News Industry: What to Do on AI Right Now
“So what should we be doing right now?”
This is the most common question I have been getting from media companies about AI and news, in the strategic consulting that I do across our industry and also in response to the article I wrote about early AI products created by teams I led at The Wall Street Journal.
This post will describe the foundational and cultural changes your teams should make to be able to play in the Data Era that we are in.
My views on AI and news are shaped by my years of digital transformation work at The Wall Street Journal and The New York Times, twenty years as a journalist, and also leadership positions both on the news editing masthead and as a chief technology officer working across business and content. Throughout my career, I have focused on audience needs. This turns out to be a powerful lens that all parts of a media business can unite around to work together for growth. I have written about audiences as a unifying force.
My focus within AI is on finding new ways to engage audiences and new ways to create impactful journalism that uncovers information that was out of reach before. AI is a continuation of a revolution that is well underway in society in which the collection and existence of copious amounts of data and ability to analyze that data are driving innovation, new insights and change. AI is the next step in the Data Era.
If past technology revolutions are a guide, AI will ultimately expand the economy and in that expansion, there are many new things that we as an industry can do to be relevant and useful to the public. So even while the legal teams at media companies are working out the copyright and web site scraping rules for the AI age, the rest of those media organizations must grapple with what they will do as AI becomes more widespread. It’s a bit of playing defense and offense at the same time: the legal terms around scraping will help media companies defend their property and capture the value of it. But media teams also need to go on the offensive and figure out how they will use AI to create new value.
To be prepared for AI, media companies need to:
- Complete the basics of Digital Transformation 101 — There are some common steps of digital transformation that many news organizations have already done, and these are more important than ever as the Data Era moves forward. These steps include breaking down some of the silos that pervade news companies, creating common audience-focused KPI’s, decreasing focus on old mediums (text, television, radio), etc. Also important is to undertake cultural change from top to bottom — digital products give news companies more dials than ever before to deliver the most relevant news at the right time. To figure out which dials to move and when, you need to look to data for at least part of the guidance. This requires cultural change not the least because senior editors at news companies often make news calls based on their gut. Read my Post on the Elements of New Company Digital Transformation here, and get going on changes if you have not already progressed in these areas.
- Get Your Data in Order — Media companies that have focused on having easy-to-access data in their data lakes are in better shape to participate in the Data Era. This is a task for the full business — content, technology, advertising, subscriptions and customer services all play an essential role in data collection and have large opportunities to use it in AI-powered innovation. What do I mean by “data?” It’s not just numbers, as some people might think. In the Data Era, data is actually any input to a model. So for a media company, data includes all customer information that is collected, all content (text, video, audio, photos, graphics), and platform usage data.This data has to be accessible and not split up among multiple hard-to-access platforms. It’s also important to have the data pipelines built between your platforms and data lake to enable your tech stack to analyze and deploy product features in real-time in response to new user actions. To do all this, you need to be sure you have hired good data scientists and data engineers with knowledge of machine learning and AI.
- Listen and Learn about AI — Media leaders in all disciplines — news, opinion, marketing, advertising, human resources and finance — need to understand more of the technology basics of AI. How could the CFO possibly discuss the large-scale investments that will be needed to do any sort of customization of a pre-trained model like Llama or Falcon? Are you saying “huh?” at this point as I mention Llama and Falcon? If so, contact me if you’d like a tutorial for your leadership team. Part of learning about AI will be actually digging in and using it. R&D is not a luxury at this point — it is a necessity. Some of the use cases you might want to try first are non-public facing internal tools … or you can find use cases in public-facing areas where the stakes are low. A central decision media companies will need to make is around what large language model they will ultimately use — something off-the-shelf or something they partially train, and whether the model they use comes from the open source world or comes via a license deal with a commercial provider. My AI Tech Basics post will give you the language to help you start thinking about this choice.
- Focus on Better and Different — AI should not be viewed solely through a prism of productivity and cost-cutting. In the newsroom and product teams, focus on what new product features and content you can offer using AI to create better audience engagement. And for internal reporting tools, focus on reporting tools to uncover impactful stories that would have been out of reach before, or on internal systems that give your business-side teams helpful intelligence to do their jobs better. As examples, here are some early AI deployments we did at the WSJ.
Ultimately, over the next five years, media companies that creatively use AI to drive more impactful journalism and audience engagement are the ones that will come out ahead.
— — — -
Please read my other posts on AI and News
