News and Media Symposium – News and Automation
6 October 2021
Prof Axel Bruns, QUT node, ADM+S (chair)
Rasmus Kleis Nielsen, Director of the Reuters Institute for the Study of Journalism
Prof Wiebke Loosen, Hans Bredow Institute
Assoc Prof Edson Tandoc Jr., Nanyang University of Technology
David Tomchak, Chief Digital Officer, Oxford Internet Institute
Watch the recording
Prof Axel Bruns:
Alright. Thank you. Thank you all very much and sorry about that slight confusion. We’ll continue straight on, and I will need my notes to introduce our next panel. This is a kind of a strange situation where all the speakers are on Zoom and the moderators in the room. So, hopefully we can make this work and I might actually sit on that side. I can see them as well, but we are delighted to have a really great panel and a highly, thoroughly, international panel, with speakers in Europe and Asia on for this next session on news and automation.
And let me just go to my notes, apologies for that slight delay, we have Rasmus Kleis Nielsen from the, sorry, the Reuters Institute for the Study of Journalism at Oxford. We have Wiebke Loosen from the Leibniz Institute for media, Hans Bredow Institute in Hamburg. We have Edson Tandoc from Nanyang Technological University in Singapore, and we have David Tomchak from the Oxford Internet Institute, where he’s a visiting policy fellow, and in his day job he is the head of digital and innovation with the new statesman.
So, we have a really fantastic high-powered diverse panel here with us today and we’ll hear from them with some opening statements to start with, before we then go to discussion and a Q and A. So, keep those questions on Slido coming, as well, as we go, but I’m going to throw first to Rasmus Kleis Nielsen. Rasmus is the director of the Reuters institute for the study of journalism and the professor of political communication at the University of Oxford. He was previously the director of research at the Reuters Institute and editor-in-chief of the international journal of press politics, which I’m sure many of you will be familiar with in this field. His work focuses on changes in the news media on political communication, the role of digital technologies in both, and he has done extensive research on journalism, American politics, and various forms of activism, as well as a significant amount of comparative work in Western Europe and beyond. And at least, I just really like to throw to Rasmus I guess, and ask him particularly also from the work that the writers institute’s doing with the annual digital news report. Yeah, where he sees the current trends in the field of news and automation. So, Rasmus, over to you.
Rasmus Kleis Nielsen:
Thank you very much. Thanks for this opportunity to share some of our research. I mean, I suppose the starting point here really is a sort of a central tenant of our work at the Reuters Institute, which is our belief that journalism exists in the context of its audience. And if we want to understand the way in which news circulates in our societies, how it’s produced, how it’s funded, how people think about it and may sometimes act on it, we need to understand the relationship between news and the audience. And that’s the underlying tenet of the annual Reuters institute digital news report, where we survey represented samples of internet users in 46 markets across the world to get a better understanding of how they use the news and how they think about news in the media. And I think that source of data which some of you will be familiar with, and it’s freely available online so please you know, use the digital news report.org website, or write to us if you’re interested in the original data. Further analysis, I think, that data can really get us as sort of a starting point to think about news and automation through the lens of us as citizens. If we start with the question of discovery – how do people come across their news? And I know that in the course of the day you talk about the world of the platforms, which is one that will re-emerge here because I think that the first thing I’ll say based on our research is that automation is still at this stage, much more down to news by platforms, that we as citizens rely on to discover news, that is done by news organizations or journalists who produce the news and publish the news. So, if we look at the question of discovery, most internet users use a number of different ways to find and access news online but when we ask people the follow-up question, what is your main way of accessing news online across the markets that we look at – which together account for about half of the world’s population – it’s less than a third who say that going direct to a news site or app is their main way of accessing news online. And the majority rely on different forms of distributed access, normally enabled by platforms such as search engines. Most importantly, of course, Google, social media, Facebook, Instagram, Youtube, Twitter, or aggregators for that matter – again, Google loot large but also Apple and Android and various competitors in that space. And of course in all these cases, automation’s absolutely central to how news is ranked, how it’s displayed, how it’s personalised, how it’s served up to individual citizens. So, this is the first point I would say, is that automation is done to the news by the platforms that we rely on as citizens. And again, I think this is a theme that’s popped up in course of your conversation today, it’s not always clear that we as citizens are aware of this. I mean it’s obvious that none of us really have a detailed technical understanding of these things, like we don’t have a technical understanding of many of the things we rely on in a complicated modern world, but even at a very basic level a few years ago when we asked a digital media literacy question of our respondents of how the things they see in their Facebook feed is ranked, about a third of the respondents flat out just said that they didn’t know. About a third correctly answered that this involves computer analysis algorithms to use a ten dollar word of data, on what each of us use and who we’re connected with and so forth. And about a third gave various answers that are incorrect, such as its selected by journalists who work for Facebook, or it’s selected by the government, for example. So, there are clear limitations still to even very basic media literacy of these forms of information. I want to also just sort of briefly share a few further data points to the set up. What we’ll hear from Wiebke and David in terms of how our research may feed into that, if they were to institute in addition to the way in which you know, we as citizens rely on platforms and internal line automation, it’s clear that of course publishers that produce and publish the news are also increasingly using various forms of automation in their own work. Now, we’ll hear more details on that from the research done by others on this panel, but I think it’s useful to think for a moment about what this looks like from the point of view of senior editors and executives who make many of the key decisions in these media organisations.
We do, every year, a survey of leaders in digital news and we have asked for example about what forms of AI and which areas of AI leaders and digital news are most interested in. And I think here it’s quite interesting to note that while from our point of view as users, as I said, before automation is very much about discovery from the point of view of news leaders, automation at this stage is essentially primarily about business and about efficiency. And the two sort of areas that we see leaders in digital news expressing most interest in investing in automation, are automatic recommendations – very sort of classic you, know you, if you’re interested in x you might also be interested in y. The sort of onwards journey at the bottom of news articles for example. They’re interested in a range of commercial uses that are often highly focused on either propensity to pay for subscription models, or ad placement and upload for advertising-based organisations. As well as things that I think for those of us who come to journalism as citizens, may seem quite sort of boring in a way, but actually incredibly important. Which is with the whole background question of automation of workflows in newsrooms, that in turn enable further automation. So, tagging for example, transcription subbing. All the ways in which news articles can cease to be 400 words and a still image that sort of floats somewhere on the internet. And in a concept management system, and become structured data that can be used in all sorts of different ways by the publishing. And these are some of the areas that we see leaders in digital news expressing interest in investing more effort in the future. So, at this stage, I think automation is still more than two news by the platforms we rely on as citizens. But I’ll also see more and more investment in automation by news organisations in the years ahead. And I think one of the questions we face there, is are those investments going to reinforce the strong winner takes most dynamics we see in the industry already, that has a split consolidation and largely sort of benefit the largest players by increasing barriers to entry? By creating data economies to scale. Or are they going to be an example of a new technology that facilitates lean, small, and focused brands by offloading expensive distribution back, office and sales functions both, or something else entirely. But with that I look forward to hearing from the other panellists into the conversation as we continue. Thank you, thank you.
Prof Axel Bruns:
Thank you very much Rasmus. You may not be able to notice but we’re in the middle of a thunderstorm here in Brisbane, so we had some wonderful max headboom style effects on you, as you were presenting. But you got through the thunder all right, so thank you very much for that. I’ll hand over straight to Wiebke Loosen. Professor Loosen is a senior journalism researcher at the Hans Bredow institute for media research in Hamburg, and a professor at the university of Hamburg. Her major areas of expertise are the transformation of journalism within a changing media environment, theories of journalism and methodology, and her current research focuses on the changing journalism audience relationship, the dartification of journalism, forms of pioneer journalism, and the emerging startup culture in journalism, as well as algorithms. Journalism-like constructions of public spheres and reality. So, Wiebke, over to you.
Prof Wiebke Loosen:
Thank you, Axel. And good morning from Hamburg to everybody. To be honest I prepared three slides to illustrate what I think is important when we talk about news and automation and would you mind if I share my screen?
Prof Axel Bruns:
Go ahead. Okay, go ahead.
Prof Wiebke Loosen:
Can you see it now? Okay. So, one slide is about the agony of defining what we are talking about in the first place, and Rasmus already highlighted some aspects and issue issues when it comes to news and automation. And the second one is about the human factor. So, it’s not only that we only talk about technology. And a third one that included three thesis on what I think these developments are about, in a broader sense, seen from a broader theoretical conceptual perspective. So, the agony of the definition.
So, when we talk about automation and AI, it is of course not unimportant what exactly we are talking about and what technologies can do. But even more crucial than hard definitions I think, is what people understand by them, and what ideas they hold about the possibilities and challenges when using such technologies. So, this definition for example is based on a survey of various newsrooms, and reflects the complexity and the broad understandings of the technologies and their various applications. And Rasmus, you already mentioned some of them. So, when we talk about AI and automation, we talk about different things that mean different things to people. So, they often they refer to the term or the ideas to advance technology. They talk about machine learning, about intelligent agents, about technologies that can to some extent, automate tasks in the newsroom.
The human factor should neither be forgotten or underestimated when it comes to technology and its implementation. For example, in newsrooms, because ultimately these ideas we have about technology are a hardly less powerful than what it can actually do. I would like to illustrate that with this quote here, it comes from a friend of mine. She is a journalist, she does tv, writes booksm and this is what she said when I told her what I’m talking about today. ‘Oh I think this is totally exciting, I would have to read up on the topic first but spontaneously I think that the journalistic feature, immediate experience, personal assessment, entertainment, everything a journalist incorporates to drive facts can never be replaced by AI.’
And if it should come to that, I don’t know if I want to witness that. The thought alone scares me. I also think the aspect is so important that this development goes hand-in-hand with self-delineation, and with self-doubt, and the shattering of identity. If an algorithm comes closer to what defines me, my empathy, my senses, and my perception, then I’m threatened in my mental existence. So, these are very strong words but we are used to talk to people that are experts in the field and that is not the case for everybody in journalism to be honest, only for a few people.
So, I think we are talking about a tension between automation of communication, and of communicative automation in a broader sense. So, since the Turing test, there’s an ongoing discussion to what extent the intelligence of automatic system is exclusively an attribution on the part of humans, or not. However, what technology can actually do is only one aspect more important than the question of how intelligent it is- I think is the question is the question. How it participates in communication, that is how it becomes a partner in communication.
When it comes to communication we are therefore dealing with a kind of what I’d like to say, soft automation in contrast to hard automation as we knew it from product manufacturing processes, in which robots built things. As this kind of soft automation stands for automation that’s built on data and digital traces. This is a by-product inherent to digitalisation. So, user traces, when it comes to for example- this characterises a kind of communicative automation that is an automation of social processes that go beyond the act of communication itself. So- and that is also a reason why the business case when it comes to data is so important. However, the automation of communication and communicative automation do not only concern personal human machine communication, they also affect public communication. Not only influencing the dynamics of public spheres through social bots, but on an already very fundamental level. Because they are becoming increasingly important for the production distribution and news of journalistic offerings, our main topic today.
So, when we talk about journalism we are dealing with an automation of communicative labour. I think this is important to keep in mind because journalism has an influence on far-reaching societal communication processes. It’s not only about journalism’s internal processes and routines, the back end of journalism as you said Rasmus. This is the basic reason why we have to be interested in the question of who is involved in the production of journalistic communication, and how this applies. Also more and more to tech firms, but also to non-humans or actons as latour says, who are taking on more and more tasks and journalism that serve to fulfill journalistic performances. So, particularly when we talk about journalism, we not only talk about the automation of communication but about a form of communicative automation. In this sense, I think it is not enough to talk about human machine interaction, it’s more about human-machine relationships as an understanding of an input/output relation. Does not go far enough to understand automation and AI and journalism, we need to develop an understanding of hybrid agency, a new form of agency that emerges in between, or in the interplay of humans and machine. This not only include an internal perspective that is what journalists, for an example, of journalists projection of agency to a machine or technology. As well as an intern/external perspective that concerns the so to say, supra individual agency of journalism, with including machines. It is not technology alone that develops agency, but it emerges in such hybrid figurations in journalism. For example, widespread ideas of robot journalism replacing people fall short. It’s about a refiguration of journalism as a whole. Thank you.
Prof Axel Bruns:
Thank you very much Wiebke. That’s a lot to think about there, too. So, thank you so much for that contribution. We’ll keep moving if we can, without being blown away in the storm, to our third speaker Edson Tandoc. Edson Tandoc JR is an Associate Professor and Associate Chair for research at the Wee Kim Wee School of Communication and Information, and director of the centre for information integrity and the internet at Nanyang Technological University in Singapore. He’s also an associate editor of Digital Journalism and Associate Editor of Human Communication Research. His research focuses on the sociology of message construction in the context of digital journalism. He has conducted studies on the construction of news and social media messages. His studies about influences on journalists have focused on the impact of journalistic roles, new technologies, and audience feedback, on the various stages of the news gatekeeping process. And this research has led him to study journalism also from the perspective of news consumers, investigating how readers make sense of critical incidents in journalism, and take part in reconsidering journalistic norms. And how changing news consumption patterns facilitate the spread of fake news. He’s also written an article with what to my mind, is still the best title of a journalism research article ever. Journalism is twerking, question mark. So, while I’m not expecting him to twerk, I’m sure he’s got lots of really interesting things to say. So, Edson. Over to you, please.
Assoc Prof Edson Tandoc:
Thank you. Thank you very much, and good afternoon. I can hear the storm from here. So, hello from Singapore. I hope you all stay safe and dry later. So, I’ll share what I’ve been thinking about, about the topic for today. And of course we know that automation in the news is not only about writing, but also to some extent is also about making sense of the news. And so what I would like to share this afternoon is how journalists make sense of how audiences respond to the news. So, many years ago when I was younger you know, I did some studies to understand how journalists were using web analytics, which was at least at that time, newly introduced in many newsrooms. So, web analytics refers to programs that track the digital footprint of audiences that users leave behind in a website. So, it provides information about audience behaviour patterns, such as time spent on an article or the website, the number of unique visitors, etc. And yes, this is an automatic process. We may say it’s an automated process, where after initial programming by human designers or coders, the system then collects, analyses, and visualises audience data and provides journalists and news managers long-term, but also real-time reports about audiences. So, as a former newspaper journalist, we never had this during my time at the newspaper. I was really intrigued by web analytics and how having access to detailed and comprehensive information about the audience can be quite powerful, if not transformative.
So, now we’re not only reading letters to the editor or relying on survey data to get a glimpse of what audiences want and what they actually do with news content, but with web analytics we can quantify audience behaviour and preferences in real time. And as we know, anything that can be quantified can be very persuasive. Soon however, based on research that has been done in this area, we realised that it is not the audience metrics per se – the metrics that are generated by web analytics – it’s not the metrics per se that is transforming news work, but the meanings that those involved in, or deal with journalism, ascribe to these metrics. So, from the audience side, studies have found that automatically generated metrics, such as number of shares or views that are shown alongside an article, can be interpreted by other readers not only as a signal of popularity, but also to some extent of credibility, which may affect their news judgment and subsequent user behaviour. So, popular stories tend to be displayed more prominently on the web page, because of that, they get shared more often which makes them even more popular. S, it’s a never ending cycle generated because of metrics. And of course, now however, we know that with trolls and bots, online popularity can also be manufactured. So, from the perspective of journalists, we have also seen how for some metrics have been regarded as the ends rather than as means to something. So, for example, some news rooms provide staff with specific targets to increase number of unique visits by a certain percentage. So, one news room that I visited many years ago, they had a goal of increasing their unique visits by 10 from the previous year. Some reporters also take pride in seeing their articles gain high engagement metrics. If when I visited some newsrooms I would be introduced to a reporter, as it is the reporter who’s bringing all these engagement with the website. And some journalists take pride in that. Some organisations also incorporate web metrics in their assessments of employee performance, that some reporters that we’ve spoken with in Singapore for example, say that they now include web analytics data in their appraisal documents every year.
One important stage in the news construction process is what some call as the interpretation stage. This is when journalists engage in introspection. So, after a news has been selected, it’s edited, published, and distributed. That’s usually followed by a form of introspection. In the newspaper that I worked for before, our editors would do a post more than meeting the following day, to reflect on today’s paper, basically looking at what they did right or what was not very effective. But in online news rooms, where there are no definite deadlines. If a story breaks it has to be up soon, there really is no chance to engage in these kinds of deep reflections about every single editorial decision that one makes, that sometimes an editor might get surprised when they look at the home page and it’s like all BTS stories been there. And I hope there are BTS fans in here. So, time spent on introspection is one thing, but the nature of introspection is another. So, during my field work for my dissertation eight years ago, and also based on field work that many other researchers have done more recently, we’ve heard how the phrase ‘doing well’ has become part of the newsroom lexicon, especially during editorial meetings. So, that story did well, this topic is going to do well, or this article is not doing well. So, we need to change it so doing well refers to performance on web analytics almost exclusively, refers to performance based on web analytics. And some editors we spoke with even had a criteria for what would constitute doing well in their news room, relative to their news organisation size. Is it 5 000 views or 50 000 views. But also, what would constitute as a golden metric for a newsroom has also evolved over time from number of hits, then the number of views, the number of unique visitors, to time spent, and now people talk about scroll depth. What is common across these different metrics is that these are automatically collected and reported by the web analytics system, and then given meaning by a human decision maker. So, in a sense, it is a form of journalistic sense-making. But now also driven by automation. So, I am there. Thank you.
Prof Axel Bruns:
Brilliant. Thank you thank you so much. That was fantastic. And in fact, I think leads in very well to our fourth and final speaker as well. Edson, you’ve talked so much about the work in newsrooms already. So, our fourth speaker is someone who has spent quite a bit of time in newsrooms. He is a visiting policy fellow with the Oxford Internet Institute, but in his day job I believe he has now taken on the role of head of digital and innovation with the New Statesman, where he is leading efforts to bring together exceptional journalism with data centric digital transformation, and accelerate the title’s growth. Prior to his appointment at the New Statesman, David Tomchak spent three years at the Evening Standard as digital editor-in-chief, leading all digital editorial content and journalism across platforms, including the coverage of two general elections in the UK. During the tenure there he also led the title through record-breaking growth, winning numerous awards, including the grand prix drum award in 2019, not long after which he became digital publisher and was responsible for the Evening Standards digital bottom line. So, hopefully, and I’m sure he will be able to provide really fascinating insights into what this looks like from the news/media professionals side. So, David, over to you.
Thank you. Excuse me. Good morning. Hi, I’m delighted to be with you this afternoon. Actually it’s early morning in London which is not a great time for me, but anyway, good afternoon. As mentioned my name is David Tomchak and I’ve spent the last 20 or so years working as a journalist and editor, running newsrooms. I’ve actually left the New Statesman media group that was after about a year with them, and I’d finished my work there. My final stint was as the Chief Operating Officer, and as mentioned amongst other things, innovation across our titles was a big part of the job, which included of course the use of machine learning. So, I come to this discussion as really a practitioner who can hopefully bring an insider’s view of how people see the technology in our industry. As Wiebke can Edson have already outlined, there are many different ways that this technology can be used in newsrooms, and by news companies. There are indeed debates over what the technology even is and of course as Rasmus said to date, it feels like AI is being done to the news by platforms, as opposed to being developed by news for newsrooms. I think to a few in the industry, to an extent, even it can feel a little overwhelming. I think I was really interested by Wiebke’s friends quote. I found that fascinating to see the scepticism in there, and I understand where that’s coming from and I’ve met many people who feel they welcome technical change but fear the impact of this technology on journalism, on quality, for example. Or where the journalists will lose their jobs. There are even bigger existential fears like will big tech use the tech, this type of technology is the final nail in the coffin for journalism. Of course a good dose of scepticism is very healthy and we wouldn’t be doing this if we weren’t sceptics.
But at a practical level for me, I think the opportunities that the technology presents are exciting and very much outweigh the risks, and thankfully I’m not alone to quote Rasmus’s latest trends and prediction report. That’s the one that he mentioned that surveys digital leaders in news, in particular. I think something around 70 percent of newsroom leaders saw AI as the most important enabling technology in journalism – Rasmus, you can correct me if that’s wrong. But it can’t be all bad if so many peers find the technology a potentially positive influence. For me, AI really sparked an interest in my newsrooms in the UK about four years ago, when big tech platforms started to roll out home voice assistant technology that was powered by AI. We’ll all remember the different home devices that were being advertised back then. I suppose we’d already incorporated AI in the form of third parties to help with some of the back-end processes, like discovering trends. Much of which has been described, it was really through simple dashboards that we were using the technology, and that came through third-party plug-and-play type of software. A lot of the stuff looked at web analytics, the way that Edson just mentioned, and I recognise very much of what he says when it comes to use of web analytics for things like setting targets for staff, or trying to make sense of the ecosystem that’s around us. But anyway, about four years ago I was Digital Editor-in-Chief of the Evening Standard, as mentioned. And at the time, I felt we could do things in collaboration with the big tech platforms that could benefit our journalism and really bring new audiences to the newspaper. And bear in mind, voice is something that was really very new at the time Siri had existed for quite some time but there wasn’t the ability to really broadcast into homes in the same way and newspapers have often been ahead in some of the adoption and testing with this technology. So, there was an environment that was right for this type of collaboration, but collaboration was required because of the complexity of the technology. In fact, big tech was even prepared to underwrite the costs, which to be honest, even I was suspicious about it first. So much so, that I actually set up a regular informal group with some friends and peers from the big titles called the AI media working group, and we just wanted to make sure that we weren’t being divided and conquered, so to speak, by this sort of trojan horse of big tech money. And it ended up being quite a good social actually. But anyway, that was about four years ago and over time – and I think this is true, many newsrooms that have been adopting this type of technology in a more hands-on way, not just using those third-party plug-ins – I think I’ve come to realise that collaboration whether it’s in the industry or the big tech platforms, is really a key to the industry’s journey, using this type of technology.
So, collaboration is really the main thing on my mind at the moment. I think day to day, other things – and Edson touched a little bit on, we’ve talked briefly about things like fake news, but day-to-day, the technology is being used to assess things like the veracity of news. But there are sort of two or three other main areas that are at the forefront of people’s thinking and they’ve already been touched on, as I say. But just to sort of clarify, there’s an idea of can this technology help with efficiency? Can it help add value? And that could mean, you know, the New York Times actually has a great example where things like archives can be trolled and used properly, and can this type of technology help us scale. So, that was something that we were doing very much at the New Statesman just before I left. Looking at how we can create extra more content from different types of data sources. And I think over time there’s another area which is becoming a little bit more interesting from my perspective, and that’s to what extent are newsrooms actually willing to share data?
I’m doing some work at the moment for the European Commission, it’s looking at how can newsrooms in Europe potentially use their combined force to use this type of technology in an effective way. And it feels like the best way of doing that is bringing data together. Of course the big issue is that data is often seen as something that needs to be guarded, and so I think the next stage will be looking at how data can be used across newsrooms in a way that’s effective for all the different platforms. So, yeah. That’s kind of where I think that, as policy makers and regulators start to catch up with some of the platforms, there’ll be more debate on that. And but, by and large, I feel like newsrooms are happy to be using this technology and collaborating with others.
Prof Axel Bruns:
I might sit here to see any questions coming through on Slido, so please if you have questions, post them now. But that was a really fascinating discussion. I thought, amongst such a wonderful panel, I think what really came through, just while we’re waiting for other questions to come through, what really came through here was for me, a kind of a sense that there is a lot of the instant interstitial and interpretive work that’s going on between the different actors – both human and non-human actors, I guess, in this network. Not to get all literary in here but you know, the different imaginations of automated technologies of AI, of algorithms, and so on, of what the metrics mean. As Edson was talking about as well, between journalists, technologies, the users themselves, various other stakeholders. There is a lot of translational work that seems to be going on basically between them about, these tools and technologies and metrics and everything else. And I do wonder, to what extent that there are still lots of misunderstandings or misinterpretations perhaps, going on between these different stakeholders. So, to what extent – to ask this quite broadly – to extend are we speaking the same language actually across these different groups, and to what extent do we still see people pushing in very different directions. I guess with these technologies that’s a very broad question, I know. But I’m wondering if any of you have any thoughts on this?
Prof Wiebke Loosen:
So, maybe I can add to to this. So, I am about to start to talk to people working in positions like the one David had. So, innovation, development, product development, and so on and so forth, in particular when it comes to AI and automation. And what I hear from these people, from these roles in media organisations is that, 80 perent of their work is explaining what it’s about. So fostering an understanding in a media organisation, and that is the most important part in their work, as they say, and takes much more time than actually doing it or implementing it. So, far beyond such technologies are entering media organisations and newsroom, there’s a lot of talk, and a lot about these imaginations as you said, about what technologies can do. And this is also influencing the further development, and that is why I think it’s so important that we not only look at the technological framework so to say, and try as social scientists to understand what it all means, but exactly, look at this interface so to say, between what people are doing with it.
Yes, thank you. Any other things to add to this?
Rasmus Kleis Nielson:
I mean maybe briefly, from my end. I think one of the currents that sort of flow around anything of importance in our society is sort of a self-promotional current. And I think we need to keep in mind that a lot of our conversations about anything important, including automation and the use of AI in news and elsewhere, is driven by self-interested actors. But not necessarily in a bad way. I mean, I have my own self-interest as well, you know, we all do. There’s nothing sort of intrinsically wrong about tha, but nonetheless, self-interested. And a lot of the conversation will reflect that when we’ve looked for example, the news coverage of AI in UK media. While it’s true that there is sort of the occasional killer robot story illustrated with a still from turning 2 or six hundred, sort of wrecking havoc, actually the vast majority of stories about AI is driven by industry sources. And many of the experts who appear in the coverage of AI are experts who hold dual appointments, who have full-time jobs or part-time jobs with tech companies in addition to the work they do as academics. And I think we should recognise that this might apply within organisations as well, that a lot of the conversation around the use of AI and automation will play out in part, as conversations between different groups within newsroom who have different priorities, want to spend money in different ways, and necessarily will promote what they prefer. That applies to the people who may want to spend on more reporters, as well as the people who want to be spending more on automation. So, I would just sort of say, I think there is an element as a promotional discourse, and of course this applies to the people who rule us as well. I mean the way in which governments talk about AI, and the way in way which I as a citizen should not all be worried in any way or shape or form, by the use of facial recognition by the police, or the use of social profiling and social services, and the like. Because you know, trust your government, you know. This same promotional discourse will also be used, I think internally, news organisations and in public discourse around us. And I think it’s been in some ways sort of social scientists are having a little bit of a field challenging these discussions with various of sophisticated literarian or post-modern moves, and warning us all against rarefying and naturalising this technology, and whatnot. But I have to say the people I meet who are most disenchanted with public discourse around AI are professors of AI who work in computer science, who think we are all off our rocker basically in the way in which we talk about it. And who I think by now has sort of basically, in many cases, given up. And talking to journalists or policymakers because they feel it’s so stupid what they have to deal with, and that they are just never taken seriously, and they’re always sort of put on platforms with various professional promotionalists, or self-styled you know critical thinkers or whatnot, many of whom you know, would struggle to explain even the most rudimentary elements of any of the technologies that they profess about. Thank you.
Can I just add to that, I think that’s really interesting. I think there’s definitely a big distinction between using the technology in the news industry, in journalism more broadly, and then how it’s reported. And that it’s, the two things are two ends of the same story. But they’re not the same. That makes sense, and in terms of using them in news rooms to Wiebke’s point, I think the way that these types of technologies end up actually gaining momentum is when people who are in the businesses demonstrate the potential value. That’s the key. So, being able to show that there’s a value to a workflow, or there’s a value to the audience, there’s a value to the commercial team, those things are very tantalizing for whoever has the budget to allocate, or similar. So, you know, it’s not – I think that the other thing is that this technology is being spoken about so much that people fear being left behind. There’s this sort of, almost a fear as well, about not just the technology and its implications, but being left behind. So, often if you’re in my shoes, it’s sometimes quite easy if you can show a tangible benefit, even if it’s a small value you know, people are aware the technology exists and they want to do stuff with it as a result. They’re quite happy to play. I think the role of the big tech platforms in that, is something that I feel still needs to be debated quite rigorously and remembering that value in a lot of this technology only exists if you’ve got data attached to it. And so, the debate really is around the data frankly, as opposed to rolling out algorithms or anything similar, is learning. Although that’s very important, that’s standing on the shoulder of the giant of data. But yeah, I think it’s fascinating. I think that putting this stuff into newsrooms is very difficult, still.
Prof Axel Bruns:
Edson, did you want to add anything?
Assoc Prof Edson Tandoc:
Yeah. Although I hope you can hear me. I think I’m having issues with my internet. It’s also started raining here where I am but yeah, not as strong as what I’m hearing from there. I think for me, I just wanted to add is in journalism studies for example, there’s the prevailing normalisation hypotheses where journalists usually normalise technology adapting them into how they do their work. I think that has been, I would say, a very fruitful framework. But I think it’s also important for us to think how journalism has been adapted for technology. I think that’s probably one constraint that we face, is that we treat journalism as something as monolithic, and that technologies have to be adjusted to fit into journalism. But maybe because of that preoccupation with that framework, we may be losing track of how journalism actually has been transformed to fit into these technologies that we’re witnessing.
Prof Axel Bruns:
Alright, thank you very much we’ve got one point to add…
Prof Wiebke Loosen:
Can I add something? So, I think it’s important that we not only think about it as adapting technology, so it’s not only adapting that technology is thrown on journalism, so to say, I think it’s a constant re-development, and redevelopment also for journalistic purposes. And so, we are journalism researchers, that is our main field of interest, but what we see there is that there’s also a big discussion of, do we need in-house solutions? Can we use the solutions that the big industry is providing? And I observe that as a constant rediscussing, what is purposeful for journalism in particular. So, that really makes a difference when it comes to different fields and society.
You know Wiebke, we did, sorry, just that we did this AI and media working group where we have someone who looks at corporate venture capital, who works in a b2b title, a guy called Jim Monson, he’s really interesting. Anyway, we do presentations to each other and he did a presentation on investment in AI and the different areas of the economies where investment comes from for this type of technology. And based just on investment, the idea that this type of technology will come from within the media industry is virtually unthinkable. So, this is why I go by study of collaboration, I think it’s really important that we – because we can’t afford it – we need to find a way of working with other groups to not have it imposed upon us, to your point. I think it’s really, no matter how you look at it, if you follow the money that’s probably where the influence will come from with something, with technology like this. So, I think your point, it’s really interesting in that you know, hopefully what you’re describing won’t happen, that we’ll be able to have more control, but I think we’ll only be able to do it if we collaborate with others.
Prof Axel Bruns:
I almost hate to break up this really interesting conversation between the panel, but we now have a number of really good questions from the audience, as well. In fact I might just ask you, there’s a handful for two specific panellists as well. So, I might just ask you to give a very quick answer if you can, to some of those and in fact David, following on from what you just said but also where you ended up in your original presentation, there are a couple of questions that ask about the kind of data sharing collaboration across organisations and where you see useful synergies, or useful benefits I guess, from doing that.
That’s it’s a great question. It’s a question that I’m grappling with at the moment, with the with the European Commission. So, I think the areas of data where I believe journalists at the moment across different organisations are really prepared to collaborate, is around content. So, you know, there are lots of one-off investigations that we could cite, where people have come together – the Panama papers is probably the most famous at the moment – but where people have come together and shared data to create stories. That’s done in a sort of one-off, almost piecemeal way, at the moment. I think that that could probably be improved. I think at the other end of the spectrum, is audience data which ironically, newsrooms already share with platforms and don’t really think about it. So, you know, we’re sharing data right now with Zoom, but you know, Google, or Facebook and all the other large tech platforms have lots of data that frankly they can amalgaming across different audience data. They can amalgamate from different organisations. Newsrooms aren’t thinking in those terms and I think if we got to a point where we could work together with data effectively at the content level, then ultimately we’d be happy to share data at the audience level. And you might wonder why aren’t we sharing data at the audience level already if Google and Facebook have it themselves? And the issue is because that’s how we currently monetise most of our content, when it comes to things like advertising. So, there’s this fear that sharing audience data will mean that we give up some of our competitive advantage commercially, and that’s not true.
In fact, if you add the data together, you can all benefit from that, I believe. So, yeah, that’s kind of where I think we’re at with data sharing.
Prof Axel Bruns:
Thank you, and that maybe leads on to a question for Edson as well, from Ayiesha asking about your experience with metrics. You talked about this move from initial scepticism to verbally normalising what’s doing well in newsrooms, which is a really interesting and ill-defined term, I guess. So, where do you see the future of metrics?
Assoc Prof Edson Tandoc:
Oh, that’s a very difficult question, and I may need a few more years to answer this question. What I would say though, now if we try to design a follow-up studies over the years, and I think for me, now that it’s considered normal in the usual – it’s no longer considered something that’s new. Almost all newsrooms use either a paid version or a Google analytics version of web analytics. I think it’s even more important to see how it’s affecting news work, now that it’s almost invisible in the newsroom. It’s something that has reached what they call a taken-for-grantedness stage. And then the effects might be even more visible for me. I’m personally interested in the use of analytics for assessment of personal, and how also trends might be transforming journalists personal attitudes about their work, and about what they think they have to do to perform well in that setting. So, I’m going into that direction. But in terms of how the technology is going to continue to improve, we also know there are limitations to what analytics can say. It can tell us what’s happening but still not explain why. And also, the metrics that are being used to assess what’s doing well is also changing. And I think that the future of web analytics would be how it can define to serve the purposes of journalism, at least from the text side. But from a journalist’s point of view, I think I’m really interested in how it is maybe quietly transforming how we even understand the work that we have to do.
Prof Axel Bruns:
Fantastic, thank you. We only have a few minutes left and it’s very much in my interest to finish on time because I still have to give a talk at 6. But we have a really, I think, interesting question here from Mark Andrejevic who’s curious to hear your response as a panel to arrangements – I guess regulation, like the Australian News Media Bargaining Code – which I’m sure you’ve followed the various debates, and the brief ban on news on Facebook that resulted from it, as well. So, yeah. Curious to see if you’ve got any thoughts on these kinds of laws and regulations. Or if no one wants to touch that hot potato.
So, just a general thought? I think that it backfired, you know. I mean honestly though, there needs to be a better effort at regulatory sort of mechanisms, or the creation of regulatory mechanisms for the use of this technology. But it goes far beyond the news industry and it’s easier for me to report on it than to probably try and get involved in the regulation. But that you know, it doesn’t feel like the experts are doing what they should be doing yet.
Rasmus Kleis Nielsen:
I mean I suppose if an Australian citizen asked me for a way of thinking about assessing the efficacy of an arrangement in this space and we first say what is the problem, and if the problem is the one he’s trying to address is the risk of market failure, in terms of local news provision, and in terms of news media that serve historically underprivileged groups, whether they’re marginalised, ethnic minorities, indigenous populations, alike- if that’s the problem one wants to solve then I suppose the policy should be judged on whether it has effectively solved that problem in a way that is transparent and accountable. And if it has failed to solve that problem, or if it just does not ensure any kind of transparency or accountability in terms of those of money, then I would probably have reservations about it as a citizen.
Prof Axel Bruns:
Alright. I think that might echo some of the views in the room as well, but I’m not going to speak for everyone. Maybe just to finish off then, we’ve got about four minutes left. There is a question ironically here, from an anonymous user, whether you could speak to the importance of trust and what makes a platform or a piece of technology trustworthy.
Perhaps there’s some views on the panel? Or not.
Prof Wiebke Loosen:
So, yeah. Maybe I can add something to that. So, what I find fascinating that we are witnessing the development of that part of AI that is called explainable AI. So, I’m not an expert in computer science but I understand that as throwing technology on technology to better understand technology, and that I find totally fascinating. And it comes with the idea as I would say, that we need trustful technology, especially when it comes to journalism. So, why is journalism research around? Because we want to understand the selection criteria of journalists and news organisations as they produce a kind of communication that is so important for society. Of course we adopt the same approach to technology that is used in this field, and also in other parts of society. So, I think that is a natural development, so to say. But it’s ironic in itself, if we try to do that by technological means, so to say.
Thank you Wiebke. Rasmus, I’m not quite sure if you have your hand up or if this is actually Zoom doing the AI thing and recognising your gesture.
Rasmus Klies Neilsen:
No, I do have my hand up. It’s just that I’m conscious that we’re about to wrap up and I thought I’d say something towards the end to sort of express an element of what I think is a fact-based hope, or cautious empirical optimism, around the role of automation in news. Because I think it’s often easy to get rid of – sort of focus on the challenges and the complications – and they are real. And there will be consequences, also negative consequences, and there will be people who will be the losing end of the transformation. That’s always the way in which creative destruction works out. And we can’t be naive about that. I would also just say that you know what I try to sort of keep in mind is that I would say that almost all journalists I meet are curious and capable individuals who you know, want to do better journalism tomorrow than was done yesterday, and who are not sort of transfixed and sort of kept captive by a picture of the past. They want to do better in the future. And it’s also the case that the news industry, while it’s taking a very serious knocking over the last two decades, is globally still a very large industry. The world association of news media estimates the total turnover of the global newspaper industry alone, at about 100 billion US dollars a year. So, we have very capable individuals, and we need to remember that while much smaller than the tech companies, there’s still a lot of money. And if the news industry invests same in R and D as say you know, car manufacturers do, and were as focused as companies – not on sort of milking declining assets or running a sensitive industry in an asset stripping fashion – but investing in the future, I think they have staff, colleagues talent in their organisations, who will have great ideas for doing journalistic things, business things, organisational things with new technologies. And I think they will find that there is a business case for doing it, if they’re willing to put up the investment and break with where I think a lot of news organisations have inadvertently ended up in recent decades which is, they’ve moved from being in the past, technology makers who made or commissioned the technologies they relied upon – printing technology was developed by newspaper publishers, broadcasting technology by broadcasters – to technology takers who were largely reliant on technologies developed by others, for other purposes that were then secondhand adapted for publishing purposes. So, I think there is an opportunity here in addition to a number of challenges. I’m not saying everyone will make it, or that everyone will benefit equally from it. But I think the opportunities are there and I think it’s up to journalists and publishers to also sort of fight that out amongst themselves internally in the industry, whether they will settle for the role as technology takers or whether the industry in the profession will seek to seize this role as technology makers again, and try to define their own terms. Also technologically moving forward, or whether they want to sort of rely on second-hand technologies that are handed to them by others.
Prof Axel Bruns:
Thank you. That’s a call to action if you ever have heard one. And I think for those of us who work in journalism or media research, or who work in journalism and the media, that’s a great provocation to end on.
So, thank you all. Thank you very much to this wonderful panel, and greetings to all of you over where you are, and hopefully we can see you again sometime in person, as well. And thank you everyone on Zoom, as well as here in the room, for what’s been a really long day. But also very exciting and stimulating day. So, this finishes today’s program as far as ADM+S is concerned, but for those of you who are here, you’re very welcome to come over to F block, grab a drink and a canape and hear me talk for a bit more before we go over to the golf club. So, thank you all and see you again tomorrow morning. Thank you.
Sorry, just before, the in-person people, the QUTX event- just noting that today the Covid restrictions did change at 4 pm so what that means is from now on, when you’re eating and drinking, you’ll have to remain seated. So, what that means for the QUTX event is that you can grab a drink and some canapes on the way in, but you’ll have to consume that seated within the lecture theatre. So, we’ll leave in about two minutes if anybody doesn’t know the way to F block, just follow me and Kathy.