A report about an earthquake was published by the Los Angeles Times three minutes after it happened back in 2014. A staffer had developed a bot (a software robot) called Quakebot because of which this feat was possible.
Software companies like Narrative Science provide articles which are produced by most of the Natural Language Generation (NLG) tools. In-house versions are developed by many media organizations. Nearly a third of the content published by Bloomberg is generated by a system called Cyborg, the Washington Post has Heliograf, and the BBC has Juicer. Graphs, tables, and spreadsheets are the data with which these systems start.
The basis of a narrative is formed by facts that are analyzed and extracted by these systems. using natural language generation software, they generate a plan for the article, and finally, they craft sentences. Highly structured data is available as an input for the articles to be produced by these systems. It includes a video of a football match or spreadsheet data from a company’s annual return. With flair, imagination, or in-depth analysis, they cannot write articles. They have not rendered thousands of journalists redundant as a result.
The number of niche articles being written has been sharply increased by them instead. As per Kenn Cukier, senior editor at The Economist, it’s not about what is best for journalists but it’s about what is best for the public. Journalists shouldn’t resist this either as they didn’t cling to the quill in the age of the typewriter. It wouldn’t be cost-effective to reach otherwise as it’s a scale play serving niche markets.
AI is used by Financial Times to spot trends within economies and markets. FT’s is looking at using other data sources and its content lends itself to this kind of analysis. FT reduces the bias within its output with the help of using bots. For the analysis of the gender ratio of the faces appearing in the paper, a system called Janetbot – named after Janet Yellen, former chair of the Federal Reserve. Its use had to be paused earlier this year but it was trained on an industry-wide sample of photos.
The sample was not racially diverse enough to detect gender reliably when it became apparent. FT is using data analytics to improve the way it sells like any well-run company that intends to remain in business. The same subscription offer as the CEO of a tech giant in Silicon Valley would receive for a student in a Birmingham bedsit during the past. Today, machine learning is used by the company to experiment with offering more appropriate packages to each.
To speed up and improve the production of market reports, Robotic Process Automation (RPA) can be used. As well as the handling of responses to customer inquiries and the fleshing out of customer databases.
