If you’d like to use technology to make your newsroom more efficient, you’ve come to the right place. Tools exist that can help you find news, manage your work in progress, and distribute your content more effectively than ever before, and we’re here to help you find the ones that are right for you.
As part of the Knight Foundation’s AI for Local News program, we worked with the Associated Press to interview dozens of news organizations in order to understand the most pressing issues news organizations face; we want to use our newfound knowledge to help you find the products and processes that will get you moving faster as quickly as we can.
Different tools match different organizations, and our recommendations depend on your previous experience with automation, computer science, and artificial intelligence. You don’t have to be an expert programmer or hire a team of developers to build and use these tools—there are plenty of great off-the-shelf solutions available, and we’re eager to help you find them.
This essay starts with some general information about newsroom automation that aims to be useful regardless of your level of previous experience. It then provides a series of questions designed to help you reflect on your existing knowledge. We use your responses to point you towards tools tailored for your particular level of expertise. After the review of selected tools, we conclude with some thoughts about reliability and ethical concerns, and some suggestions for further reading.
How do these tools work?
Several tools we recommend depend on artificial intelligence. If that term seems intimidating, fear not: most researchers who work in that field define it as “the science of making computers do things that require intelligence when done by humans.” You probably use technologies that rely on AI every day:
- Siri, Alexa, and other smart assistants rely on algorithms from natural language processing to understand your questions.
- YouTube, Facebook, and other similar sites use recommendation engines to find the most interesting content for you to enjoy.
- The New York Times, Reddit, and other platforms use machine learning to automatically detect and moderate toxic comments.
Practically speaking, journalists using AI systems do not have to understand these concepts completely. One can buy tools like transcription services and story generators off-the-shelf without worrying about the underlying details.
Note, too, that much of the software a newsroom can benefit from relies on nothing more complex than ordinary software engineering. Systems designed to help you archive your stories or publish new pieces onto a website, for example, need not rely on artificial intelligence at all — the technology powering them is no different than the technology used to power most other websites. Even if you’re hesitant about AI, adopting some of these tools gradually can help boost your productivity tremendously.
What tools are right for us?
If you already feel confident about your level of background knowledge, feel free to skip to the part of this essay that points to your level of expertise. However, if you are unsure where you belong, please read the statements below and count the number applicable to you.
- Reporters at my organization know how to effectively search the web to find newsworthy material.
- Online databases and specialized search engines help reporters at my organization find content for stories.
- My organization takes advantage of systems like Google Alerts to help us find relevant news automatically.
- To help reporters manage their stories, my organization built or hired others to build custom systems and infrastructure.
- My organization uses natural language generation to produce stories that our reporters can then edit.
- Automatic triggers run when my organization releases a story; for example, publishing an article automatically creates posts on platforms like Facebook.
- Data analytics helps my organization understand how our audience responds to our content.
If fewer than 3 of these statements apply to your organization, we suggest you read the section for organizations less experienced with automation. If between 3 and 5 of these statements apply, consider reading the section about adding additional tools; if more than 5 apply, the section about advanced tools is for you.
Once you are done reviewing the tools, scroll down to view potential concerns of AI and further reading on the subject.
Note that we have selected the tools below as examples to help you begin exploring. Their inclusion in this document is not an endorsement, and these tools were not evaluated by direct comparison against other tools which may provide similar services. Please feel free to examine other options that better fit your organization.
Newsroom AI For Beginners
Automation, artificial intelligence, and their applications to journalism are broad and ever-expanding fields. This is great news for the potential possibilities of work-load reduction, but the expansiveness and overwhelmingness of AI can be intimidating for getting started. We have found that newsrooms operating at your level of knowledge value user-friendly interfaces.
The tools in this section are not designed to complicate your newsroom by adding extra steps or replace human employees, but rather assist with many of the reasons newsrooms are unable to provide coverage that they wish. This includes speeding up processes and providing alerts to new information. The following list of tools is by no means complete, but is intended to provide a starting place to either begin to implement automation into your newsroom or further the progress you’ve already made, as well as augment your existing newsfinding, production and distribution processes.
Based on newsrooms we’ve interviewed, we think the following tools could be helpful to get you started with integrating AI into your journalistic processes.
Otter.ai
What it does: Otter is a transcription service that works with uploads or in real-time. Video and audio files can be uploaded and Otter will produce a transcription with timestamps and speaker identification.
How to use it: Simply upload your file, or press the record button, and Otter will begin transcribing what it hears. The transcript can be edited manually once the file is processed if Otter misidentifies any words, which can happen often.
Price: Basic Plan: Free, Pro Plan: $8.33/month, Business Plan: $20/user/month
Descript
What it does: Descript is a tool for audio and video journalism: the service transcribes recordings and then allows you to edit them within the site. Used by major outlets like the BBC, NPR, and CNN, Descript can be a helpful tool to cut down on time spent transferring files between applications. Plus, you can collaboratively edit podcasts with your colleagues. No audio editing experience required.
How to use it: Go to the Descript site, and upload or record your content. From there, Descript matches the transcripts with the audio files, and sections can be dragged and dropped to create a story.
Price: Free, Creator: $12/editor/month, Pro: $24/editor/month
Google Alerts
What it does: Google Alerts sends you updates whenever webpages, articles, research, and more get changed. Alerts can be sent via email or directly to an RSS feed (Really Simple Syndication: a file that carries website updates).
How to use it: Log in with your Google account and type in what you want to receive alerts about.
Price: Free
Meltwater
What it does: Another topic alert service, Meltwater, has more abilities than Google Alerts. With Meltwater you can track when topics are trending, when high profile social media users mention your topic, when major news outlets mention your topic, and more. Meltwater can also be integrated with Microsoft Teams and Slack.
How to use it: After obtaining access to the service, click the bell icon or navigate to “alerts.” From there, choose the type of alert, how you’d like to receive it, and what information you are looking for.
Price: Variable, use this form for a free consultation
Social Pilot
What it does: Similar to popular social media management and marketing app, Hootsuite, (which many expressed frustration with in our interviews), Social Pilot allows you to schedule content across social media platforms, check analytics, and engage with followers.
How to use it: Create an account or start a free trial, then connect your social media accounts from the SocialPilot panel. From there, you can begin to add your posts to a queue.
Price: Professional: $25.50/month, Small Team: $42.50/month, Studio: $85/month, Agency: $127.50/month
Newsroom AI for Intermediates
For those newsrooms with an “intermediate” understanding of AI in journalism, some of the tools suggested in the previous section may already be familiar. Expanding on this list, here are more off-the-shelf tools that illustrate the myriad ways to automate tasks in the newsroom without needing a technical background.
Narrativa
What it does: Narrativa’s Natural Language Generation service, called Gabriele, generates narratives using data from the biggest western markets. Gabriele fulfills one part of the news production process; using the text it generates, writers and editors add their own edits, analysis, and information.
How to use it: Gabriele and Narrativa’s other NLG services automatically generate text content on a variety of topics.
Price: Narrativa does not publish the cost of its Gabriele and other products, but to learn more, you can book a demo.
Flourish
What it does: Tool that can generate various animated or interactive visualizations (such as bar chart races, story maps, and chord diagrams) given datasets.
How to use it: Flourish provides a library of flexible templates for users to explore their datasets and generate visualizations in line with their brand guidelines.
Price: Free and paid tiers (e.g. $69/month for personal use)
IFTTT
What it does: Like Hootsuite, Social Pilot, and other social media management tools, IFTTT streamlines the distribution of content over different social media channels through automation. In particular, IFTTT has specific functions for syncing content between these platforms, such as keeping Facebook and Twitter profiles in sync.
How to use it: Connect to Facebook, Twitter, and other apps.
Price: Free, or about $3/month for the pro version
Lighthouse
What it does: Google Lighthouse is an open-source, automated tool that runs a series of audits on a given webpage and generates a report on the page’s performance, accessibility, search engine optimization, and more.
How to use it: Access the tool through a web UI, Chrome DevTools, or other workflows.
Price: Free
Newsroom AI For Experts
Newsrooms that have an “expert” level of understanding of AI in journalism means that your newsroom is very experienced in understanding and working with largely automated tools and technologies which help ease the workload of the menial tasks and activities that could be easily done by a machine. Your newsroom is not intimidated by using AI in journalism because of this prior experience and are looking for more advanced tools that will further the work you are already doing. Some of the tools in this section may require some programming skills in order to use them or more experience with AI. We are expecting the newsrooms that identify with this section to either have a team or a person who is quite knowledgeable in programming that can understand these aspects, whether that be someone who builds custom systems or someone who is generally more knowledgeable about working with tools that require programming or higher-level skills.
Based on newsrooms we’ve interviewed, we think the following tools could be helpful to further your experience with integrating AI into your journalistic processes.
Perspective API
What it does: This tool uses machine learning models to identify abusive comments online. According to its website, these models “score a phrase based on the perceived impact the text may have in a conversation” based on certain attributes such as profanity, identity attack, threat, sexually explicit, etc.
How to use it: Use your Google account and create a project. Then, fill out a form on the site to gain access to the API.
Price: Free
new/s/leak
What it does: new/s/leak (NetWork of Searchable Leaks) can aid in the information collection process by allowing users to explore large amounts of textual data. More specifically, the tool uses both language technology and data visualization to illustrate the relationships between important entities (people, organizations, places) within datasets like those distributed by Wikileaks.
How to use it: Access publicly available demos of the tool here
Price: Free
NLTK
What it does: This Python library provides a suite of extremely useful functions and corpuses for analyzing natural language. If you need to build a custom tool that requires tokenizing sentences or finding the antonyms and synonyms of words, few other products make doing so through code so seamless.
How to use it: First, install it inside your Python project using the package manager of your choice. Then, download the datasets upon which NLTK relies by following the instructions from the official documentation. From there, import it like any other module in the standard library.
Price: Free, Apache License 2.0
Scikit-Learn
What it does: This Python library provides a number of robust off-the-shelf machine learning models. It also provides functions for transforming raw text into vectors that those machine learning models can understand.
How to use it: After downloading scikit-learn with a Python package manager, import the model of your choice into your Python project. From there, embed the text you would like into vectors using your favorite strategy and run the models to see their results.
Price: Free, 3-Clause BSD License.
Conclusion
While many newsrooms are excited about the potential opportunities provided by artificial intelligence and automation in the newsroom, they are also skeptical about the ethics and accuracy of their uses. The level and type of skepticism often varies depending on an individual’s knowledge of artificial intelligence. For some, anxiety stems from the mystery surrounding AI. Others have a very clear idea of what AI can do, and are therefore able to foresee how it can be abused.
A common concern among newsrooms is the reliability of tools. Many newsrooms would welcome artificial intelligence as long as the quality of their work is not compromised. Journalists need to be able to trust the tools they are using to give them accurate information and cannot waste valuable time and resources on a tool that is unreliable. The point of adding automation to their work is so they can spend their time doing valuable work instead of mindlessly sifting through information, which is only helpful as long as journalists do not have to go back and check that the information is correct. Newsrooms do not want AI tools to create even more work than necessary.
While trusting AI tools to do work without much supervision is essential in some functions, many newsrooms agree that it should not replace the personalized and nuanced work of a journalist. While AI can be used for data sifting, interpretation is still a human task, and the possibility that automation could completely replace the role of a journalist is concerning to many in the industry. There is no proxy for the human mind, and the idea of a ‘robot’ writing an article by itself is very off-putting to some journalists. Also, many newsrooms agree that there should still be some form of human revision to any output from AI.
While humans certainly have their share of bias, that same bias is often built into algorithms. Newsrooms want to make sure that tools are collecting data in an ethical and legal way that does not violate the privacy of the public. The laws regarding data use are not as robust as they often need to be, in some cases allowing for the abuse of information. Algorithms need to be able to understand the nuances of issues such as censoring source names for the sake of protection and racial bias in facial recognition. In terms of audience engagement, some newsrooms are worried that algorithms will make decisions solely based on getting more clicks instead of the quality of content, or highlighting stories based on their polarity and creating echo chambers.
Despite ethical concerns, many newsrooms are very eager to incorporate AI into their businesses. It would be especially useful for smaller newsrooms that are low on resources and time. These newsrooms are often the ones who need automation tools the most in order to increase efficiency. However, they are also least likely to incorporate new technologies that are not accessible and realistic in terms of the time and money needed to start using them. Small newsrooms can be the hardest ones to convince to do the training for a tool unless it directly fits into their mission and workflow. While many newsrooms are willing to carve out the time needed to learn about new technologies, small-scale integrations that are inexpensive, off-the-shelf, and require a small amount of staffers would be best for newsrooms of any size. Even larger newsrooms are open to expanding their data teams, but can only do so with the right tools in mind.
Further reading
This tool is one of many resources that can help you think about incorporating AI/automated tools in your newsroom. For a broader look at automation in the journalism industry, Automating the News by Nick Diakopoulos provides an overview of AI and automated tools at various points of the journalistic process. Another resource that can be helpful is Meredith Broussard’s Artificial Unintelligence, about ethical issues, or Francesco Marconi’s Newsmaker: Artificial Intelligence and the Future of Journalism which discusses the future impact of AI on journalism. “The present and potential of AI in journalism” from the Knight Foundation provides an industry survey of how AI and automation are currently being used, and the related Journalism AI Case Study database, created by the JournalismAI project, includes a myriad of other tools that show how AI and automation can be applied toward journalism.
