Nick Diakopoulos

How can you know if your content was used to train an AI model like GPT-3 or 4? In this post I show that OpenAI's models appear to have memorized boilerplate text from the New York Times, Washington Post, and The Economist. https://generative-ai-newsroom.com/finding-evidence-of-memorized-news-content-in-gpt-models-d11a73576d2

Interested in doing a PhD in Human Computer Interaction, Computational Social Science, or something else at the intersection of Tech + Society? The School of Communication and School of Engineering joint program in Technology and Social Behavior (TSB) is hosting a virtual information session to answer your questions. Register and join to learn more! http://tinyurl.com/tsbphd

@cyberlyra@mastodon.social @josh@fediscience.org @jeffjarvis@mastodon.social there are bridging initiatives like this one: https://journalistsresource.org. If you want your own work part of the public debate IMO you’re best off writing the op ed yourself and pitching it. Why be mediated when you can write it directly as the expert?

What Could ChatGPT do for News Production? I went through the Associated Press's list of news tasks to see how generative AI could be used. I would not recommend using ChatGPT for these tasks without an editorial process for fact-checking before publication. @ndiakopoulos@medium.com

One of the questions that comes up over and over again for newsrooms is what guidelines to put in place for using generative AI. Working together with @hannescools@mastodon.social we analyzed 21 existing guidelines documents and also made a few suggestions for others thinking about developing their own: https://generative-ai-newsroom.com/towards-guidelines-for-guidelines-on-the-use-of-generative-ai-in-newsrooms-55b0c2c1d960

The goal of the Generative AI in the Newsroom project is to collectively develop the capacity for responsible use of generative AI in news production. Follow here: https://generative-ai-newsroom.com/the-generative-ai-in-the-newsroom-project-2b6682baf2e8

"it is science that has failed to understand the public, in terms of what and how to communicate" This article presents a helpful model for thinking about how to address this: https://issues.org/mental-models-scientists-risk-communication-morgan-fischhoff/

"In this way we were able to add very useful content to the rule-based auto-generated articles, and augment the articles with insights and explanations presented in the company’s financial reports."

News document summaries in Danish using Open AI Models, published by Ekstra Bladet. With enough tries it works, but still very much a need for human in the loop. New from Tore G. C. Rich on the Generative AI in the Newsroom Blog: https://generative-ai-newsroom.com/summaries-in-danish-with-openai-cbb814a119f2

Can GPT-4 help cover news events? Here's an event summary of a recent AI Media Democracy Lab panel on the legal aspects of generative AI, using Whisper for transcription and a few targeted prompts to GPT-4. Don't forget the human in the loop: https://generative-ai-newsroom.com/can-gpt-4-help-cover-events-564c1e87e987

Ever wondered about using generative AI but weren't sure about the terms of use? Here's @Nhelberger@someone.elses.computer with a deep dive legal analysis of the ToU for five model providers: Open AI, Midjourney, Anthropic, Hugging Face, and StabilityAI: https://medium.com/generative-ai-in-the-newsroom/generative-ai-in-media-journalism-think-big-but-read-the-small-print-first-375f2ecb1256

Generative AI is a potentially powerful tool to help illustrate news articles and visually draw in readers. Check out @lydiachilton's new primer on how to get started: https://generative-ai-newsroom.com/news-illustration-with-generative-ai-4955a71cbd83

Here's the call for participation for the Computation + Journalism / DataJ conference: http://2023.cplusj.org -- March 30th deadline for abstracts, workshops, panels, and refereed papers! Conference in Zurich June 22-24.

Here's an extended primer and FAQ on copyright law and generative AI written for journalists and news organizations, with suggestions for responsible use. By João Pedro Quintais and me. Read here: https://medium.com/generative-ai-in-the-newsroom/a-primer-and-faq-on-copyright-law-and-generative-ai-for-news-media-f1349f514883

@j2kun@mathstodon.xyz I believe there are also some result out there about visuals attracting homepage clicks. Makes sense that people’s attention is attracted to pictures…

What if AI could scan the world for events and information and send an alert when something looked interesting? In this post I look at how to use @OpenAI models to rate the newsworthiness of scientific abstracts. TLDR, it works pretty well! https://generative-ai-newsroom.com/finding-newsworthy-documents-using-generative-ai-a43a41a90e30

Can We Trust Search Engines with Generative AI? A Closer Look at Bing’s Accuracy for News Queries. @ndiakopoulos@medium.com

🚨New Essay🚨 ChatGPT and the AI Act -- in which
@Nhelberger@someone.elses.computer and I think through how the AI Act falls short when trying to address generative and general purpose AI: https://policyreview.info/essay/chatgpt-and-ai-act

New from @sachnishal.bsky.social on the Generative AI in the Newsroom blog -- fine-tuning, prompting, and evaluation of news angle generation: https://generative-ai-newsroom.com/getting-the-science-and-the-scoop-with-news-angles-from-gpt-3-b1b4fcd7f6b1

Here's an inspiring story of how @MarshallProj is using generative AI to help structure and summarize complex bureaucratic policy documents for its audience: https://generative-ai-newsroom.com/decoding-bureaucracy-5b0c1411171

News article headlines suggested using OpenAI's models in the YESEO app were helpful for a majority of journalists in their editorial process. Read about the study from Ryan Restivo here: https://generative-ai-newsroom.com/how-the-yeseo-app-tested-gpt-models-with-users-94838bad328

@j2kun@mathstodon.xyz yes, there are SEO results on the value of visual illustration for attracting attention (clicks)