Richard Gingras, Vice President of News at Google, delivered these reflections at WAN-IFRA’s World News Media Congress 2023 in Taipei. Gingras gave this speech on 29 June, the same day Google announced it would remove news links from its Search and other products in Canada when the Online News Act goes into effect in the country.
Thank you, to WAN-IFRA, to Vincent, Thomas, Cherilyn, Kah Whye, and the board. It’s an honor to be here. I appreciate the opportunity to share thoughts on the present and future of news in our societies.
I’ve worked at the intersection of journalism, technology, and policy for five decades – from first-of-their-kind digital news products, to pre-Internet and post-Internet online services, from leading an early search engine called Excite to working on a current search engine called Google.
No past experience comes close to the complexity of opportunities and challenges facing our digital societies today.
I’m not alone in this room (or maybe I am?!?) in having been alive long enough to remember earlier tranquil times. The era of the pre-eminence of print. Heady days of when rivers of classified advertising gold flowed deep and fast. When publishers controlled deep and fast rivers of distribution. We might also remember the sunny and optimistic epoch of the dotcom boom, when the audience reach of the Internet would fuel further growth and revenues. Then we realized the frictionless distribution of the Internet also created opportunities for others. Other publishers. Other marketplaces for products. For advertising. For information.
It would also enable diverse voices, in good ways and not-so-good ways. Some are angels. Some are not.
Those times were far from boring, but in the rearview mirror they look serene and innocent compared with what we see today.
I’d like to talk about three matters: the use of generative AI relative to news; matters of Internet policy, and a few bonus observations about the future of journalism in our societies.
—
Over the last six months the subject of AI has burst into our consciousness, stimulating our imagination, stoking new fears, causing us to puzzle through direct and indirect consequences. At some point, we hope, our minds will settle into a better understanding of the realities versus the surrealities.
I seem to have lost the battle of lexicon. I far prefer the more accurate term of machine learning versus the projectional nature of the phrase Artificial Intelligence. These are language models of probability, not creators of new concepts or intellectual ideas.
I also dislike the term “hallucination”. These engines of probabilistic expression may be modelled from human expression but they are not sentient. Calling errors in probabilistic expression hallucinations exacerbates the perception of sentience by adding a dark cloak of temporary insanity.
There is no doubt that the scale and complexity of the information space will expand further with machine learning and generative language models.
The technology is upon us. It has been for a long time. There are many language models, including many from sources we are not aware of. The need and the challenge is to develop better models to generate constructive content and help identify the malevolent content that will be created.
The tools we see today are, for the most part, language models, not knowledge models. They are predicting the next word in strings of text based on massive modeling of human expression.
That will change.
Language models will continue to evolve
Language models will be mapped against known “knowledge” models,
And the various providers of those models will evolve how they analyze and interpret the queries each of these tools are asked to address.
That last vector is at least as important as the others. At Google, query understanding has been key to our work with Search for twenty-five years — understanding the query, determining how best to answer that query.
There is a big difference in answering a mundane query like “how many ounces in a cup” versus “how should we address climate change”. Matters relating to personal and societal health require special consideration. Matters of competing societal perspectives are more wisely addressed with diverse voices who are expressing diverse perspectives.
This is the approach we have long taken with news queries in search that drive traffic to you. They will be core to our principled approach going forward. We are gradually working through the uses and implications of AI into our products. I expect that will be an ever evolving quest.
As we develop new Search and News products using generative AI, we’ll continue to work in collaboration with news publishers and hopefully prioritize approaches that will send valuable traffic to you.
Will the many available generative AI tools further facilitate the creation of misinformation and disinformation? Assume they will. That should not surprise us. The Internet does fully enable the breadth of free expression as codified within the world’s sovereign domains – from awful to lawful.
Yes, determination of the provenance of content, whether it be an image or a paragraph, will be key. In most cases, the more useful path will be determining the reputation of the source publishing that information. What signals measure a source’s reputation? How does context impact that assessment? How might we detect patterns of endorsement or, conversely, patterns of coordinated disinformation behavior?
This impact of generative AI on disinformation will add further urgency to regulate such behavior — which will engender its own set of challenging consequences..
AI and the Future of Newsrooms
Like all technology, generative AI has value, even as it intrinsically has no values.
There is value in generative AI.
Some of that value will be in gaining efficiencies. Automation will be applied to elements of content production ranging from summarization to headline creation to copy editing. But a recent WAN-IFRA survey noted it can play a role in topic ideation and research too.
This suggests it will also allow journalists to focus on more creative and complex work – investigative journalism, enriched media, what have you.
It can also help news organizations grow audiences by translating and transforming content into various languages and formats.
And it can offer value in generating useful service journalism, thus expanding a publication’s coverage of its community, and expanding the publication’s reach and value.
I think that’s particularly relevant in the domain of local news. I’ve worked closely with emerging digital players. A particularly successful one is Jeff Elgie and his ten-year-old company Village Media in Canada. He now owns profitable, ad-supported outlets in more than 30 cities. His sophisticated platform supports more than a hundred partners.
Village has developed and is now testing Generative AI tools to enhance newsroom output:
To suggest headlines optimized for search and social audiences
To craft story summaries for social posts.
To draft service journalism stories that follow their editorial policies and style drawn from a press announcement of a new restaurant or local business.
To draft narrative stories off of data, be that local police blotters or the box scores of high school basketball games.
Jeff’s objective is to expand how Village addresses the broad range of information desired by the communities Village serves. Jeff’s proven belief is that the key to trust and the path to sustainability is to become the “go to” information source for his communities.
He believes his news sites must go beyond accountability journalism to gain broad audience reach, reach that will generate ad revenue to pay for the journalistic work and broaden the audience for accountability journalism.
I began my career in public broadcasting. My mentor was Hartford Gunn. He founded PBS in the United States. He inspired the Ford Foundation to do the seminal work that led to greater funding, including taxpayer support, for public media.
In that day of Newton Minnow’s “vast wasteland”, the dire need for quality television was clear. But over time, Hartford struggled with public media being a silo of elite content for elite audiences.
I share that concern. I see many erstwhile digital news sources who focus solely on “serious accountability journalism”. That’s important.
However, I fear that ignores the larger need to enrich the very fabric of a community by addressing its overall information needs. I fear if we don’t bridge the divides in our societies, the important accountability journalism will not be heard beyond the depths of its own silo.
Public Policy
While the journalism community strives to rethink how to perform its role in our digital societies, the governments of the world are considering various approaches to how that digital society should be managed.
Many of these policies have been put forward by the news industry itself. While such policy initiatives may be constructive given the impact of the Internet, those efforts also stimulate a broad range of questions relating to their near and longterm impact on a free press, on free expression, and on the openness of the global Internet itself.
I have spent the last decade circling the world, working with journalists, working with publishers, working with various stakeholders on evolving Internet policy. It has been and continues to be an astonishing and illuminating experience. No past experience comes close to the complexity of challenges our digital societies face. Big complex questions. No simple answers. Challenges that are deep and structural. We need to be thinking in years and decades, not weeks and months. Quick fixes are often no more than wishful thinking. At worst, they may result in damaging the foundations of the free press itself.
In my global travel I’ve learned many lessons. I’ve learned that perspectives on free expression vary. I’ve learned that approaches to governance in open societies vary.
I’ve learned it’s unwise to suggest that the United States codification of free expression with its First Amendment is the only and correct one.
I’ve learned it’s unwise to imply that the United States form of democracy is the only and correct one. Societies are different. The challenges are different. There is no one-size-fits all set of answers.
Yet I understand that both you and your governments are understandably impatient. Societies right now are fragile. In a span of a month I visited two countries whose capitol buildings were stormed by rioters angry about election outcomes. There are broad declines in measures of global societal stability. Governments are worried about their foundational legitimacy. It is a tough decade, a tough century to be a true statesman or stateswoman.
While no one would argue that governments have the right and responsibility to craft Internet policy, we should recognize the potential consequences of such policy — on matters of free expression, on matters of freedom of the press, on matters of human freedom.
At core, Internet policy is about a government’s approach to managing the information space of its society. Pursuit of digital policy raises several paradoxical questions:
How does one manage free expression in a digital world?
How does one manage press freedom in a digital world?
I believe that last question can and should be refined further: what flavor of press freedom do you want for your society?
My personal perspective presumes that a truly open and independent press is one where the role of a journalist isn’t defined by law, and where the independence of a free press can best be accomplished with diverse sources of financial support, not a dependence on one.
But, my opinion does not matter. In truth, Google’s opinion does not matter. Governments will decide.
What flavor of press freedom do you want for your society?
Around the world, there are an array of evolving approaches to Internet policy. Each in its own way will impact how support for the press is accomplished. Each in its own way will impact press freedom itself.
The devils, and or the angels, are in the details. The details matter. The intricate “design decisions” matter. A closer inspection of each might yield more intentional outcomes:
Is your flavor of press freedom one where press freedom is managed by a press association or associations? What principles do the associations stand for? What are their journalistic standards? Who adjudicates those standards? How does the association represent or not represent the emergence of new voices and new players?
Is your approach to the financial support of press freedom built upon copyright structures? Intrinsically, these approaches cannot and will not differentiate on matters of quality or topicality. Will they stimulate the important journalism many feel is desired? Or will they largely and disproportionately support high-volume producers of content, including content that some will consider misinformation?Will such all-content-is-equal approaches disadvantage those who seek to produce quality journalism for their communities?
Is your model of support for press freedom addressed by bargaining codes between the press and specific tech companies? Will bargaining codes fuel change and innovation versus supporting embedded players? Will the outcome be neutral and avoid partisan outcomes? Might bargaining codes create disproportionate distributions similar to copyright structures? Will the process have sufficient transparency to avoid suspicion that the scale of determined indifference has been skewed? Will it undermine the nature of the web as an open ecosystem?
Is your flavor of press freedom enhanced by a government regulator? What is their role? How does one define what journalism is or who does it? How deep into the inner workings of the press might the government go before it breaks the independence of the press we say we want to preserve?
Can one’s definition of press freedom be maintained as legal mechanisms are crafted to address misinformation? Will such laws provide detailed determinations of the nature of disinformation such that it can be identified and curtailed? Can such approaches be effective if exemptions are granted for politicians and/or media? How do we address the overlap between what Jamie Susskind describes as the “hinterland of naughtiness” with a walled garden of legal expression? Where does one draw the lines between awful versus lawful? Might such mechanisms just as easily be used against the press by less well-intentioned leaders?
What flavor of press freedom do you want for the society you live in? What checks and balances must exist to ensure that the letter of the law is not warped over time by the parochial interests of a politically influential few?
I believe that public policy for the direct financial support of the press should not have singular companies, like Google, in the middle of it. Indeed I struggle, as I said recently in testimony before the Canadian Senate, with the presumed wisdom of any country being comfortable with notable financial support for that country’s press being intrinsically linked to a specific private company or companies.
Do NOT interpret my statement to suggest Google is not in favor of further support for news or that Google should not be part of the solution. We have worked to be part of the solution. We will continue to do so.
What might be the better approach to gathering and dispersing financial support for journalism — if that is what a society deems important to address? The answer has a huge impact on the larger question of what flavor of free press do you want in your society?
From Google’s perspective, an independent framework for the distribution of financial support might consider the following principles :
A framework whose governance does not include Google, or other sources of funding.
A framework that is driven by thoughtful criteria that can support the creation of quality journalism for the communities journalists serve.
A framework that can support innovation in journalism.
A framework that can support the digital transformation of legacy providers.
A framework that can enable the emergence of new voices, which would seem critical in our fast-changing digital societies.
A framework whose financial resources might be more durably drawn from a broader class of activity than singular companies whose prospects may change over time. Who knows what the Internet looks like in 10, 20, 50, 100 years? A regulatory regime for a space as important as the press must be more durable than the current business models and margins of a couple of companies.
Our recent approach here in Taiwan with the Digital Co-Prosperity Fund adheres to many of those principles, one difference being that Google is currently the only financial resource. We hope others that support its mission will join us. It was thoughtfully crafted with broad stakeholder participation and the endorsement of the government.
But again, that’s not for Google to decide. It will be decided by individual societies and their governments. It will be decided, I hope, with the informed influence of the journalism community and of civil societies, an informed influence that considers the long term view of press freedom and its future.
We are at a seminal point in the evolution of our digital societies. We are at a seminal point in the role of the press in those societies. May we consider these questions with the thoughtfulness and care they deserve.
Relevance
Journalism is facing a crisis of relevance.
How can journalism play its role in enabling informed societies if its efforts are not perceived to be relevant, useful and valued?
The Reuters Institute tells us less than 10% of our societies regularly consume what we might call serious news. Google tells us less than 2% of Search queries are about matters of news.
How can we reach those who don’t care or lost interest? We hear it from friends. They avoid the news. It makes them sad, or anxious, or fearful. They find solace in other ways, bingeing the latest on Netflix or feeding their addiction to TikTok. I do as well.
How might we adapt to the media forms our cultures are adopting?
The underlying assumption of a democratic society and the role of journalism, is this: IF we express our ideas with the right words and logical arguments, IF enough people read those words, THEN our democracies will be effective, the world will be a better place.
Again, the Internet and new media forms have rearranged social, political, and cultural structures. We see it with social media. We see it with short-form video. The messages get shorter. An inescapable progression, or digression, of how we communicate, how we understand the society we live in. We can’t ignore it.
How might we better understand how our societies value or not value serious journalism? Do our audiences understand the role of journalism? Do they know which sources to trust with their precious time and money? Could the explosion of inexpensive but popular news and opinion smother the credibility of fact-based coverage? Is the drift toward partisan news making the problem worse?
Might we rethink the models, the formats, the linguistics used in our journalistic work to allow our societies to see journalism for it’s the more constructive role it can play it societies?
The word “constructive” is key. It’s not news that makes you “feel good”. But news that goes beyond the typical coverage model, with clear signals and clear intent, to include the necessary context, the hows and whys, and importantly, a consideration of how the calamitous event could be prevented.
Might we avoid linguistics that drive divisiveness and exacerbate news avoidance? Might we evolve journalism to be seen as it could be: as a constructive advocate in addressing a societies challenges and its information needs. Do hot, confrontational chat shows with names like Crossfire diffuse or drive divisiveness. Doesn’t it matter? Politicians study the nuances of linguistics. Journalists seem less inclined to do so. That seems unwise.
—
Two closing observations.
First, the relationship between Google and the news industry is wobbling on and off track. It is my hope that it can return to a more collaborative than combative equilibrium. I’d suggest that would be beneficial as we mutually address the evolution of meds and technology over the short, medium, and long term.
I fear the demands of the news industry are stretching beyond workable for Google. I fear the underlying policy constructs we are seeing, for instance in Canada, will have untoward secondary consequences. I fear that such legislation and the parallel evolution of new AI-driven product models may put in doubt the very viability of a search engine operating against the open Web. I have a hard time seeing how that’s a good thing for press freedom or for the long term benefit of the creation of quality journalism.
Last observation: I’ve been reading a lot of history of late. Always interesting. Always the same. At every opportunity we decline to learn from experience. Instead, we determinedly repeat, indeed acclaim, past misbehavior. It is rather damning of the future of our species.
Today people speculate about the dangers of the Internet, or more recently, the dangers of AI. It would be more accurate to admit that we are concerned about the dangers of ourselves, not the machines. The AI models being built are learning from the vast expression of humankind, both principled and not, sometimes in quest of the common good, often not. But that is the history of our species, is it not?
Regulatory approaches can fall victim to the same dynamic. What is the underlying principle being pursued? To what extent are those principles truly about the common good versus specific or self-interests?
We must think beyond the memes.
We are at a seminal point in the evolution of our digital societies. We are at a seminal point in the role of the press in those societies. May we consider these questions with the thoughtfulness and care they deserve.
We will soon publish a separate summary of the Q&A that followed Gingras’ speech.
The post ‘We must think beyond the memes’: Google’s Richard Gingras on the future of news in our societies appeared first on WAN-IFRA.