This article summarizes an article titled “AI’s Next Great Challenge: Understanding the Nuances of Language”. To refer full article click here
“Will AI ever be able to master the nuances and subtleties of human language”-This has been one of the most common and frequently asked questions for the past few years among the several developers community throughout the globe. The rate of progress in the field of artificial intelligence has increased by considerable amount in the past few years. Ground breaking creations like ‘ALEXA’ and the robot ‘SOFIA’ are few of the many instances of AI.
Through different tests and observations, scientists have claimed that the self-automated machines still lack the ability to have natural conversations and be more intuitive in replying to a question and take things less literally. Natural Language Processing has been the primary focus of the scientists for many years. In this process Scientists have pointed out three major areas which needs thorough examination and advancement:-
- Sentiment Analysis
- Question Answering
- Joint Multi-task Learning
- Sentiment Analysis :-
A word or a sentence can be interpreted differently or can express different feelings under different circumstances. Sentiment Analysis deals with understanding the different perspectives of a word or sentence and this can be used in AI to express and understand the intentions of the speaker and how they feel.
Natural Language Processing has helped in the advancement of sentiment analysis although it needs more and more attention in the years to come. Real-life example of sentiment analysis in AI is found in ‘Google Assistant’ , ’Siri’ which understand and identify whether a statement is happy or rude or sad and produce a relevant response.
- Question Answering:-
Although Natural Language Processing has helped machines understand the meaning of the sentence, it is seen that these AI oriented machines and devices have failed to produce a desired and relevant response in context to the questions asked. Digital assistants like ‘Cortana’ and ‘Siri’
are tested to have around 40 IQ which is not a bad number for a digital assistant. However, the accuracy of the response provided greatly depends on the manner that you ask the question…if you ask in a more formal manner, it will provide a more accurate response.
- Joint Multi-Task Learning:-
One of the many targets that scientists want to accomplish in the field of machine learning is Multi-Task Learning in which an AI model is taught to learn to do new tasks and integrate new tasks along with the old ones in order to solve complex problems.
Several machine learning models and algorithms like ‘Linear-regression’, ‘Decision Tree ‘ can be used to classify data from a given data set, develop neural networks and make desired predictions.
Though, Natural Language Processing and repeated studies have immensely improved the way of interaction with machines, it is still a tremendous challenge which the scientists have undertaken in order to make machines understand and process information with increased accuracy.