This content was not written by a robot. This is a warning that should perhaps be placed above every publication in the future. Last week, Morningstar announced that it would begin using robots to write investment research. Morningstar evaluates mutual funds using two different ratings: quantitative by stars and the Morningstar Analyst Rating (MAR), a qualitative rating based on content analysis by a forward-looking fund analyst. This fund analyst is now partly replaced by a robot. If there is no MAR available, a Morningstar Analyst Rating (MQR) follows. This has the same qualification of Gold, Silver, Bronze, Neutral or Negative, but with a tiny Q behind it. A quantitative rating is packaged as a qualitative rating. In real qualitative research, you find out in-depth information about people’s motivations, thoughts, and expectations. Qualitative research is about words, not numbers. It requires in-depth interviews where the focus is on finding out the answers to ‘why’ questions. The analyst asks more questions when there is a lot to say about a certain subject. With quantitative research, you compare numerical insights. It answers questions that can be expressed in quantities. So as soon as a Q is added to the qualification Gold, Silver, Bronze, Neutral or Negative, the reader is dealing with a quantitative judgment in addition to the well-known quantitative star judgment.
Now I have some experience with research generated by robots. If you spend half an hour reading such texts, you soon see patterns of sentences that keep recurring. There was an attempt to introduce some variation, but it was always traced back to the results of quantitative research. As a reader, you feel that you are not taken seriously anymore. Now, this is an experience from a few years ago. Artificial intelligence has developed exponentially since then. It will now be much harder to tell whether a text has been generated by a robot or a human. Yet it is not the case that such content has taken over the Internet. We do live in the era of fake news, but that is news that has not been written by robots. Still, there is a great temptation to let robots write texts. The content of those texts is used by Google search engines to find websites more easily. More content quickly means that you will be found more often and a robot will continue to write. Nevertheless, Google’s web crawlers are perfectly capable of recognising robot-generated text. A robot is used to identify another robot. Brave New World.
The characteristic of robot-generated texts is that they lack substance, literally substantial content. The robots train themselves by reading millions of texts, more than a human can ever read alone, but they make no attempt to interpret the text and thereby offer a new insight. Ultimately, the text depends on the command a human gives the robot. As such, a robot can be used to answer questions, supplement text, make the text understandable or even summarise the text. However, do not count on an original finding, something that distinguishes excellent research from mediocre research. New insights found by thinking outside the box. That is something a robot cannot do, simply because there are too many paths. Humans continue to play a vital role in an acceptable outcome. Even a living person who has to write a text needs more than just a title, to begin with. They either direct the text themselves or are directed in a certain direction by the client. There are plenty of people who can be labeled as walking encyclopedias and in terms of knowledge, they are most comparable to robots. This knowledge gets in the way when they have to write an original text. After all, too much knowledge quickly leads to conformism and that is the enemy of creativity. It leads to a greater fear of coming up with a successful original idea, often only to be told afterward ‘I had thought of that too but didn’t say it out loud’. Wharton’s research also shows that it is better to play a game first than to start writing immediately. The delay gives the brain time to think more freely. Now, it is not the case that a robot will replace all Morningstar analysts. The 130 analysts will continue to analyse the largest funds. Morningstar itself admits that humans are needed to add colour to a story, something that Morningstar says is impossible using a quantitative process. To avoid wasting readers’ time, however, my advice is to include a Q at the beginning of the text to indicate whether or not it was generated by robots.