march 17, 2014 Quakebot, the software developed by journalist Ken Schwencke has in fact thrown down from the bed his creator at 6.25 in the morning because rivedesse the piece on the earthquake of degree 4.7 with the epicenter in Beverly Hills. The total time of data collection, writing and inserting in the editorial system Quakebot was less than 15 seconds thanks to the access to the information on the Usgs (the homologue of the Ingv Italian). To Schwencke is enough to re-read the piece and press "enter" to be the first on the news, less than 5 minutes after the shock and without errors or inaccuracies. The earthquakes are not the only sector in which a lot of media, supported by a growing number of stratup, are betting on algorithms and a utomation to reduce time and costs of the information. Last August, Thompson Reuters, specializing in financial information, has announced an alliance with the startup california Graphiq, specialized in semantic analysis and visualization to automate the production of interactive graphics to accompany his analysis of the market. The american agency is only the latest to point in this direction and there is good reason to believe that he is right the computer scientist Kris Hammond, who has wagered that by 2025, 90% of the news read by the general public will be produced by a computer with little or no human contribution. Hammond, in fact, knows what he is talking about. His Narrative Sciences, which launched in 2010, to date, has raised over 10 million dollars of investment and its software Quill produces more than a year of articles and short throws on a quarterly basis, stock exchange indices and the other to Forbes. Associated Press – the u.s. agency supported by 1400 newspaper s in the u.s. and thousands of broadcasters around the world – has followed a similar route by choosing the software Wordsmith developed by the american company Automated Insights to accelerate the publication of the analysis of the stock exchange and the results of the quarterly and is considering to apply the same approach to the sports results.
The playing field perfect for the systems of IA, is, in fact, any industry in which there are data is plentiful and structured in a precise manner, maybe from other machines. But things are changing quickly. The spread of the network and the cloud systems are also, by multiplying the sources of data and the possibility to "train" the software on different terrain. Furthermore, the replacement of neurons with the bits, until a few years ago, a game essentially american, is taking hold on this side of the Atlantic.
The Financial Times of London, Emma, the software of the newly formed Stealth Shaunak Khire, last spring, he ripped his colleagues in the flesh, and the bones of the editorial, financial, which, however, does not seem to be worried about because they are learning how to handle his pieces, titolandoli and making them more interesting for a human reader. In France, The Team is developing a series of bots to do the chronicle of social games of rugby after the success of similar systems for football, while some of the big tv channels are looking to the software of the automated production of video as Wibbitz to get video in real-time and low-cost. The automation of video production, increasingly demanded by online channels, is one of the most interesting new frontiers of automation of editorial and many people look to platforms like that of the Ban.jo to David Patton, who raccogkie automatically public content, and geotagged from Instagram, Twitter, Facebook, GooglePlus, Vcontake and others, to produce in just a few seconds video of an event in the news. The leap into the world of the IA editorial, however, is not free of pitfalls and is not sudden. Tribune Publishing, at the beginning of the summer had announced plans to increase to 50 times the volume of its production of videos thanks to the Ia, but it has been found that users abandon without delay the video automated after the first few seconds.
© REPRODUCTION RESERVED
No comments:
Post a Comment