Can we still handle the truth? Journalism, ‘alternative facts’ and the rise of AI
We all have moments in life when we know something big is happening, that we are stepping into a new and consequential experience, and our mind takes a mental Polaroid, an intensely clear snapshot of what that moment looks like and how it feels, and then stores it away in a file marked “important”.
Well, my mind does anyway, and in my professional life so far there have been three.
It’s late 1987. I’m nearing the end of a cadetship at the Canberra Times and it is finally my turn to work in the newspaper’s press gallery bureau at Old Parliament House. I had chosen the Canberra Times over other opportunities because I figured it would be the fastest path to being a political reporter, the only kind of reporting I was really interested in; and here I was, on a sunny spring morning, walking past the roses and straight in the front door of that white building I’d seen every night on the news, because I worked there. I had to pause to calm my nerves. All I could think was: “Now I can start.”It was more than a decade after Watergate, but like most reporters of my generation that investigation was an already mythologised touchstone for the power of journalism to effect change. I absorbed that quote from Woodward and Bernstein: “All good reporting is the same thing – the best obtainable version of the truth.”
From the starry-eyed perspective of an excited cadet, that idea seemed simple. Over the next 36 years, I came to understand how slippery and difficult it could be. As the waves of financial and technological change battered and shifted how and what we did, when the very concept of truth as the foundation of public debate was challenged, achieving the best obtainable version of the truth became very complicated indeed.
Journalism, balance and truth
In the late 1980s, the news cycle still moved with the sun. Papers dropped on the lawn in the morning and we had until the evening to file news for the following day. There was more time to ask questions and talk to contacts. More time to think. And as we all know, in those days, the media held a near monopoly on the attention of readers, a commodity that ensured those “rivers of gold” in advertising revenue kept flowing.
This manageable tempo made it easier to find and test information, but the gatekeeping role also conferred on publishers considerable power to shape public opinion and determine which ideas, and whose ideas, got to be heard, how the obtainable version of the truth was portrayed and framed. News could be angled to suit the commercial or ideological interests of owners or toned down to ensure no advertisers complained.All the big mastheads in Australia held that traditional gatekeeping power, but it was wielded most overtly by News Corporation, an organisation I had joined as a junior reporter in 1989. As successive governments approved mergers that steadily reduced the number of publishers and increased concentration in media ownership over the ensuing decades, urged on by the editorial columns of some of those same papers, concern that power was being exercised in a way that harmed Australia’s democracy eventually grew to the point that more than 1 million Australians signed a petition supported by two former prime ministers, from opposing parties, calling for a royal commission. But that’s all to come.
Then, as now, there were diligent reporters at the Australian working hard and breaking news, and then, as now, it took an editorial view on issues and politics, as it is entitled to. But, as has been extensively traversed over the years, including in Robert Manne’s 2011 Quarterly essay Bad News, the exercise of power did at times appear to override the truth.
The organisation’s coverage of global heating is an obvious example. It highlighted the views of climate sceptics long after the science was clear and those views were discredited, fomenting doubt and confusion to back an editorial position often in favour of Australia delaying effective domestic action, a goal that sadly, it helped to achieve. The positioning became too much even for James Murdoch at the height of the 2020 bushfire crisis, when he and his wife, Kathryn, released a statement saying they were “particularly disappointed with the ongoing denial among the news outlets in Australia given obvious evidence to the contrary”.
The power of publishers to influence the news was of course not a new phenomenon.
As evidenced by the Guardian’s reckoning with the connections of its founders with enslavement through their commercial ties to the cotton trade, it had long been the case.
In the 1800s, the Manchester Guardian’s editorial columns deplored the existence of slavery in the US, but then failed to support prescriptions to end it.
As the global editor-in-chief, Katharine Viner, wrote in our editorial series on the issue “Cotton Capital”, it is “difficult to avoid the conclusion that the commercial interests of the founders may have influenced the paper’s editorial policy”. The establishment of the Scott Trust in 1936 as the Guardian’s sole shareholder has protected our editorial independence since that time.
Coming back to Australia in the 1990s – press gallery bureaus were big enough for reporters to have specialist rounds and there was healthy competition for quite detailed policy coverage across all mastheads. But over time, I also saw how some senior reporters sacrificed fully scrutinised assessments for easy exclusives from politicians with whom they were close. When I was appointed political correspondent for the Australian in 1994, a senior press gallery colleague invited me to coffee to inform me that “they”, by which I believe he meant some of the blokes who held most of the senior roles in the press gallery, didn’t think I would cut it, because I didn’t understand how it worked. To succeed as a political correspondent you had to “do favours and be owed favours” he said, develop sources who would help you, and you them.
I was, he said, far too straight down the line, too much like Michelle Grattan. I continue to admire Michelle Grattan and the many other press gallery reporters who are straight down the line. Sources and trusted relationships are necessary to do that job, but never at the expense of fairness and facts. Getting too close to sources is a danger, and not only for the politics beat.For me the unbearable pressure point came when I briefly returned to the Australian in 2009 and was assigned to cover the Copenhagen climate conference. For several days running I was asked to leave the meeting with 150 world leaders and 20,000 delegates with the fate of a global agreement to reduce emissions hanging in the balance, to travel across the city to the Copenhagen Climate Challenge. This was a gathering of a handful of climate deniers including the Australian geologist Prof Ian Plimer who said he was in Copenhagen to try to stop the world engaging in the “global collective madness”. The request, I was told, was being made in the interests of “balanced coverage”.Eventually, along with a few other Australian reporters, I went and filed a gently sassy sketch, having discovered a few old fellows drinking fruity lexia at lunchtime and talking about how the weather had seemed much hotter when they were lads. I believed, given the overwhelming heft of scientific evidence, that continuing to elevate the views of climate deniers was distortion, not balance. And yet it’s still happening. A 2022 report by the Institute for Strategic Dialogue found Australia’s Sky News channel has become a central source for climate science misinformation around the world.
But, if we dismiss that view of balance and objectivity as an extreme example of false equivalence, what is the alternative?
I fall back on the simple formula from Bill Kovach and Tom Rosenstiel in their 2007 book, taught in most journalism schools, The Elements of Journalism.
They make the case that objectivity is not something required of each journalist personally, there should be no expectation that we come to our desks devoid of all views or thoughts or opinions, but rather it is something required of the method each journalist applies in conducting their work. Objectivity means employing observable, repeatable methods of verification. It means always following the facts. Our methods have to be objective and, as far as possible, transparent to our readers. We should care deeply about where the truth lies. But we have to be clear about how we go about finding and assessing it, how we amend or change or correct a story when new information arrives or when we have got something wrong or when our assumptions are challenged. It can be elusive and we need to document the hunt.
As Alan Sunderland, the journalist and former ABC executive, wrote in Meanjin in 2019, “regurgitating the views of others without assessing their factual basis is not journalism. Balancing a smart well-informed view with an ignorant ill-informed view and giving them the same weight is not journalism. Failing to care about where the truth lies is not journalism.”
We all come to this task with life experiences that inform what we know and how we view the world, our gender, our background and education, our race or ethnicity, our sexual orientation, and those experiences make a newsroom richer and better able to tell stories, find sources, understand what is going on around us.
I don’t want reporters to leave their true selves and experiences at home. We can’t possibly see what we need to if we are only looking at the world through the eyes of middle-class white folk. And at the moment, Guardian Australia has a way to go on that front.At the same time no view or perspective can override the facts, or result in us refusing to consider a different point of view, or not telling a story because we wish it wasn’t true or because it challenges something we believe. It can never mean we only include the sources who agree with what we think if there are reasoned alternative views. And yes, that does involve judgments on our part, which we need to be able to defend. It makes it all the more important that we never rush to pre-emptive or ill-informed judgment on social media. Truth might be hard to pin down, but it is not just what we want it to be. Shying away from stories because we wish the world was otherwise is doing the audience just as much a disservice as pulling punches to protect a friendly source.
As Rosenstiel said in a 2020 Twitter thread on the issue: “If we mistake subjectivity for truth, we will have wounded an already weakened profession at a critical time. If we lose the ability to understand other points of view we will have allowed our passions to overwhelm the purpose democratic society requires of its press.”
That exchange came as both the media and democratic society were changing at dizzying speed. Digital platforms were eroding the gatekeeping power of the old publishers and upending their business models. Some politicians didn’t just spin, they lied. And then they attacked the press for calling them out. They advanced “alternative facts”. Some news organisations began reporting falsehoods as facts, and then found themselves unable to stop doing it because their audience now believed so completely in the falsehoods. Attaining the best version of the truth kept getting harder. But again, I’m getting way ahead of the story.
I’m sitting with Katharine Murphy on the grass beside the big pear sculptures outside the National Gallery. We’ve both been offered jobs at the soon-to-launch Guardian Australia by Katharine Viner, who was Guardian Australia’s launch editor: me as political editor and Murph as my deputy. In the UK, with its billion-pound Scott Trust and, at that stage, 190-year history, the Guardian was an institution for the ages. But Guardian Australia was a philanthropically funded experiment.
I was, at the time, chief political correspondent for the Sydney Morning Herald, Murph was at the Age and the then Fairfax chief executive, Greg Hywood, had told us we were mad, that the Guardian might be a good brand in some suburbs in London, but the only news brands that would ever have any clout in Australia were Fairfax and News Corp. In other words, Australia’s concentration of ownership could never be broken. We knew he might be right. It was a risk. But we had the chance to prove him wrong, and if we did we would be helping to create a new voice, and an unashamedly progressive voice; not partisan, but happy to plant a stake for the truth. In a last-ditch attempt to dissuade us from resigning, Hywood had sent an executive to Canberra to have coffee with each of us that weekend; a nice bloke who we met, one after the other, and then we regrouped there on the lawn by the pears. We each typed our resignations, counted to three and pressed send.I was attracted to the idea of a digital-only offering at a media organisation at the forefront of working out how to manage a news cycle which no longer moved with the sun, but rather changed by the minute. This was a time when we were optimistic about the possibilities of how digital journalism could help us understand our readers and learn from them. It was a time when Facebook and Twitter were useful places to get information, talk to people who would otherwise be out of our view, discuss and debate and distribute our news. We were still hopeful about this new town square, and there were good reasons to be. The media’s gatekeeping power was being challenged. Being factchecked and challenged in real time, facing competition from bloggers and startups, seemed like a potentially positive kind of shake-up. But we could see we were facing big changes.And a lot has changed. We still do benefit from interaction and challenge and ideas from readers but over time the social platforms became addictive traps. We depended on them for distribution even as the migration of eyeballs to their sites eroded our revenue base. Virality favoured the uncomplicated, the emotional and the dogmatic. Nuance seldom went viral.
The logic of online interaction, the algorithms that favoured emotional extremes, the news sites that monetised clicks by feeding readers the ultimately unsatisfying sugar-hit of titillating easy consumption, it all formed a cycle, a kind of centrifugal force that pushed online debate towards ever more toxic extremes and an increasingly polarised readership. Civilised disagreement was swamped by divisiveness and anger and people entered closed-loop information systems impervious to alternative ideas. Some politicians fomented and capitalised on the whole mess and the polarisation corroded trust, that essential glue in a democracy, including trust in the press.
In 2018 Trump’s erstwhile strategist explained a plan to push this democratic destruction even further.
“We got elected on Drain the Swamp, Lock Her Up, Build a Wall. This was pure anger. Anger and fear is what gets people to the polls,” Steve Bannon told Bloomberg’s Michael Lewis. “The Democrats don’t matter. The real opposition is the media. And the way to deal with them is to flood the zone with shit.”
The news cycle was already moving too fast, already angry and partisan, but flooding the zone went further, aiming to deliberately overwhelm people with conflicting information to the point where they gave up on ever finding the truth, or turned off news altogether, or chose to blindly follow a side, a team. And once people identify with a side, once opinions become integrated into an understanding of personal belonging, they are heavily and instinctively motivated not to assess available information but rather to seek out information that supports their team view, which, on the internet, is always possible to find. I think many of us know good people who disappeared into that vortex during the traumatic years of the pandemic. And so we spin into a polarised world where there are chunks of the population that can’t tell fact from fiction, where a president can convince 40% of the population he won an election that he obviously lost, and encourage some of them to stage an insurrection.
n an essay for a conference about authoritarianism and faultlines in democracy, Jelani Cobb, the dean of the Columbia journalism school, wrote: “The crisis of American democracy has been facilitated by the crises confronting American journalism.
“Hindered by declining revenues, a diminished public trust, a shrinking labour force, and the emergence of a sophisticated disinformation ecosystem, the press, as the debates over the COVID vaccines and the 2020 election indicate, has had difficulty convincing many Americans of basic facts.”
How do we hold on to the truth if the best obtainable version of the truth no longer forms guardrails of public debate and disagreement? How do we filter a firehose of disinformation, or factcheck a torrent of lies? These were questions I was soon going to have to deal with in an entirely new way. But again I’m jumping ahead in the story.
Steering the course
It’s now May 2016. A crisp clear evening, the sun just setting, and I am standing outside Parliament House, waiting for a lift. The tufty-haired gang-gang cockatoos are hanging from the nearby trees, noisily eating red berries, my proximity not deterring them at all from their feast. Katharine Viner, now the Guardian’s global editor-in-chief, calls from London to offer me the editorship of Guardian Australia. It was a call I’d been hoping for but I was momentarily overwhelmed. We were still so small and the task was so big.
A few years before this, Nine Entertainment had been allowed to merge with Fairfax, concentrating the major media market even more than before, shrinking the sources of Australian news (other than us) from five to four – Nine, News Corp, Seven West Media and the ABC. The Australian Competition and Consumer Commission at the time it approved the takeover acknowledged that that deal reduced competition in the market for Australian news and information, but said it did not reduce competition sufficiently to be in breach of the law because “other players, albeit smaller, now provide some degree of competitive constraint”. The first listed “other player” was us, Guardian Australia. Could we be competitive enough to make a real difference?The first challenge was financial. With the digital platforms eating away at advertising revenue we launched reader revenue, a slightly counterintuitive plan to ask readers to pay for something when they didn’t have to. It is now more than half our revenue. A few years later, the news media bargaining code also forced Google and Facebook to the negotiating table to do deals with publishers, including us, a model that jurisdictions around the world are seeking to replicate.
The second challenge was how to grow. We needed more journalists and deeper expertise. Some generous philanthropy helped us bring forward expansion plans, we turned a profit and paid back the philanthropic loan that had helped us launch. We expanded a little more, and then the code helped us grow again. When we started we had about 20 editorial staff, about 40 in total including commercial and operations. Now that stands at 167.
The third, and by far the trickiest, was how to steer the course.
If readers were overwhelmed, perhaps some information could be framed differently, as explainers, inserts in stories to quickly give context and background, data representations, lists of the main points.
If some readers, young readers for example, were never going to come to our painstakingly curated news site, then we needed to present our stories as podcasts, or insert our news into their social media feeds. When we do, they consume it. There are ongoing regulatory questions about TikTok, but with minimal resourcing we are reaching more than 2 million viewers a month with quite serious topics. Last year, for example, we had 1.5m views to an explainer about the conflict in Tigray.
But how to navigate a polarised world, where we could factcheck misinformation all day and never report on the issues we consider important and where countering an untruth can help to draw attention to it. These are judgments to be made every day.
But there are also reasons for hope that we might be able to hang on to fact-based debate.
n Australia at least, the pandemic seemed to shock politicians out of their reflexive divisiveness and governments responded with an unusual degree of cooperation. Our job was made easier because, for the most part, Australian decision-makers followed expert advice and sidelined the shrill voices urging them to do otherwise.
And readers responded to the pandemic with a seemingly insatiable need for information – they reached for the best available facts and knowledge provided by all the mainstream news sites as something solid to cling to as our former lives and plans dissolved into uncertainty.And it turns out, this more measured fact-based discourse enhanced trust in institutions. Social cohesion, as measured by the Scanlon report and trust as measured by the Reuters Digital Media report, and our own Guardian Essential poll, all rose. People reported greater trust in all institutions, including the media, although the latest reports suggest the Covid trust bump might now be waning somewhat.
Meanwhile, Anthony Albanese has so far made a better fist than some of his predecessors of delivering on the promise to “change the way politics operates” by dialling down the temperature of the public debate, implementing his policies and empowering his cabinet, and the polls suggest the electorate likes this too.
Voters are also responding to the kitchen-table politics template set by Cathy McGowan when she won the seat of Indi in 2013, and followed with such stunning success by the teals in 2022; they like the grassroots doorknocking campaigning of the Greens, and the community conversation strategy that was so successful in the same-sex marriage plebiscites, and is being re-employed by the yes campaign in the referendum for a First Nations voice to parliament. Perhaps calmly talking things out and respectfully discussing differences can prevail over traditional campaigning tactics if we can bypass the commercial and technological pressures incentivising division.
And, as Judith Brett examined in her book From Secret Ballot to Democracy Sausage, Australia has a degree of inoculation from the polarisation infecting politics elsewhere by virtue of a system that compels people to at least turn up to vote. I’d add our shaky and very inadequate but still extant social safety net as another brake on the worst extremes of inequality that fuel an entirely justifiable mistrust in existing institutions and systems.
Taken together, I think there is room to hope that we can hold on to factual discourse in Australia, that evidence-based policymaking has somehow re-emerged from more than a decade of leadership coups and sloganeering.
AI and the tsunami of misinformation
But now it’s 2023 and I can sense another Polaroid moment coming. I’m just not quite sure what it is yet. Or whether the picture will be real.
Artificial intelligence, AI, is developing faster than we can keep up with it.
n essence, our ability to train computers to learn in a way that imitates human knowledge and behaviour is developing exponentially. In recent months a new form of AI has captured the global imagination. Generative AI tools synthesise new content from existing content, whether that’s text, video, code, imagery or audio.
They are built on Large Language Models which ingest billions of pieces of information and intellectual property from across the web, identify patterns and respond to our prompts, suggestions and questions. This gives them an incredible facility for language which is perhaps their most dazzling trick: we can talk to them and instruct them as we would another person and they talk back to us in fluid and seductive ways. Many of us would have experimented with one of those chatbots like ChatGPT (the initials stand for Generative Pre-trained Transformer). About 100 million people around the world have done so, and have been amazed at the incredible speed with which it produces highly plausible-sounding answers and boggled by the possibilities of the technology.
For us in journalism there are many potential uses and some very obvious dangers. It’s an opportunity and a potential threat and it’s happening whether we like it or not, and so very fast.
As a simplistic example exercise which I could factcheck very easily, I asked ChatGPT to write a profile of me. I must say it was very flattering, but it got a couple of things wrong, including my birth year, the year I started journalism, where I have worked, when I was awarded a Walkley award and what it was for, apparently I won the award for covering the Goss government’s first budget in 1995, when I have actually never covered Queensland politics and anyway the first budget was in 1990. The mistakes went on, culminating in the “fact” that in 2018 I was apparently awarded “the Press Freedom Medal by the Australian Journalists’ Association” in recognition of my “outstanding contribution to the defence of media freedom and the public’s right to know”, which would have been lovely but the AJA has not existed as a standalone entity since 1992. While I think the press council might occasionally award a similar-sounding medal, I have certainly never won it. It was all written in a highly credible emulation of real human prose.
But ChatGPT is not human, it has no critical faculty and certainly no commitment to the truth. It doesn’t know what the truth is. It is basically predictive text on hyper-steroids. As Melissa Heikkilä said in the MIT technology review in February: “AI language models are notorious bullshitters, often presenting falsehoods as facts. They are excellent at predicting the next word in a sentence, but they have no knowledge of what the sentence actually means. That makes it incredibly dangerous to combine them with search, where it’s crucial to get the facts straight.”
As our head of editorial innovation, Chris Moran, wrote: “The question for responsible news organisations is simple, and urgent: what can this technology do right now, and how can it benefit responsible reporting at a time when the wider information ecosystem is already under pressure from misinformation, polarisation and bad actors.”
The opportunities are exciting, but as Moran says, the concerns extend beyond how the media uses ChatGTP.
We are also worried about how ChatGTP and other programs use us.
For example, ChatGPT provides citations to books and academic papers and articles, but some of them are also made up, including citations to nonexistent articles allegedly published by the Guardian. These fabrications are termed “hallucinations” but that really is anthropomorphising things; the computer models do not have brains that temporarily lose their grip on reality, they have no actual understanding at all. This raises obvious questions, including how we respond to the fact that these large language models are crawling the web, including all of our licensed content, without recompense, to chew it up and spit it back out in a kind of huge information terrine that may at times completely falsify our work, or provide what appears to be an alternative to our work which is in no way the best obtainable version of the truth and may well be false.
Newsguard, a company that gives trust rankings to online news companies, found 49 websites that appeared to be publishing hundreds of articles a day that look like news stories but seem to have been written by AI with little or no human oversight, specifically to generate advertising revenue, complete with apparently AI-generated “about us” pages, disclaimers and copyright notices.The obvious potential is for a torrent of unreliable information that would leave Steve Bannon in the shade, vast quantities of articles, books and reviews so laced with mistakes and falsehoods that we completely lose sight of where truth lies and consumers of news and information are even more baffled and overwhelmed and potentially misinformed than they were before.
As Emily Bell, director of the Tow Center for Digital Journalism at Columbia University’s graduate school of journalism and Guardian board member, wrote this year: “The real peril lies outside the world of instantaneous deception, which can be easily debunked, and in the area of creating both confusion and exhaustion by ‘flooding the zone’ with material that overwhelms the truth or at least drowns out more balanced perspectives.
“This is all moving so fast it is more like a blur than a moment of clarity.”
Through all of this it seems the need for factual, quality news is greater than ever.
The flood of non-information makes the assembling and digging for facts, the critical analysis of information, the reporting of things that need to be known and would not be, but for our work, even more important than before.
We do it as part of a community of readers now, open to suggestions, correction and debate, we distribute our work in different ways, and we urgently need to find new ways to distinguish it and stand by it amid the emerging tsunami of misinformation. Unlike an algorithm, we care about where the truth lies and we must make the loudest possible case for its importance. Because the principles and essence and purpose of this job are the same as when I was a baby reporter walking into Old Parliament House for the first time and I am as excited by it now as I was back then. It feels good to be able to say that after 36 years.
I’m also privileged to lead a team of incredibly talented, innovative, determined and principled reporters and editors who very regularly have better ideas about how to fulfil our mission and confront these issues than I do.