One of the best inventions of the mankind is artificial intelligence (AI). Scientists and experts have collaborated with tech giants and trying to apply AI in almost every field. The development of machine learning algorithm that trains itself over the period of time and gets smarter with a huge database has been key focus of many companies. Though it is the fact that humans are irreplaceable in some cases; but replacing mundane and low-skilled jobs with AI has become possible. The AI has advanced so much that it has been used in healthcare, banking, and many other sectors. In the information age, it has become very difficult to consume a vast amount of knowledge generated by the world. The fast pace of life has made people difficult to read entire documents, news stories, and lengthy articles. Summarization of these documents and informative stories could save a lot of time. AI has stepped up for the job.
Researchers at Salesforce have been developing an algorithm that can summarize lengthy articles and documents. The algorithm uses machine learning tricks to summarize long pieces of articles. Though the development of such software is in initial stage, the software by Salesforce has been far better than any other algorithms developed yet.
Richard Socher, chief scientist at Salesforce, said, “I don’t think I’ve ever seen such a large improvement in any [natural-language-processing] task.”
The software is in initial stage and has a long way to go in terms of matching ability of humans of summarizing documents. Summarization requires commonsense and mastery of language. This task becomes more difficult as higher level of intelligence and logical thinking are necessities of the task. Parsing language is one of the toughest challenges for AI. But this challenge holds the greatest market potential. Personal computing can be transformed if AI possesses an ability to parse written queries and respond in coherent ways.
Caiming Xiong, a research scientist at Salesforce acknowledged that there are imperfections in the algorithm. But it has ability to summarize news and articles along with offering context of customer emails. This algorithm can be used to gain insights of emails on Salesforce’s own platform. Reinforcement learning has been used in this algorithm to train computers and obtain results. Various companies are also working on applying reinforcement learning in their algorithms to improve functionality of their systems.
Kristian Hammond, a professor at Northwestern University, appreciates the advancement in the research by Salesforce. He outlined that their research also highlighted the limits of relying entirely on statistical machine learning. Hammond said, “At some point, we have to admit that we need a little bit of semantics and a little bit of syntactic knowledge in these systems in order for them to be fluid and fluent.”
Improvement of language skills of computers is essential for advancement in AI. Maluuba, a startup acquired by Microsoft in early 2017, has been working on system that asks questions from the text. The team at the startup firm has been using a combination of reinforcement learning and supervised learning. Adam Trischler, senior research scientist at Maluuba outlined that it is important to develop inquisitive machines to make them intelligent.
The application of AI to summarize will be essential in determining fake news and hate speech for social media platforms and other websites asking users to contribute content. Moreover, offering insights on huge amount of information would be helpful in fields such as medicine, law, scientific research, and healthcare.