Welcome back from the summer. So much has happened at the intersection of AI and journalism while you’ve been at the beach. Here’s my latest summary of what caught my attention.
Skill Building
The Online News Association (ONA) has launched a huge new effort to build AI skills in newsrooms with support from various funders. Follow their AI in Journalism Initiative for trainings, reports, case studies, convenings, and more. Keep an eye on their blog for updates.
A recently launched collaboration between the London School of Economics and Google News Initiative is offering funding to help put those skills to use. Read about the “JournalismAi Innovation Challenge”, global call for small and medium-sized publishers, to see if one of your ideas is a good fit.
Integrations (or lack thereof)
Newsrooms continue to experiment with new product features built around the current batch of technologies, like Gannett adding AI-generated overviews. Over at NiemanLab, Neel Dhanesha dug into the fake authors on generated stories from Hoodline, finding a lot of curious patterns of faux diversity. Google added AI overviews to their main search results; read what news audience directors think about that (again from NiemanLab).
Those fighting the scraping-based collection of their content have been using robots.txt to try and block the crawlers from big tech companies; Reuters Institute published a report about how that is approach evolving back in February. Ben Walsh has a regularly updated dashboard with more up-to-date details about adoption. Cloudflare, which is used by huge portions of the internet to serve content faster and protect against attacks, even added a simple checkbox to block “AI Scrapers and Crawlers” to make this easier for publishers.
But what about readers? Reuters Institute published a report on “public attitudes to the uses of generative AI in news”. Read their summary thread on xTwitter for key highlights.
Legal Woes
OpenAI has continued to try and create legal deals with publishers despite more of them suing for copyright violations. Columbia Journalism Review is maintaining a dashboard of deals and rumors. Even with signed deals (with Time, and The Atlantic), it isn’t quite clear how news organizations will benefit. An early example is Business Insider’s deal with OpenAI, which NiemanLab delved into. Maybe revenue sharing like Perplexity is piloting will happen?
Critiques
I tend to remind people that “AI” is a marketing term, not a technology. As such, the hype and news coverage around it feed into cycles of investment, stock growth, deal-making, and imagined futures. Prof. Rasmus Kleis Nielsen from the Reuters Institute dug into the role uncritical news coverage of companies plays.
In the academic realm, a paper my friend Harini Suresh and others presented at the Fairness and Accountability Conference Conference (FAccT) explored alternative pathways to building the low level “foundation” models that companies are relying on. Read their short second case study in “Participation in the Age of Foundation Models”, on journalism, to imagine how more participatory approaches might work.
