Fix Journalism Using AI
[1st quick draft]
Problems with the news industry have been present from the time our country was founded. The author of our Declaration of Independence wrote:
I deplore, with you, the putrid state into which our newspapers have passed, and the malignity, the vulgarity, and mendacious spirit of those who write for them; and I enclose you a recent sample, the production of a New England judge, as a proof of the abyss of degradation into which we are fallen. These ordures are rapidly depraving the public taste, and lessening its relish for sound food. As vehicles of information, and a curb on our functionaries, they have rendered themselves useless, by forfeiting all title to belief. That this has, in a great degree, been produced by the violence and malignity of party spirit, I agree with you; and I have read with great pleasure the paper you enclosed me on that subject, which I now return. It is at the same time a perfect model of the style of discussion which candor and decency should observe, of the tone which renders difference of opinion even amiable, and a succinct, correct, and dispassionate history of the origin and progress of party among us.
- Thomas Jefferson Letter to Doctor Walter Jones, January 2, 1814
Yet despite their flaws Jefferson viewed them as crucial:
The basis of our governments being the opinion of the people, the very first object should be to keep that right; and were it left to me to decide whether we should have a government without newspapers, or newspapers without a government, I should not hesitate a moment to prefer the latter. But I should mean that every man should receive those papers and be capable of reading them.
-Thomas Jefferson Letter to Edward Carrington, 16 January 1787
Declining Product Quality Is a Market Opportunity
Unfortunately, in an era of innovation improving many industries, the news media has been declining in quality and losing money by not finding a way to successfully adapt to technological change.
A study by Gallup and the Knight Foundation found that in 2020 only 26% of Americans reported a favorable opinion of the news media, and that they were very concerned about the rising level of political bias. In the 1970s around 70% of Americans trusted the news media “a great deal” or a “fair amount”, which dropped to 34% this year, with one study reporting US trust in news media was at the bottom of the 46 countries studied. The U.S. Census Bureau estimated that newspaper publisher’s revenue fell 52% from 2002 to 2020 due to factors like the internet and dissatisfaction with the product.
A journalist explained in a Washington Post column that she stopped reading news, noting that research shows she was not alone in her choice. News media in this country is widely viewed as providing a flawed product in general. Reuters Institute reported that 42% of Americans either sometimes or often actively avoid the news, higher than 30 other countries with media that manage to better attract customers. In most industries poor consumer satisfaction leads companies to improve their products to avoid losing market share. When they do not do so quickly enough, new competitors arise to seize the market opening with better products.
An entrepreneur who was a pioneer of the early commercial internet and is now a venture capitalist, Marc Andreessen, observed, the news industry has not behaved like most rational industries: “This is precisely what the existing media industry is not doing; the product is now virtually indistinguishable by publisher, and most media companies are suffering financially in exactly the way you’d expect..” The news industry collectively has not figured out how to respond to obvious incentives to improve their products.
The problem may be partly that for many years most national and local media markets in the United States were monopolies or oligopolies. Business cultures were entrenched that did not need to worry about change, so they were unprepared to adapt well to the rise of the internet. One author suggests the switch to internet business models that focused more on subscriptions than advertising reduced the pressure advertisers placed on publications to appeal to a broad audience.
Pick One: Niche Publication or General Audience
Many internet e-commerce ventures use the internet’s ability to target niche markets as a competitive advantage. Publications may have followed this strategy without fully considering its implications. Instead of writing content that strives to be more objective and neutral, which is a difficult goal, many journalists have opted for content that shows bias and spin, which appeals to a loyal audience. This allows the easy creation of marketing clickbait that emotionally attracts one political demographic. However, this approach undermines the public’s trust in the news media, which could be restored by providing more calmly neutral coverage that appeals to a broader audience, but does not lend itself to clickbait style marketing.
There is a market for such publications targeting an individual political faction, in fact some media outlets may find a greater audience by just admitting they are targeting a particular biased demographic and explicitly marketing to them. The public will likely be more willing to trust outlets that are honest about the nature of their content. However other outlets might take notice of the large market opportunity presented by most of the public that prefers less biased news.
Regardless of whether the level of bias or quality has indeed changed in recent decades: the internet allows people to contrast the same news story from varied sources to spot any comparative bias or flaws they would have been unaware of in the past. Twenty years ago, author Michael Chrichton suggested most people suffered from what he termed the Gell-Mann Amnesia effect. People would read a news article on a topic they were well informed about and decide “the journalist has absolutely no understanding of either the facts or the issues. Often, the article is so wrong it actually presents the story backward—reversing cause and effect.”
Then they would go on to read an article on a topic where they had no expertise and forget their concerns about the publication’s quality and trust it as if it were probably completely accurate. Perhaps they implicitly rationalized they had no way to know for sure about the content and may as well hope for the best. They did not think it worth the effort to explore alternatives in the era before many options were available at their fingertips.
Now people can merely take a moment to view other online news sources to see how coverage of an issue compares and spot flaws in articles outside their areas of expertise. No matter how well intentioned they are, journalists are fallible like all humans. They are not experts in everything and have biases and blind spots, and the public can discover this now.
Writing For Multiple Political Languages
A study confirmed prior work indicating Americans in general tend to have a flawed, distorted understanding of the views of those with political world-views different from their own. They referred to a “Perception Gap” between what people perceive their opponents think, and what they actually think. Oddly this research found those who read the news “most of the time” had perceptions “nearly three times more distorted” than those who read the news infrequently, “only now and then.” They reported “Democrats without a high school diploma are three times more accurate than those with a postgraduate degree.”
Journalists of course are likely to be educated frequent news readers and are not immune to these issues. Even if they were, they are still going to do a better job expressing views they truly believe in, regardless of any attempt to be neutral. Consumers benefit from having those who truly believe a particular political perspective ensure an article accurately represents those views.
John Stuart Mill, the most influential English language philosopher of the 1800s and a key figure in classical liberalism observed:
Nor is it enough that he should hear the arguments of adversaries from his own teachers, presented as they state them, and accompanied by what they offer as refutations. This is not the way to do justice to the arguments, or bring them into real contact with his own mind. He must be able to hear them from persons who actually believe them; who defend them in earnest, and do their very utmost for them. He must know them in their most plausible and persuasive form; he must feel the whole force of the difficulty which the true view of the subject has to encounter and dispose of, else he will never really possess himself of the portion of truth which meets and removes that difficulty.
Teachers who understand a topic well, and experts in specialized fields who need to explain concepts to the public, struggle to deal with what is called the curse of knowledge or the curse of expertise. They take concepts for granted that their audience may not have learned. Once you understand a topic well it can be difficult to think through how to explain it to someone who does not have all the same background information you possess.
The studying you have done may lead you to change some assumptions about the world that the less informed public hasn't had reason to change. That can lead the audience to misunderstand you. You may use phrases that have specialized connotations within a field that differ from how the public uses them, misleading them subtly.
Similarly, those who hold a particular political point of view may struggle to communicate well to those with a different political worldview that makes different underlying assumptions about how society works that they take for granted. They may use language with subtly different connotations for the same words, have different underlying values, experiences and interests they use to interpret what others communicate. A liberal may struggle to write a story that connects well with a conservative audience, and vice versa.
If someone who speaks only English tries to communicate with someone who speaks only Russian, they are at least acutely aware of the communication gap and focus on bridging the divide. Many political discussions in our time seem unproductive because each side may be speaking English, but they cannot seem to get through to each other due to underlying world-view differences. Even journalists who are concerned with advocacy rather than neutrality should be aware that to persuade their opponents, they may benefit from assistance in translating their work to appeal to those who think differently.
Human Or Machine Translation Assistance
One way to reduce the chance a political news story is viewed as biased is to have people who believe each point of view involved in creating the article to ensure their opinions are accurately represented. Some software developers use a practice called pair programming where two programmers work together at the same workstation to try to achieve better results than each would by themselves. News outlets might consider having pairs of people with different political views work together on stories to reduce bias, but that may not be practical with if their staff does not contain the right mix of viewpoints, and it is unclear how it would impact operating costs.
A more pragmatic approach is for them to adapt the idea of pair journalism to work with AI systems that play the role of someone with a different political viewpoint to help nudge articles towards neutrality. AI systems were trained on massive amounts of text that include ideas from across the political spectrum. Search engine company Neeva created an AI system demonstrates they can make use of this knowledge. Neeva provided a slider control for users to indicate what political perspective they wished to see, and listed news stories from that perspective and summarized them. AI chatbots and writing tools exhibit this knowledge when you tell them to create content from a particular political perspective.
These systems are not yet perfect, but they are good enough to be useful and will improve. One professor suggests treating them as interns that are still learning so their work needs to be double checked.
Unfortunately, many people are concerned the fine-tuning AI vendors do to make their systems more user friendly has introduced political biases. Even if the underlying data they were trained on contains myriad viewpoints, the vendor’s guidance may over-ride the AIs ability to tap into that knowledge. On some topics the AI may try to stick to the viewpoint the vendor has chosen for them rather than viewpoint you would like assistance with. Concern over these issues may lead to the creation of new AIs that each provide a different perspective, or generic AIs more willing to express any of the political views contained in the vast body of knowledge they were initially trained on.
Create AIs Trained for Journalism
Writing a news story has different challenges than writing about topics where the AI has access to the full story from either the history it was trained on or from reading the web. An AI system needs to rely on a reporter to inform them of all the details. Without other information it can't know for certain if anything is missing or inaccurate. A journalist may unintentionally exhibit bias when they neglect to mention negative news facts about their side, or positive facts about another side, or they are otherwise biased in their presentation of the basic facts to the AI. Systems may be trained to spot hints that the reporter hasn't given them a full unbiased story so they ask questions of the reporter to ensure they have not accidentally left anything out.
AIs can be trained on how to coach journalists in general, not merely how to help reduce bias. If the reporter is not the first one to break a story, future AI tools might be adapted to read other publicly available news coverage on the topic to search for signs of discrepancies, and to nudge a journalist towards improving a story and writing follow-ups.
On many topics there are more than two points of view, which can provide pair journalism with AI an advantage over pair journalism with two humans. There are multiple factions within the two major political parties, almost half the public does not identify as belonging to either major party, and some belong to minor parties. Given so many perspectives, it is unclear what would constitute an appropriately “neutral” story.
Rather than one “neutral” story, in some cases an AI may interact with journalists to create multiple separate articles on the same news item from varied perspectives. A friendly user interface for consumers, perhaps including an AI that learns the user’s preferences, could either provide readers one of those static articles the journalist & AI team created, or the reader’s AI could come up with an individually tailored version of the story that matches the user's preferences and knowledge. A user who does not know much about foreign policy might benefit from background information being added to the article. Someone who has followed an ongoing story closely may not need a recap of past events the default version of a news story contains.
Neutral organizations outside the news media can use AI to help the public assess the bias of media outlets. If they confirm a news source has become less biased, that may help restore consumer trust to earn back an audience. No solution is perfect, but adopting AI can be a step in a productive direction.
It is in the Media’s Self Interest to Change
The staff of news outlets obviously chose to evolve the media into what it is today. Some entrenched industry factions may not wish to change. Many journalists would like to be political advocates and feel they are advancing their cause by letting bias enter their work. However, that approach has led to a declining audience and lack of trust, which may undermine their goal.
If advocates wish their journalism convince political opponents to change their minds, they need those opponents need to see their work. They need to convince their opponents to trust their media outlet. Advocates should be interested in using AI (or other humans) to help them explain stories in ways their opponents will find of interest. If journalists truly believe their political views are right, they should let the public make up its own mind after seeing all sides reported.
Journalists who favor advocacy in their work need to consider whether it should be up to them to decide what is best for the public to believe by withholding other viewpoints from them. If the public does not trust that they have been given the whole story, they may be less likely to side with the journalist’s politics.
AI Can Help Local News Sites
Many areas have no local news outlet, or only limited news options. AI tools can provide a new local news site a greater chance of being able to earn enough to operate by lowering costs through making the staff more productive.
It can also enable new types of features for existing outlets. AI tools can enable fuller, or more efficient, coverage local government meetings and other public forums. Audio recorded at events can be fed to AI transcription services, or new services being created to analyze recordings of government meetings, so AI assistants can summarize meetings for news stories, and provide the whole transcript for anyone interested. AI systems can help comb through local government and publicly available corporate documents looking for potential problems to investigate.
Most of what people turn to local news for does not require investigative journalists to go out and find stories. Much of the content comes in from the public in press releases and volunteer created content that government entities, businesses, and members of the public wish others in the community to hear about. AI tools can help sort through all that incoming information more efficiently, and optionally allow the public to see more of it than was practical in the past when there was not enough human staff time to handle it. AI tools may be flawed: but many consumers would prefer intern quality news about something to no news at all. AIs can be trained to allow consumers to assist a news site in correcting any mistakes they spot.