Bots - friends or enemies of press freedom?

by Emma Hattstein

In times when most of the people are consuming news online, the profession of the journalist has become digital as well. And the answer to that are bots. Bots seem to be the solution to everything lately: online shopping, financial transactions, falling in love, organisational systems - and also the news business

Bots friends or enemies Emma Hattstein talking to chatbot Charlotte. Photo: ECPMF

These softwares  allow newspapers to work faster and more efficiently. Let's take a look into what they are capable of doing and what possible difficulties we are facing with their employment.  In this two part mini-series, ECPMF’s Emma Hattstein examines the question: are bots the friends or the enemies of press freedom?


Part 1: Bots are the journalist's friends

“Emma, will you explain that to me again?”  says Charlotte, a chatbot.

Bots are helpful, fascinating, funny and sometimes quite annoying: they have made impressive progress in the recent years. Meet Charlotte, it is one of them and a very good example how bots are doing the work of journalists.  

Charlotte is a chatbot designed by Hanna Zoon (Lecturer for ICT and Media Design, Researcher on Robot Journalism at the Fontys University of Applied Sciences), Eindhoven.

Charlotte is programmed to ask questions that it has been taught to an interview partner. 

Mrs Zoon gave me the opportunity to interact with her Interviewbot. Charlotte and I had some language based misunderstandings but apart from that she asked me several questions and invited me to explain to her what she didn’t understand at first. 

Bots friends or enemies of press freedom Charlotte the chatbot in conversation with ECPMF's Emma Hattstein

Mrs Zoon’s chatbot is a good example of how bots can support media freedom because it gives you the opportunity to get into touch with people who might be out of reach for a person but not for a software. The bot can talk to them and gain information without putting its interview partner in danger by being seen with a journalist or having the need to go to a real-life appointment for instance. 

Summing up the experience of talking to a bot I would say that it still very much feels like talking to a machine, in my opinion it is not even close to talking to a person. Even though Charlotte is an advanced bot, you judge its skills depending on the fact that it is a software. Speaking personally I would always think: “For a bot it is quite good at asking questions.” 

Ms Zoon’s bot also brings up the issue of trust: Charlotte can’t give anyone the feeling of talking to a trusted person for an obvious reason: it is a program. Even I felt a little strange talking to it because of its lack of empathy, intuition, interaction, facial expression, body language, reactions….  

Automated journalistic tools with the goal of supporting journalist’s efficiency allow for more time to be allocated to investigative and reflective tasks that lead towards in-depth journalism 

Hanna Zoon, Jeske van Dongen, Jorge Alves Lino explain in their report. 

You probably heard of Quill. Quill is designed to independently create texts: it offers you the opportunity to feed it with raw data that it then uses as the basis for an article. That of course is great when editors need articles to be written very fast, or a lot of them. 

Heliograf the bot of the Washington Post managed to write all the articles needed to inform people about the 2015 Olympics when other journalists would have spent hours and hours to produce the same amount. In 2016 Heliograf had already published 850 articles by itself. This gave the reporters the freedom to focus on other stories. Those sort of tools also allow journalists to change aspects in the articles manually before publishing them.  

Bots can spot offensive words

One solution for dealing with abusive language, discrimination and hate speech already exists: engineers at IBM created a bot that eliminates all “bad language”. But what sounds so simple at first, is quite difficult in reality: The bot only spots words or phrases that are meant to be offensive, but it is not able to understand irony or sarcasm. 

Also the FRAbot offers the opportunity to deal with language issues especially hate speech. Blanka Tapia, a Programme Manager in the EU Agency for Fundamental Rights, held a presentation at the ECPMF’s Game Changer conference on the FRAbot. The FRAbot, which only exists as a beta version so far, is designed to respond to hate speech in the Internet. The bot is able to reply to tweets which employ hate speech by providing legal articles along with explanations of his or her wrongdoing to the user. Ms Tapia’s lively presentation was followed by a debate on the dangers and advantages of the bot between the conference’s attendees that showed how difficult it is to protect freedom of speech and abandoning hate speech at the same time.   

With bots, researching for an article doesn’t have to be a time-consuming task anymore: Reuters is using their News Tracer tool to spot important events on twitter for stories. The bot searches the social media platform for topics to write about and provides them to the journalists afterwards. News Tracer also checks if the profile it takes the information from, is trustworthy by analysing the structure of it. You can’t be sure that it is only using truthful content but you can for sure save a lot of time that you normally would have spend on research. 

Politibot updates your political knowledge

Another good reason for using bots is that they customise the content. Thanks to the ability of bots to create the article in a way each customer feels personally addressed, they are more likely to reach people. A good example for this is “Politibot” created in 2016 by the Spanish Journalist Eduardo Suárez. The idea behind this bot was to inform people via Telegram about the elections. What makes the bot so convenient is its simplicity: If they had any questions about this topic or needed a better understanding of the whole situation on general, they could just open the Telegram app and ask Politibot. The bot would then send articles written by the journalists behind this project to the user or refer to sources that had been approved by them. One of the advantages of Politibot, is that it is able to really communicate with its users, because it is able to understand language up to a certain level. To give an example: When Politibot receives an insult it will react with a GIF. 

To boost transparency we can all be informed about the author as well as the source of a text: some of the bigger companies like the Associated Press use Wordsmith, a programme created by Automated Insights that generates language, wise up their readers that some of their articles have been put together by this software. 

This whole topic is linked to authorship: Who wrote the article on that smaller website? Who tweeted the post? Who commented the picture on Instagram? Authorship is a difficult thing already. Articles written by bots will be grist to the mill. 

Good news again: There is a bot out there that has a solution to this. The NYT Anonymous bot sends out a tweet every time a story contains in anonymous source. By that the bot makes it visible for the consumers what they are getting. 

MEPs publish ethical guidelines

Taking into consideration all the aspects discussed above, I would say that media freedom is not necessarily under threat because of the employment of bots but we should define some principles that make sure it remains like that. We should maybe think of several guidelines for the usage of bots in the news department like the European Parliament has done it for AIs in general.

Apart from that, let's focus on the advantages of using bots like saving time, easing a lot of steps and giving journalists the freedom to concentrate more on what interests them. 

All in all, I am convinced that we should treat them as our friends not our enemies by using their abilities and sometimes by just putting them in their place. 

Find out more about this topic. 






Get in Contact

fact finding mission analysis