Artificial Intelligence, why isn't anyone from our country Nepal making any notable progress in the field of Artificial Intelligence? A simple answer to this question would be the lack of Data that we face in our country.
Artificial Intelligence is the intelligence displayed by the machines. The dated history of AI goes back to 1956 AD when John McCarthy coined the term "Artificial Intelligence" as the topic of the Dartmouth Conference. And since then there have been many developments in the field of AI. As the computational power of machines keeps on increasing, the tasks previously considered to require intelligence are often removed from the definition. This is known as AI effect and has lead to a saying "AI is whatever hasn't been done today". This great advancement in AI has been possible when the AI researchers changed their focus from AI algorithms to data as well. Previously the focus was much given to the algorithms being used but nowadays a great deal of focus is given to the data too. A computer can learn from the data we have and if we don't have enough data then no matter how better the algorithm we create, we can't get a good result. The use of AI in the field of Speech Recognition, Image Recognition, etc is due to the abundant data that we have. The huge data that we have has given rise to the new field of AI or Machine Learning called Deep Learning. All the AI-related things that we see around us is due to the abundant data that we have. Let's take the example of photo tagging on Facebook. Nowadays when we wanna tag someone in our photo, the suggestion pops up. This is because Facebook already has enough photo of many people and now it can use machine learning algorithms to just compare the photo to find the correct person but it wouldn't have been possible if there were not enough photo data available for Facebook.
While the whole world is making such big progress in the field of AI and people have already started excluding the tasks like Optical Character Recognition from artificial intelligence, then why don't we still have a good Nepali Optical Character Recognizer? The task that has been considered trivial around the globe is still not possible in our country. Why is this so? I got the answer to this question the hard way while working on two of my projects. The first project was a Nepali optical character recognizer and the second one was a Nepali speech recognizer.
Nepali optical character recognizer is a software that would take image input of the written Nepali text and give the textual form of it. It could be used to convert the old paper-based record to computer-based records. While working on this project, the algorithms that we used were state of the art algorithms, that are being used around the globe but our results were not as accurate as we desired. It was due to the lack of data that we had. We didn't have any source to get the image version of the written text and their textual form. We had to create the data ourself and with a small team, it was difficult for us to create a huge data repository. The second project that we did was Nepali speech recognizer, which takes Nepali speech as input and recognizes the input to give its textual form. The problem that we faced in this project was also the same. The lack of data and had to create our own data repository which led to poor accuracy. This problem of lack of data is a serious problem and until this problem is not solved, we can't expect a great deal of progress in the context of Artificial Intelligence in our country.
In other countries, the universities play a big role in solving this problem. The universities create a data repository from the data available from the previous projects and also invest their resources to create the data that is not available. This chain will always work as all the new researchers will create some new data and that will be added to the repository. The government also plays there part and makes data available for the students and public to use but there is no such practice in our country. For the progress in the field of AI in our country, it is high time that institutes like IOE Pulchowk Campus, Kathmandu University, etc focus in creating a data repository for AI-related projects and even the government supports this by making data that they have and can make available for students and public. Stock Market prediction is an example of AI project and has been successfully implemented in other countries but not much progress has been made in our country due to lack of the proper data. Stock Market data is a public data but is not available easily in the form required. AI researchers require the time series of stock market data that they need to feed into there algorithms so that the machine can make future predictions of the stock market but the data is not available in the way required and if also available its hard to get.
So for the field of Artificial Intelligence to develop in our country like it has done around the globe, it is a necessity that we have abundant data for AI-related research & projects and Educational Institutes & government should play there part to solve this problem.
© Copyrights Prabin Bhandari. All Rights Reserved