At the Associated Press, the news department leaders were the first to suggest trying artificial intelligence. They were motivated by two mega trends in the media business: “the relentless increase in news to be covered and the human constraints associated with covering it.”
In 2013, the AP teamed up with Durham, North Carolina-based Automated Insights to automate the production of certain types of news stories directly from data. They started with sports and then extended the initiative to corporate earnings reports. Since then, the AP has continued to experiment with AI and recently published “A guide for newsrooms in the age of smart machines” based upon on its own learnings as well interviews with dozens of experts in the fields of journalism, technology, academia and entrepreneurship.
What they’ve found is that “artificial intelligence can do much more than churn out straightforward sports briefs and corporate earnings stories. It can enable journalists to analyze data; identify patterns, trends and actionable insights from multiple sources; see things that the naked eye can’t see; turn data and spoken words into text; text into audio and video; understand sentiment; analyze scenes for objects, faces, text or colors — and more.”
Among the key takeaways in the report is that, in the field of journalism, AI has potential to:
Attend to menial tasks and free journalists to engage in more complex, qualitative reporting.
Enhance communication and collaboration among journalists.
Enable journalists to sift through large corpuses of data, text, images, and videos.
Help journalists better communicate and engage with their audience.
Empower the creation of entirely new types of journalism.
In addition to providing insights into the practical journalistic applications of AI, the report covers the relevant technologies in this field—including machine learning, supervised learning, natural language processing, robotics, computer vision, and more. It looks at the potential impact of AI on journalists and journalism. It also considers ethical, philosophical and practical implications of implementing AI within media environments.
Above all, the report focuses on the potential for AI-human “collaboration” in which journalists are freed from many of the craft’s more mundane or repetitive tasks. Rather, leveraging AI can save organizations time and money, while being better equipped “to keep pace with the with the growing scale and scope of the news itself.”
Signal Media co-founders: Miguel Martinez and David Benigson
In December, Hearst Ventures participated in the Series A round of funding for Signal Media, an artificial intelligence-powered information company that helps its clients monitor the world’s news media. Signal CEO David Benigson offers insights into the platform’s technology, which is designed to help businesses make smarter and faster decisions.
How does Signal track and monitor news cycles across global markets?
David Benigson: Signal is an artificial intelligence company with the goal of transforming the world’s news into accessible, actionable bits of knowledge. We apply cutting-edge technology that we have developed here, in house, to enable clients to monitor the news for whatever they deem important: company mentions, client news, trending storylines and more. Our motoring tool analyzes news in real time across over 100 markets and 40 different languages. We are trying to bring clarity to the news and get users the information they need to know, along with information they didn’t even know they needed.
What types of clients does Signal serve?
Benigson: We have found success with clients across a range of fields—from big financial institutions to communications firms. One of our initial areas of focus was transforming the way public relations departments work and receive coverage concerning their brands. Increasingly, we are seeing the value of our service diversify into areas that are beyond the scope of our initial plans, including client intelligence, horizon scanning, and regulatory change.
Ultimately, we enable our users to search for companies, topics and themes of interest. This provides them with the ability to track their reputation, understand wider market insights and operate more effectively across the board.
How do you see clients most effectively using Signal on a daily basis?
Benigson: Let’s take a big wealth management firm as an example: Previously, they were only able to track mentions of themselves in the news. Today, they are able to not only monitor each of the subdivisions of their very large company, but also monitor their competition, key spokespeople, clients and key trending topics that impact the regulation of their business. And they are able to do all of this within our single platform.
What we have been able to do is allow our clients to have specific and narrow searches which provide only relevant information. Our search results give a holistic view of the interests of the company and the sphere in which they are operating. Additionally, users are getting this information in real time and from all across the globe.
Can you share some details about the experiences that led you to create Signal?
Benigson: Initially,I studied law and then worked with a few startups. After that, I worked with chef Jamie Oliver, who himself created a media company. I founded Signal when I was 24 years old, so I had very little direct managing experience. The process of launching a company has been both amazing and challenging. I’ve learned so much over the past three and half years as the company has scaled up.
The original idea for Signal came from speaking with people in the industry about how they were receiving news every day and what tools they were using to obtain that information. It rapidly became clear that there was a big gap in the market for a tool that could more effectively help people make sense of the vast and ever-growing web of information available online.
What makes Signal different from the other services that are offered in this field?
Benigson: It really starts with our artificial intelligence technology. Signal is looking to automate things that have, up until now, been done manually. Artificial intelligence means that, in a sense, we can be uniquely ambitious. There are millions and millions of new documents added to the internet each day, and processing that data and connecting that information to users in real time is only possible because of our powerful, intelligent technology.
On top of that, our user experience and customer service is a real draw. We have spent a lot of time thinking about what happens when a client is trying to work with Signal, and we want to make that process as frictionless as possible. Our strength is that we are able to combine the power of our technology with the human experience of using our platform. Because of that, our product delivers an unparalleled experience that sets us apart from the competition.
Benigson: We have gone through quite seismic changes since developing the initial concept, and we are now servicing around 100 corporate clients. Signal has concentrated on adding value to our users by identifying what they continue to struggle with when it comes to monitoring their brands. We want to provide end-to-end solutions, so we continue to seek user feedback. Signal has employed user-centric design processes that ensure we have regular interactions with our clients—this all feeds directly into the product development process.
Who makes up the Signal team?
Benigson: Our team has grown to around 50 people. Two years ago there were only 10 of us, so we are expanding rather quickly. Half of the team is made up of product developers, tech engineers and data scientists, and the other half is made up of sales, marketing and client relations employees. Our latest round of fundraising will allow us to invest in both of these key areas at a larger scale.
How are you working to secure more clients based outside of the U.K.?
Benigson: Signal has a small operation already running in New York, but we want to continue scaling that up. We see the U.S. as being the largest and most attractive market for us, and we are extremely keen to make that work. We are always working with our sales and marketing teams to improve how we reach people outside of our current core areas.
What’s next for Signal in 2017?
Benigson: We are continuing to expand the data sets that we use, including in the legislative and research fields. The broader the selection of data we aggregate, the more we are able to apply our products to people in large organizations. We are also looking to launch new products on top of our core platform. Right now, we are gearing up to launch a specific product that allows our clients to track changes in regulation as they happen, helping them to remain compliant. We also have a mobile app that we are preparing to launch, as well as a few new tools within the platform itself.
From the Signal perspective, what are some of the biggest benefits to your partnership with Hearst?
Benigson: A big part of Signal’s platform is how we leverage news media, and there is no better organization at the edge of innovation in news distribution than Hearst. They have offered unparalleled strategic advice when it comes to our expansion and how to leverage technology to get the most value. I also think that when looking at U.S. expansion, Hearst will be a key player in helping us build a network. We are really excited for the opportunity to work with, and learn from, the Hearst Ventures team.
Robots aren’t taking over the media (yet) but Artificial Intelligence and automation are lending a helping hand.
The challenges of running a profitable media company have resulted in a “streamlining” of these organizations, often requiring staff reductions that create holes in coverage, particularly in areas that require domain expertise or a nuanced interpretation. What if there was a way to free up the increasingly valuable reporters from some of their tedious or time-consuming tasks to focus on telling complex stories or digging into subjects deeply? Enter Artificial Intelligence (AI) for journalism. An increasing number of media companies are employing robot journalists—or, more accurately, artificially intelligent algorithms adept at analyzing data and writing serviceable prose – to do just that. (Fusion’s Kevin Roose does a terrific job explaining how it works.)
Here are some of the ways AI is making its mark on journalism:
The Associated Press, working with Automated Insights and Zacks Investment Research, automatically generates almost 4,000 stories about U.S. corporate earnings each quarter—about twelve times what AP reporters and editors previously created. For a small fraction of these stories, editors add “the human touch” to provide more nuanced interpretations.
More recently, the AP added sports reporting to its automated journalistic endeavors. The move allowed the company to expand its coverage of Minor League Baseball (and could offer similar possibilities for other sports coverage). AP baseball editors and reporters worked with Automated Insights to configure the Wordsmith platform to ensure that it would follow AP standards for baseball coverage.
The Washington Post is also taking a crack at automated journalism, developing its own proprietary artificial intelligence tool, “Heliograf,” which automatically generates brief, multi-sentence updates. The Washington Post will roll it out to report key information from the 2016 Rio Olympics, including results of medal events. Heliograf, which will continue to be developed by Post engineers to enhance storytelling for large-scale projects, will also be leveraged for data-driven coverage of major news events, including the U.S. election. This technology can process a combination of different data sources, such as crime and real estate numbers, customize stories depending on individual user actions, and help look for anomalies in data to alert journalists to a potential story—all of which The Washington Post plans to leverage as it further develops Heliograf’s utility.
Algorithmic journalism is not limited to sports and finance, however. One of the longest-running robo-journalists in the game is Quakebot, which was written by Los Angeles Times Writer Ken Scwenkie. According to Business Insider, Quakebot is an automated system that resides on the Los Angeles Times’s servers, which receives emails from the US Geological Survey, runs that data through an information checklist, and then determines if it’s newsworthy based on the magnitude. It then parses out content from the email and inputs it into the Los Angeles Times’s content management system. Posts are structured using a formula based on previous posts.
It is interesting to note the LA Times also uses its own brand of automated journalism for The Homicide Report, which tracks every homicide victim in LA County. This sort of approach (visualized data journalism) is particularly suited to the algorithmic approach in that analyzing huge (and possibly constantly growing or changing) data sets is a massive undertaking. With solid structure and oversight, these can be reliably created, updated and presented in a dynamic visual way by leveraging AI.
The momentum for automated journalism has been building for years now and is definitely picking up steam. The New York Times reports that Tronc, Hearst, Bonnier, Gannett and the Weather Channel use automation to do everything from streamlining video production to analyzing and summarizing text and finding photographs and video clips to go with it. And, in an April memo to Bloomberg’s staff, Editor-in-Chief John Micklethwait announced that the news organization is creating a 10-person team to determine how automation can be used throughout the company’s portfolio of editorial products.
As Dr. Andreas Graefe, author of the Tow Center for Journalism report “Guide to Automated Journalism” noted, automated journalism is particularly useful in generating routine news stories for repetitive topics for which clean, accurate, and structured data are available. He also outlines the potential for algorithms to generate large volumes of news quickly, potentially in multiple angles and personalized for reader preference.
But, as the report points out, there are limitations to automation that must be carefully considered as more organizations use artificial intelligence to augment their journalism. Given many of the current implementations reliance on data, the quality of that data is paramount. And, as Micklethwait points out in his memo, “one irony of automation is that it is only as good as humans make it.” That, of course, extends to the media organizations that deploy algorithmic approaches, which must balance efficiency, transparency, and the consumer demands of more information, faster—but who also love a good read.
Artificial intelligence can help journalists automatically adapt stories to the
personalities, moods and locations of their readers.
The news and information ecosystem is in the midst of change — again.
Mobile-first consumption is on the rise, smart homes are becoming mainstream and connected cars will soon take over the roads of major cities around the world.
Smart devices will require “smart content”. It’s only a matter of time before artificial intelligence becomes the backbone of the media industry of the future.
A change in how we interact with news Today, most people find information via search or social. And while these two channels are radically different in functionality, they have one thing in common — any given article surfaced through these platforms is exactly the same for everyone in the world.
Content today is one size fits all. And why wouldn’t it be? A journalist writes a story hoping to reach as many people as possible.
Search and social help tailor information choices to individuals to a degree, but Google, Facebook and Twitter know that artificial intelligence will fundamentally change the equation. That’s why, since 2013, these companies have been investing substantial resources into the space and acquiring startups.
In Facebook Messenger, for example, several news organization such as CNN and The Wall Street Journal are already using bots and some level of automation to deliver news through the platform.
Beyond automation: What is artificial intelligence? Artificial intelligence understands the environment it operates in and performs certain actions as a result of it. AI seeks to learn what its users want and how they want it.
In the specific case of news media, articles can be processed through algorithms that analyze readers’ locations, social media posts and other publicly available data. They can then be served content tailored to their personality, mood and social economic status, among other things.
AI allows journalists and media companies to create infinite versions of an article, resulting in increasingly relevant information that speaks directly to individuals — ultimately forming a more engaged audience.
Examples of artificial intelligence, machine learning and automation in content creation Crystal is a program that adapts emails you write to the personality of recipients. For example, if you’re sending a note to a more laid-back person, the software suggests a change in tone from a formal introduction such as “Dear John” to a more colloquial “Hi” or “Hey.”
Crystal uses previous emails to that recipient as well as their social media posts to recommend certain language, tone and sentiment. It’s easy to see how this approach can be adapted for a newsroom –in fact, it can build off of pioneering efforts already underway such as automated earnings reports by The Associated Press.
Artificial intelligence can even localize stories. If you live in California, you might not read a story entitled “Texas residents poisoned by toxic waste plant for years.” But if the story included an automated note highlighting a similar past incident in your city, you would probably be more inclined to look at it.
The future of journalism powered by artificial intelligence
Beyond tailoring content to users, AI can help journalists do more investigative work by analyzing massive sets of data and pointing to relationships not easily visible to even the most experienced reporter.
While this technology can improve efficiencies in newsrooms, though, it should work in tandem with journalists, not replace them. Going forward, the challenge will be to make sure we continue to adhere to our standards and ethics.
What will the future of news will look like when it becomes powered by AI? We will soon find out.
I’m the Strategy Manager for The Associated Press and fellow at Columbia Journalism School. I write about media, storytelling and innovation. Let’s connect. (@fpmarconi)