Earlier this month Econsultancy published an article about why automated content will never replace human writers. The article, ‘Automated content is a thing – but should it be?’. The article was written by Magnus Linklater of Bespoke Digital, which calls itself a digital marketing consultancy. While we agree mostly with Linklater, he has made some assumptions in his post that are not entirely true.
We can begin with where we agree. Linklater writes, “The problem with all of this is obvious: an algorithm simply cannot replace a human being when it comes to tone of voice, empathy and an innate understanding of an audience’s needs. Well, not yet.”
True. We are under no illusion that any text we produce is going to win a Nobel or Pulitzer prize. It would be nice if we did but, let us be honest, that is not going to happen here anytime soon. Robot journalism does have its limitations: our ‘robots’ cannot conduct interviews and are limited to the data pool available to them.
In our view, Linklater makes two mistaken assumptions.
The first is that all content needs a human touch. This is just wrong. Some texts just need facts. If, for example, a financial news service just wanted to list the performance of three stocks on a given day, a reporter could sit at their desk; look up the information; calculate the changes in stock price; compare those changes to previous days, weeks, and months; and then sit and write a report.
That may look something like this, “The day’s best-performing stocks were all in the XXX industry, which has seen an average gain per stock in the past two weeks of x per cent. Those gaining the most today were A, B, and C, which saw individual increases of X, Y, and Z per cent. The rest of the market, however, has seen a slight fall on average of x per cent across individual stocks.”
The function of that sentence is to inform, not entertain. And so it does not need to be well-written (although it would be nice if it was). It is therefore a passage that can be easily automated, especially since it deals in facts and figures. In fact, we have a project now producing similar texts at lightning speed.
This brings us to the second assumption. Linklater thinks that speed is not an issue. It is, especially in news. To write the passage above might take a reporter thirty-to-forty-five minutes. Our natural language generation system, the rtr textengine, can spit out multiple reports like this PER SECOND. Here, it is not analysis that is most important, but speed. And computers beat humans on this every time.
There are other advantages to automation. One of these is the personalising of news. To go back to a stock market example, it means that texts can be tailored to the reader. So if the report is being generated for someone who works in the automotive industry, the system could give them news about the big players in that market rather than include other non-related sectors such as pharmaceuticals and foodstuffs. Automation means that a greater variety of subjects can be covered.
But what the author of the article does not see is where computers surpass humans. Our systems are able to calculate data and spot anomalies in one split-second, and produce a report in the next. Humans cannot do that.
The supposition that machine writers will replace human writers is a false one. Sure, they will replace them in some aspects of their work but that is just the nature of machines. But they are not going to be able to replace them completely. Instead, our machines will buttress the work already done by humans, taking over the heavy lifting that can be easily automated. Think of it like wearing glasses–when you first get them, you do not stop using your eyes. Instead, they act as a support to an everyday function.
Automation is not a hypothetical; it is already here. One statistic says nearly four-out-of-five companies have been using content automation for the last two years. The Los Angeles Times, for examples, automates its earthquake reports. This is not the future. It is now. But they will not kill industries much like vending machines did not kill off corner shops.
But where should the author be happy? Well, this response was written by a human as we do not have the capability to have our robots engage in a dialogue. So, in that respect, professional writers–your jobs are, and will remain, safe.