Imagine, if you will, a writer sitting at their desk, pen in hand, composing a masterpiece. The words flow from their mind and onto the page, each one a unique and beautiful snowflake, a testament to their individuality and creativity. Now contrast that with a machine, churning out words with cold, calculated precision, devoid of emotion or personal experience.

As we read an article written by AI, we are confronted with a worldview that is fundamentally different from our own. While we as humans are bound by our emotions and personal experiences, the machine is unfettered by such limitations. It is a strange and unsettling sensation, as if we are looking at the world through a distorted lens, one that is both familiar and yet fundamentally alien.

But the consequences of acquiring information that has been machine generated go beyond simply the way it makes us feel. It forces us to confront the very nature of knowledge itself and ask ourselves what it truly means to understand the world around us. Do we rely on the unique perspective of our fellow humans, or do we cede that role to the cold, calculating machines? The answer, as with so much in life, is a complex and nuanced one. But one thing is certain: as we move towards a future where AI plays an increasingly prominent role in our lives, we must not lose sight of the importance of human perspective and the value of individual experience.

So, who is the true author of this article? The human who provided the prompts, or the machine that transformed them into words?

The headline and body of this article have been algorithmically generated by chatGPT using the following prompts:

Headline Prompt: Make this headline sound more newsy: “Will you read this thoughtful article written by AI?”

Body Prompt #1 “Describe the paradox of a human reading an article written by AI, highlighting the consequences of acquiring information that has been machine generated. Write it using the literary style of J Scott Fitzgerald.”

Body Prompt #2: “In the second paragraph, juxtapose the process of human writing and machine writing”

Body Prompt #3: “Add a thought provoking sentence at the end questioning the authorship of this article.”

New York, N.Y. (March, 2022) — Applied XL, a real-time information company tracking disruptions to the health of people, places, and the planet, today announced it had raised $3.5 million in seed funding led by Hearst Ventures with the participation of the Boston Globe Media Partners and STAT, (which are owned by the Henry family), Alumni Ventures, DataPower Ventures as well as renewed support of all its pre-seed investors including Tuesday Capital, Correlation Ventures, Scott Cohen, Robin Lingg, Michael Reitermann, Sandiip Bhammer, Christina Topsøe and Geoffrey Hoguet.

The latest round brings Applied XL’s total funds raised to date to $5 million in the last 12 months. The company will use the new funding to scale its first vertical serving pharma and biotech professionals, expanding its offerings for the life science analytics market, whose value is projected to reach $42 billion by 2025.

“Professionals in this sector are overwhelmed with an ever-growing amount of unstructured data, missing out on crucial information, and slowing down their decision-making. We are developing event detection algorithms that replicate the rigorous research process of biotech journalists in order to surface important signals and deprioritize noise,” said Applied XL’s CEO and co-founder, Francesco Marconi. “We are deeply grateful to Hearst Ventures and all of our investors who support our vision of building smart systems for eliminating information overload.”

Applied XL uses machine learning to monitor dozens of industry databases and make sense of disparate data streams, surfacing signals before they become news. Rather than providing static snapshots and summaries of existing information, the system dynamically categorizes events, ranks them, and assigns proprietary labels that provide immediate insight into when and why important changes occur. The company has also developed a human-in-the-loop system in which an expert network contributes with data input, which is then used to retrain the system’s machine learning algorithms to achieve increased levels of accuracy.

“Hearst recognizes the transformative potential of B2B information solutions tailored for specific industries. Hearst has already experienced marked success with our investments in the healthcare, transportation, and financial services sectors,” said Ken Bronfin, senior managing director at Hearst Ventures. “Our investment in Applied XL reflects our broader strategic focus on leveraging the power of big data to power information products serving professionals across sectors.”

In January 2022, Applied XL partnered with STAT, the leading media company reporting on health, science, and medicine, to launch STAT Trials Pulse. Users include professionals spanning pharmaceutical, health care, biotech and financial services companies who use the data to inform regulatory strategy, clinical development and investment decisions. For example, the platform helps identify patterns in patient enrollment changes and timeline shifts to anticipate interruptions to clinical trials.

“Since the launch of STAT in 2015, Boston Globe Media Partners has already seen the impact of a specialized product targeting small but influential audiences and charging them for high-end news, information, and insights.” said Linda Pizzuti Henry, CEO Boston Globe Media Partners. “And now, we’re excited to build on this work by investing and partnering with Applied XL to continue developing its technology and provide industry leaders with the data-driven insights to inform critical business decisions that will allow them to grow and remain competitive.”

Applied XL was launched in 2020 through the Newlab Venture Studio by computational journalists who formerly led artificial intelligence efforts at the Wall Street Journal and Associated Press.

About Applied XL

Applied XL is a real-time information company tracking disruptions to the health of people, places, and the planet, so decision makers can act faster and plan ahead. Applied XL develops event detection systems powered by machine learning and editorial algorithms to find early signals before they become news. Learn more at appliedxl.com.

About Hearst Ventures

Hearst Ventures is the global venture capital division of Hearst. Founded in 1995, the group has grown to become one of the most active and successful corporate venture funds with more than $1 billion invested to date.

Hearst is a leading global, diversified information, services and media company with more than 360 businesses. Its major interests include global financial services leader Fitch Group; Hearst Health, a group of medical information and services businesses; Hearst Transportation, which includes CAMP Systems International, a major provider of software-as-a-service solutions for managing maintenance of jets and helicopters; ownership in cable television networks such as A&E, HISTORY, Lifetime and ESPN; 33 television stations; 24 daily and 52 weekly newspapers; digital services businesses; and nearly 250 magazines around the world. Follow us on Twitter @Hearst. To learn more about Hearst, visit hearst.com.

About Boston Globe Media Partners, LLC

Boston Globe Media Partners, LLC (“BGMP”) provides news and information, entertainment, opinion and analysis through its multimedia properties. BGMP includes The Boston Globe, Globe.com, Boston.com, STAT, and Globe Direct.

About STAT:

Founded in 2015, STAT is a national digital media brand that focuses on delivering fast, deep, and tough-minded journalism about the life sciences industries to over six million monthly site visitors and an additional 20 million readers on the Apple News app. STAT takes you inside academic labs, biotech boardrooms, and political backrooms, casting a critical eye on scientific discoveries, scrutinizing corporate strategies, and chronicling the roiling battles for talent, money, and market share. With an award-winning newsroom, STAT provides indispensable insights and exclusive stories on the technologies, personalities, power brokers, and political forces driving massive changes in the life sciences industry — and a revolution in human health.

50

STAT Trials Pulse provides biotech and pharma professionals with the earliest signals about clinical development through machine learning and editorial algorithms.

NEW YORK, NY — Today, STAT, the nation’s must-read health, science, and medicine publication, and AI company Applied XL, announced the launch of STAT Trials Pulse, an artificial intelligence platform that uses machine learning and editorial-driven algorithms to help life sciences professionals identify and prioritize the most important updates happening in the clinical trials space. The platform was developed by Applied XL, a company that builds real-time information systems rooted in computational journalism principles, in collaboration with STAT’s biotech reporting team.

“Our mission at STAT has always been to provide readers with leading coverage on health, science, and medicine. Now, STAT is expanding beyond reporting. With STAT Trials Pulse, our organization has taken a huge step in providing industry professionals with the insights and knowledge they need to succeed. Today’s announcement is a game changer for how we use technology to acquire information,” said Rick Berke, Co-Founder & Executive Editor of STAT.

STAT Trials Pulse includes the following features:

• Personalized user experience: the platform creates tailored feeds that monitor specific conditions, interventions, and organizations that matter most to you, categorized and ranked by proprietary data points.
• Timelines and smart updates: STAT Trials Pulse uses these data points to build dynamic timelines for each clinical trial, notifying users of trial stops, delays, and completions. Smart updates alert users about trial changes as they happen and provide explanatory context.
• Trends: the platform indicates trends based on clinical trial event outliers, allowing users to surface emerging new research activities, potential disruptions to clinical pipelines and regulatory roadblocks.

“We leveraged our proprietary event and trend detection systems to build a product that translates the best practices of computational journalism into a next generation business intelligence platform. There is a growing demand among pharma operators and biotech investors for tools that can help cut through information overload and enable faster decision-making. STAT Trials Pulse fulfills these needs, giving users an unbeatable competitive advantage by surfacing signals before they become news,” said Francesco Marconi, Co-Founder & CEO of Applied XL.

Additionally, a team of STAT journalists will be curating diverse and timely feeds relevant to researchers, scientists, and journalists. The team includes biotech senior writer Adam Feuerstein, medicine senior writer Matthew Herper, and general assignment reporter Kate Sheridan. STAT Trials Pulse is available for a 4-week free trial to all new users that sign up. The platform’s annual cost is $1500.

This interview guide offers a framework for understanding how to audit artificial intelligence tools implemented across the public and private sector.

As these systems proliferate all parts of life, it is important to understand how to question specific algorithms, just as we question human decisions.

In a world where algorithms become ubiquitous in government and business, it’s critical to develop a more nuanced understanding of AI. By understanding the intricacies of smart machines and knowing what questions to ask, we can help keep these systems accountable. Here are important questions to ask:

1. How is the organization using artificial intelligence?
2. Are any decision trees or rule based systems being used and framed as AI?
3. What is the technology used by the algorithm? (eg. machine learning, natural language, speech recognition, computer vision, robotics et.)
4. What is the purpose of the algorithm? (eg. prediction, classification, optimization, pattern recognition, event detection etc. )
5. What are all forms of human intervention and how do they impact the decision-making process? (eg. collect data, label data/ annotation, model development and training, validation, test, productization)
6. What are the procedures in place for human oversight and/or override of algorithmic decisions?
7. What libraries have been used to build the system?
8. How much of the code is open-source?
9. Has the author of the code been given due credit?
10. What types of auditing procedures are in place?
11. Is there an internal team responsible for auditing the algorithm (eg. specialized data science team, automation editors, subject matter experts etc.)
12. How are algorithm deliverables communicated to technical and non technical staff of the contracting company?
13. What criteria is used to measure the AI system’s effectiveness?
14. Which of the many algorithm variables are the determining factors in a model’s performance?
15. What decision-making concerns were considered during the data collection and model generation phases?
16. What were the initial errors during data collection and algorithm testing phase?
17. What changes have been made to the algorithmic development workflow to address any of such errors? How and why?
18. Has recent academic research and development informed the improvement of the algorithmic workflow?
19. How does the algorithm address any potential concerns related to discrimination or bias? (eg. racial bias, gender bias, economic bias etc.)?
20. How can the algorithm be exploited and what correction measures have been implemented?
21. During model training, has enough data been provided to the algorithm/model to account for unusual cases?
22. How were the criteria and process for data collection determined?
23. Was public data used to train the algorithm?
24. Was private data used to train the system? What approvals were required to access the data?
25. Was the data and copyright owners informed of the potential use of the data? (eg. terms of use, email communications etc.)
26. What other ethical and legal concerns were addressed before and during data collection?
27. Was the data collected before or after the model was built?
28. How much of the data was collected after the model was built? And how did it perform on that data? (Models trained on retroactive data perform well on trained sets. The best way to assess a model is to collect fresh data and test it.)
29. Has the company paid for licenses to access the training data from a third party? If so, has that partner been vetted?
30. What happens to data that has been collected for a particular algorithm and used for training?
31. How is the quality of collected data assessed?
32. Have datasets containing private identifiable information been anonymized before use?
33. How did the organization in question manipulate or sanitize the data for the algorithm?
34. How did it decide what data is relevant to the model?
35. Are end users informed of what inferences are made using the data? (eg. algorithmic disclaimers)
36. If the company discloses this information to users, is the communication clear, made public and easy to understand?
37. Can the collected data be used for predatory practices?
38. Can the data be used to influence or limit an individual’s choice in the above markets or the public realm?
39. Can this data be used by a government or private surveillance body?
40. Is the data being collected by a private organization to exploit user engagement?
41. If so, what mechanisms or triggers are being used to retain user attention? (eg. fear, conspiracy theories, provocative, violent or user-sensitive content)
42. Are there any laws in place against these practices?
43. If not, are there any data privacy groups working on addressing these concerns? And has the organization engaged with them?
44. What are the ongoing bills addressing these concerns?
45. Can the data collected by companies expose sensitive details about individuals? (eg.exact residence location, gender identity, sexual orientation etc)
46. What measures has the company taken to address potential security or privacy leaks?
47. Has the company been subject to an independent expert or platform auditing its data?
48. Are there any federal agencies overseeing the practices of data collection in the company’s field of operation?
49. Are owners of the personal data given rights to access their personal information or withdraw access privileges to the company or the data collector?
50. Has there been a manual review of the partial or complete dataset?

1. In 1889, Jules Vernes foresaw live news: “Earth Chronicle is every morning spoken to subscribers, who, from interesting conversations with reporters, statesmen and scientists, learn the news of the day.” In 1980, CNN launched the first TV channel providing 24-hour news coverage.

2. Hugo Gernsback described news personalization back in 1911! “Morning newspapers transmitted to the sleeping subscriber by wire at 5am. The newspaper office, notified by each subscriber subscriber what kind of news is desirable, furnished only such news”.

3. In 1939, Harl Vincent wrote a book called Newscast featuring the “Polyceltron Iconoscope Televisor”, a portable camera and microphone setup that could broadcast on-the-spot news. Today, news organizations use LiveU transmission technology to broadcast directly from the field.

4. News summarization was introduced in 1942 by Robert Heinlein: the “News Roundup” device presents brief story outlines, any of which can be chosen to find details. Today, publishers use Natural Language Processing to automatically generate news summaries.

5. In 1961, H. Beam Piper wrote about a device called the “Transmit Camera”, a system that both takes pictures and uploads it directly to a news station. This sounds a lot like crowdsourcing journalism platforms like Stringr.

6. In 1963, Philip K. Dick imagined the “Homeostatic Newspaper”, an autonomous news-gathering and publishing entity, that produces a newspaper without human assistance. “News Receptors” were the devices used by homeostatic newspapers to gather news automatically. My favorite!

7. In 1967, Philip Jose Farmer described the combination of a television, a news camera and surveillance device — a gadget named Fido — which to me, sounds a lot like the Citizen app.

8. Google News Alerts were conceptualized in 1978 by Arthur C. Clarke who wrote about “Personal Interest Profiles”, a set of topics about which people wanted to know about, cutting through the flood of global information.

9. The idea of synthetic media was introduced in 1988 by Jeff Kleiner + Diana Walczak who created the 1st digital actor for their film “Sexton for President”. They also coined the term “Synthespian”, a synthetic media personality. Xinhua News launched its own synthetic anchor in 2018.

10. In the 2002 movie Minority Report (the original book is from 1956), Steven Spielberg shows a paper-thin, flexible LCD-style display that was large enough to display a full newspaper page. Not sure if USA Today is currently working on an ePaper but they do exist, for example Remarkable Paper.

Humans are producing more information right now than we did at any point in history. We’re up to our eyeballs in data — data from the web, from sensors, from mobile devices and satellites, all growing exponentially. The amount of data created over the next three years will be more than the data created over the past three decades. In the next five years, these technologies will produce more than three times the information that they did in the previous five.

What does this flood of data mean for the people whose job it is to make information accessible? How can we leverage emerging practices of journalism to derive business intelligence?

The ever-rising sea of data presents both a dramatic challenge and, up to this point, largely untapped opportunity for the information industry. The field’s long-standing, mostly rigid methods do not have the capacity to interpret or contextualize the enormous amount of data our society is poised to produce. We need faster and more sophisticated information gathering if we don’t want to drown in big data.

Determining what is deemed “newsworthy” will evolve into a process focused on finding and analyzing statistical outliers with clarity, while articulating meaning by providing the human context behind the data. News becomes information when it demonstrates clear utility for decision making.

How should information gathering adapt?

The concept of “drowning in data” sounds abstract, but there are real stakes involved in ensuring that the methods of journalism adapt to the reality of today’s information landscape. In the next decade, society will face extraordinary challenges — from the increasing probability of global pandemics to crumbling infrastructure and the effects of climate change. The disruptions to come call for an urgent reassessment of how information is sourced, because good information plays a vital role in creating a shared reality — a common frame of reference among the public, which in turn enables individuals, governments and organizations to work towards solutions from a place of shared “ground truth.”

To meet the demands of the next decade, we must move towards a new type of journalism that is more scientific and analytical, a journalism that contextualizes raw data by continuously tracking multiple sources and by processing and vetting it in real-time. This is computational journalism.

The sector that currently gets closer to this notion is financial information; stock prices and economic indices guide coverage and provide organizations and the world with a “ground truth” about the markets. But this has not always been the case. A number of companies, such as Dow Jones and Bloomberg, were originally built with the goal of collecting financial data, normalizing it, and providing it to individuals and organizations. Since their inception, these institutions evolved into fundamental cornerstones of the global economic infrastructure.

But what if we could do something similar for sectors to help drive growth beyond just capital — for example, to spur progress across the life sciences, energy, infrastructure, and supply chain industries? What if we could define a “ground truth” in real time that illustrates how our healthcare system operates, or provides an assessment of an organization’s environmental footprint, or conveys the state of local infrastructure? What if we could track and contextualize significant “state changes” about the health of people, places, and the planet with the same rigor and analytical detail as we do the financial markets?

As companies and governments increasingly move their operations online, they leave data trails that, if properly mined, can provide unique insights into how these entities work. Massive troves of data across sectors hold vast potential for shaping how organizations function — *all* that is needed is a systematic approach to consolidating and contextualizing that data to provide actionable insights.

It is a short step from where we are today to a future defined by new systems that allow us to detect and contextualize data from a diverse array of sources at the continuous rate with which we track stock prices. However, challenges remain: for example, such systems require scale that is not currently supported by traditional newsrooms, due to the need for monitoring multiple data sources simultaneously and at a pace surpassing manual capabilities.

At the same time, these systems need transparency that tech companies have failed to offer. The same way journalists ask questions to human sources, they should be able to ask questions of their algorithms, to try to understand their inner workings. This will allow for the creation of reliable algorithms, and help avoid the pitfalls of purely autonomous systems, such as machine bias. By applying journalistic principles in the development of technology we can mitigate unwanted AI biases and ensure fairness.

Editorial algorithms like the ones that we are developing at Applied XL are needed to make sense of enormous quantities of data while maintaining high journalistic standards.

At Applied XL, we are creating information systems that will work 24/7 to monitor billions of data points. Our intelligent systems are guided by principles defined by the computational journalists who program them and are vetted by experts in specific domains, combining both the high standards of a newsroom with the scale of Silicon Valley platforms.

Our first swarm of editorial algorithms is currently monitoring clinical trials data to identify newsworthy events and state changes, in many instances before those insights appear in press releases or news coverage.

Given the expanding accessibility of artificial intelligence and the proliferation of cloud computing, we are now confident that we can scale the collection raw data automatically and reliably at a speed previously unimaginable, making it possible to keep up with the explosion of data. This does not mean that AI will take over for human journalists, but it will undoubtedly transform their roles: the journalists of the near future will be information officers and algorithmic editors.

How Humans Will Remain in the Loop

Naturally, the prospect of such a paradigm shift is bound to set off alarm bells in some quarters of the media. If everything is reduced to numbers, won’t we lose sight of the human context? To answer this question we need to first understand the origin and evolution of journalism. In its early days, journalism was designed to provide context in a period of information scarcity — wherein a relatively limited amount of readily available information could be manually sourced.

Information scarcity made it feasible for reporters to rely on human connections for intel, which they acquired through interviews, requesting documents from their contacts, and attending press conferences. Eventually news organizations had to become more standardized to keep pace with the growing complexity of the information landscape, developing new practices such as “beats,” the inverted pyramid, and planning-based news cycles. These techniques and processes have been adapted time and again with the advent of new technologies, including the internet. But we have reached an inflection point where human interaction alone, even if optimized, is no longer enough to provide journalists — and their consumers — with the knowledge they need. Processes that were once useful are quickly becoming obsolete for an environment of information abundance, which we often refer as “data vomit”.

It’s true that numbers, on their own, are insufficient to tell a full story. It’s for this reason that journalists and other knowledge experts play an integral part as the arbiters of algorithmic truth.

In fact, journalism has always been about data. It’s just that today, the times have outstripped the industry’s methods. And from a purely data science perspective, journalism is the detection and documentation of statistically significant events. With editorial algorithms, journalists as well as experts are still the ones who decide what weights, parameters and transparency principles to apply to their machine learning models.

If this is the inevitable future of information, why are established companies not investing in creating these systems?

Most information companies, especially legacy ones, have consolidated their processes over the decades in such a rigid way that it has become difficult to introduce new technologies that would update existing workflows. They may be able to improve components of their systems, but very unlikely their entire workflow. The reality is that it’s too costly for them to re-engineer their systems and retrain their staff.

The specialized B2B information industry has seen immense growth in the last several years as companies have sought to gain an edge. But the professionals these companies serve report being overwhelmed by the sheer volume of information, the lack of reliable real-time context for this data, and the complexity of navigating it. For the information industry to thrive in the next decade, we need entirely new approaches able to serve the needs of a new generation of consumers and decision makers where speed and accuracy is essential.

In early 2020, I left my job as R&D Chief at the Wall Street Journal to do precisely that. I was joined by Erin Riglin who was previously WSJ’s automation editor and now serves as our Chief Technology Officer. For the last several months we have been incubating the company at Newlab, a community of experts and innovators applying transformative technology to the world’s biggest challenges — and the innovation epicenter of New York City — with the support of Applied XL’s chairman Scott Cohen.

Today we are excited to announce that we raised $1.5 million, a seed investment led by Tuesday Capital to build an information company designed for the era of big data, a company that fuses journalistic integrity and computational speed. Other outstanding investors that believe in our vision include Frog Ventures, Correlation Ventures, Team Europe, the investing arm of entrepreneur Lukasz Gadowski (co-founder of Delivery Hero), Ringier executive Robin Lingg, a global media and marketplaces specialist as well as other experts in the information and life sciences industry.

With this seed investment, Applied XL is building its event detection systems to source data using custom algorithms rooted in editorial principles, which are constantly calibrated and tuned by experts in the loop. Our growing catalog of algorithms draws on a set of very specific newsworthiness criteria that, when triggered, dynamically generate news alerts with context and precision.

Our first vertical will focus on life sciences, an industry with specific information needs currently not being fully met by existing private and public data sources; for example, many existing data providers cannot deliver the kind of real-time context life science organizations require to make decisions on clinical trial development, competitive positioning and commercialization. Applied XL will address this need by mining data on clinical trials, pharmaceutical industry regulation, and healthcare policy, and automate the generation of alerts and reports. We will initially work with select partners, including professionals working in strategy functions at biotech and pharma companies, and will open our data platform to life science experts, including scientists and journalists, enabling them to access data and calibrate machine-driven insights. Following the launch of our life science product, Applied XL will tailor its technology to meet the evolving needs of a range of other industries.

Applied XL aims to use editorial algorithms to track the health of people, places, and the planet. Our mission? To democratize information.

If you would like to get in touch, you can contact us at pilot@appliedxl.com

1. From a machine learning perspective, journalism is *simply* event detection applied to continuously updating data.
2. Let’s look at an hypothetical example where there’s a lot of reliable data: Sports. If we were to monitor all the professional soccer/football matches around the world, occurring over a given weekend, how do we determine what the most newsworthy games were?
3. That decision is traditionally made by a journalist based on their domain knowledge and context of the subject — they may emphasize matches that had the biggest offset, focus on top ranked teams, or even pay more attention to teams where highly paid athletes play, among many other considerations.
4. In this sense, newsworthiness is triggered when predefined editorial criteria is met. Now, assume that the logic of those principles can be written down and translated into algorithms. Who determines what is newsworthy and what’s not?
5. It is certainly not the machine. It is instead human editors who decide what weights and parameters to apply to the machine learning models. These are called “editorial algorithms”.
6. Beyond Sports, only a few other coverage areas such as Finance or Weather are fields where data is abundant and for the most part, objective. The same cannot be said for beats such as politics, health, science, world events, among others — yet.
7. As everything around us is moved to the digital realm, new data trails and records are being created that can provide ways to constantly monitor fields that today are not quantifiable. Natural language processing is crucial here, but I will leave that for another thread.
8. Within the next decade, we will be able to source and gather news through machines at a speed that today is unimaginable (think Associated Press meets Palantir). This does not mean that AI takes over human journalists, quite the opposite.
9. The development of editorial algorithms will require humans at the helm (and to clarify, a journalist does not need to be technical in order to define reliable algorithmic parameters — but at the very least they have to understand how machine learning works).
10. In this new reality, algorithmic transparency will become one of the most crucial functions of modern newsrooms.

1. The pandemic forced many news organizations to innovate how they use technology, rethink how they manage remote staff and make sense of huge amounts of complex information. Covid also demonstrated the importance of tracking data in real-time and making it a central reference for all reporting!

2. The sad reality is that the frequency of how often pandemics occur will only accelerate with climate change, which means newsrooms will need to be at least 10x more efficient in how they collect data — or risk their very own survival.

3. As temperatures rise and the global population grows to 8.5 billion by 2030, humans will live in closer proximity with wild animals that put us at risk of covid-like diseases.

4. Floods will lead to an increase in the number of people getting sick from infectious diseases spread by mosquitoes including Malaria and Dengue — even in regions that are not currently affected by these maladies.

5. Diarrheal disease will be rampant in areas where water has been contaminated as a result of extreme precipitation patterns, extreme rainfall events, and seasonal variations.

6. Finally, as ice melts we will come in contact with soil that has been frozen for 1000s of years, and it is in the realm of possibility that ancient viruses and bacteria could emerge.

7. What does all of this have to do with journalism? Public officials, organizations and individuals all rely on information to make decisions. The faster we can source information and contextualize it, the more time we have to prepare, evaluate and mitigate risks.

8. For newsrooms and information companies of all types to survive in the next decade, they will need to reinvent news gathering and to develop real-time baselines for tracking the health of people, places and the planet. Right now a lot of this data is still collected manually and even managed in spreadsheets…

9. Solutions include: leveraging AI to make sense of large amounts of scientific publications and deployment of sensors to directly collect data from the field. Incorporating satellite observations is an also a new concept for epidemiologists that can be applied to journalism.

10. Deep collaboration between journalists and experts will be crucial for the future of news. Undoubtedly, journalism must evolve into becoming “information science”!

Author’s note: for more on how AI is changing and journalism, check out “Newsmakers: Artificial Intelligence and the Future of Journalism”.

How the Wall Street Journal is using Deep Learning to Inform Content Strategy

2.The explosion of data from the web, sensors, mobile devices and satellites, combined with the accessibility to machine learning, creates the ideal environment for transforming how all information around us is sourced.

3. Advances in AI research are already allowing computational journalists to measure issues that were previously impossible to quantify (like counting trees in a desert), and as a result surfacing entirely new analytical baselines for the world around us.

AI has just revealed there are over 1.8 billion trees in the Sahara Desert

4. Currently, financial news is one of the areas of coverage that gets closer to the notion of *ground truth*: stock prices and economic indices provide journalists with shared reality for the markets.

5. What if we could define real-time baselines for how our healthcare system is operating? or for knowing the resilience level of our communities? What if we could report the news with the same analytical detail we follow covid infection rates?

6. I believe the traditional news process needs to become more scientific. The next frontier in news innovation will be focused on developing *self-gathering information systems*: continuously tracking data, processing it for insight and vetted by humans in the loop.

7. What we call “news” today, in the next decades will evolve into a process exponentially more focused on finding and explaining statistical outliers: the mathematically unpredictable and the scientifically unlikely.

8. This doesn’t mean that the future of journalism is about letting the machines run loose. Bien au contraire.

9. Optimal computational journalistic performance happens somewhere between finding these new data signals and having humans validating and contextualizing them.

10. In the future, journalists will become information officers and arbiters of algorithmic truth.

Author’s note: for more on how AI is changing and journalism, check out “Newsmakers: Artificial Intelligence and the Future of Journalism”.

The growing demand for information during the pandemic has not been met with an adequate increase in journalistic resources. In fact, the opposite has been true. According to the Poynter Institute, in the US alone over 33,000 journalists have been laid off, furloughed, or given pay cuts as a direct consequence of the coronavirus, resulting in a reduced capacity to cover fundamental public health issues. Despite the dire situation, the crisis has shown a secondary, positive side effect: accelerating the speed of journalistic innovation.

Take the case of British local wire service Radar AI. The joint venture between Press Association and Urbs Media leverages machine learning to dynamically generate regionalized COVID-19 news digests, distributed to local news media in the UK.

“In May alone we produced 17,000 local news stories. We realized that we had an important role to play in helping our subscribers tell that story to readers at a local level. Usage of AI-driven content by our subscribers have been at record levels throughout, climbing each month since April,” said Gary Rogers co-founder and Editor in Chief of Radar.

Covid’s impact on ad revenues

But this accelerated tech adoption on the newsgathering side is not enough to alleviate the immense pressure local news industry faces, exacerbated by a steady decline in advertising revenues. According to FTI Consulting, newspaper ad revenues in Q2 of 2020 have declined by 20%-30% compared to the same period last year.

“The challenge is that there has been no market response to the growth of digital information. In particular local advertising has collapsed, I would say disappeared, as a result of the financial crisis in the sector. A long and serious economic crisis awaits us,” said Stefano De Alessandri CEO and Managing Director of the Italian news agency ANSA that recently launched an initiative in partnership with Applied XL to support local newspapers with Covid-19 related coverage powered by AI.

A weakened local news ecosystem leads to a public that is less informed, which is problematic, especially during a pandemic, where analysis of public health data is essential for many communities to stay up-to-date on the virus’ impact on their towns and local health care systems.

The ultimate data story

Elsewhere in the industry, other news leaders recognize that coronavirus represents an inflection point for journalism. In fact, this pandemic is the “ultimate data story,” as Bloomberg senior executive editor Chris Collins explained in a recent conversation with the London School of Economics JournalismAI research initiative. Indeed, Bloomberg is one of many organizations using AI alerting systems to monitor social media and other data sources for coronavirus related insights to notify reporters as soon as a trend or anomaly emerges from data.

British Philosopher Alain de Botton’s thought-provoking description of journalism as “the determined pursuit of the anomalous” is taking on a literal meaning, as news organizations race to adopt algorithmic practices. Statistical outliers in the data help anticipate newsworthy events, leading journalists to their next story: Where is the next Covid-19 outbreak going to be? Will the healthcare infrastructure be prepared to deal with an influx in patients?

News organizations have been looking for signals in data for a long time by analyzing movements of stock indexes and deducing their potential impact on the economy. What has changed is that journalists now have a multitude of data indexes — measures of ground truth — at their disposal, and powerful AI tools to derive insights. During the Covid-19 pandemic, reporters have been using economic data and health metrics across countries, regions, and cities to find the next outliers of this ongoing story.

“This crisis changed our business. Key areas where we were focusing on, such as sports coverage, became obsolete within days of the pandemic disrupting normal life. We had to reinvent our business and focus on creating value for our customers, namely providing Covid-19 content,” said David Llorente, CEO of AI-powered news agency Narrativa which is headquarted in Spain.

Alternative ways for contextualizing reporting

Leading health news publication STAT News which has grounded some of its reporting on data tools including its Covid-19 tracker and the County Health Preparedness score, which estimates the impact of the pandemic on the healthcare system by leveraging public data and statistical models. This new measurement derived through data science developed by Applied XL and data from the Center on Rural Innovation is not only providing context to STAT’s reporting; it is also being leveraged by organizations such as US News & World Report and FOX News to quantify the impact on health infrastructure in rural areas.

“The beauty of this use of data is that it speaks for itself. It’s essentially unfiltered so readers can explore and interpret and analyze the information themselves in real-time. As much as we write critical news and feature stories on Covid-19, our tracker has been vital in helping readers grasp the enormity of a public health crisis whose scope and scale sometimes seems beyond comprehension,” said Rick Berke, co-founder and executive editor at STAT News.

Newsrooms are quickly identifying new ways of contextualizing information beyond traditional metrics. For example, digital publication Axios is using data on the share of beer taps pouring at open locations to measure consumers’ confidence about reopening cities amid the pandemic.

This form of data-led reporting through alternative indexes is here to stay after the pandemic ends, with ever-growing data sources that extend to other parts of news organizations’ coverage: transportation, housing, local economy, law enforcement, education, and more.

Bridging the innovation gap in small newsrooms

Local news organizations can take advantage of open-source data analysis tools and free online courses for journalists to leverage machine learning libraries for reporting. Collaboration with other newsrooms and partnerships with academic institutions to share knowledge and resources can also accelerate the pace of adoption. London School of Economics’ media think-tank Polis, for example, recently announced a partnership of more than 20 international news organizations to prototype AI-powered journalistic solutions and discuss the impact of AI on journalism during the Covid-19 pandemic.

“Given the importance of data in telling the story of COVID19, both nationally, and locally, there are some interesting ways to leverage automated content production to bolster local news about the crisis. A nice example of this is the Big Local News Case Mapper, which offers an embeddable map and chart that local outlets can use. There are perhaps untapped opportunities for AI to be applied here as well, such as to detect and alert reporters when particular locations exhibit spikes in cases or have a worrisome trend,”said Nick Diakopoulos, computational journalism professor at Northwestern University.

By leveraging data science and artificial intelligence, news organizations will be able to provide in a more effective manner with real-time data of events and information catered to a reader’s specific city or neighborhood. In an industry beset by declining revenue, local news on public health can indeed become a crucial entry point for driving audience engagement and future revenue.

Author’s note: for more on how AI is changing and journalism, check out “Newsmakers: Artificial Intelligence and the Future of Journalism”.