There is a new range of software out there that is claiming to be almost human in the way it writes. They tend to work on an artificial basis to cherry-pick elements from a wordset and then use them to create what they perceive to be a human-sounding product. The various offers claim they can use emotional language, adjust diction and play with syntax to convince you.
The evidence seems to show that these platforms can indeed generate articles driven by data at a speed that even the fastest author would find hard to match. A wide range of companies are already using versions of the software including Yahoo that uses it for Fantasy Sport, and Associated Press that uses it to generate a large part of its US corporate earnings stories. The essential approach being that the machines should be allowed to create the simple stuff that has little added value.
More recently these firms have started offering a self-service option where you can upload your own data and article templates. Content does however still have to be data-driven but it is improving all the time.
What seems to be the problem is educating the market about the potential of these solutions and to allay the fears that if a machine can write it then who needs writers anymore. It still seems to hold that true journalism is not just about providing information – it’s about putting it into the context.
It seems that it the ultimate ‘why’ of the story that is the missing piece of the puzzle for AI.