I know. It’s not summer everywhere. But this week, we’re changing up the fare around here. You can think of this one as a super-soaked version “Further afield on the wide, wide Web,” the link roundup section of my newsletter.

The content here isn’t necessarily fresh from the Internet presses — though some of it is — but rather these are pieces that are either good foundations, good overviews, or great think pieces that stand the test of time (time in these parts is rather accelerated since the technology moves quickly). 

I understand that you’re probably at the pool, hesitating between watching your kid show off their newly minted diving skills, a good gossip magazine, and some of the links on this list. I can assure you, however, that you’re not going to learn anything about the Kardashians that could possibly surprise you at this point, but yes, this newsletter cannot compete with watching the fruit of your kid’s efforts.

But first, news this week (vacation edition)

Just so you can keep up with the latest … 

• OpenAI commits 5 million dollars to local news partnerships with the American Journalism Project (Venture Beat). Taking a page from the playbook of other large tech platforms before it. 

• How Do the White House’s A.I. Commitments Stack Up?  (The New York Times). Last week, the White House announced it had secured a “voluntary commitment’ ” from leading tech companies on how they would use and develop their technology.

Catching you up with a sober, clear view on generative AI

Let’s get situated, first of all. Certain things on the Internet are available in abundance: cat videos, racist rants, and hours upon hours of folks streaming their video game prowess. And now, also, armchair AI experts with either a piece of software to sell to you or a potion to project you against AI.

• “The next generation of AI?” (InPublishing). Here’s Charlie Beckett, the director of the JournalismAI project at the LSE, providing an efficient analysis of just how much of a revolution — and how much not of a revolution — we’re talking about as we enter this age of generative AI. 

Our robots, ourselves

• You are not a Parrot (New York Magazine). This article got suggested to me by several people … who don’t work in data. Ever a good sign. But don’t think this is easy access because in order to truly appreciate it, you’ll be pulling in thoughts from semantics, linguistics, and even philosophy (what does the idea of “meaning” mean when we think of the meaning of words?) 

• AI is a lot of work (New York Magazine/The Verge). A look at the world of companies which provide the cheap workforce that’s at work, tagging anything and everything on the Internet.
• My AI-writing robot (The New Yorker). A staff writer for The New Yorker had a robot trained using her own material so it could acquire her writing “voice.” But, it turns out, your thoughts — and not just your voice — are what makes your words yours.
• Transformers: the Google scientists who pioneered an AI revolution (The Financial Times). The story of a system architecture, dubbed only the “transformer,” which enabled an inflection point of all large language models. This story is great if you want nerdy insights into the breakthroughs of transformer; but it’s also great as a human story of innovation.

One thing is for sure: death, taxes, and more junk on the Internet

• AI is killing the old web, and the new web struggles to be born (The Verge). You’ve certainly by now heard the view that one place of concern for AI content was the specific combination of inaccurate content multiplied by the ability to scale. In this essay, the author looks at various examples of how machine-generated content is evaluated and how certain companies that use user-generated content (like Stack Overflow) are approaching it. In this piece, by the way, we’re “the old Web.” But I’d like to think of us as the “artisanal web.”

• How to Prepare for the Deluge of Generative AI on Social Media (Knight First Amendment Institute at Columbia University). An extensive review of the various nefarious or problematic ways AI can be leveraged but also some ideas and approaches for mitigating and detecting these uses. While technically an academic paper, this one is practical, easy to read, and will bring you up to speed even as it will also, probably, bring you down.

Guidelines and helpful tips

• Towards Guidelines for Guidelines on the Use of Generative AI in Newsrooms (Generative AI in the Newsroom). A great collection of examples from news organisations around the world taking a stab at introducing and scaling the usage of AI in their organisations. 
• A Primer and FAQ on Copyright Law and Generative AI for News Media (Generative AI in the Newsroom). This guide from Nick Diakopoulos and João Pedro Quintais got a strong shout-out at the Nordic AI conference I attended a few weeks ago: a primer on copyright law as it pertains to generative AI. I have added it to my Notion app; it’s just that good and useful.

The law and compliance

• Stanford’s Center for Research on Foundation Models published an analysis of the current level of compliance of various large language models. I shared this recently but it’s worth a rerun. A cheat sheet with a matrix of the level of compliance of various commercial LLMs against the current draft state of the EU’s AI act. 

• Generative AI in Media & Journalism: Think Big, But Read the Small Print First (Generative AI in the Newsroom). Professor Natali Helberger, distinguished university professor/law and digital technology at the University of Amsterdam, reviews the term of use of five large generative AI platforms: Open AI, Midjourney, Anthropic, Hugging Face, and StabilityAI.

What our speakers recommend

And some great recommendations from some of our speakers for October’s Data master class series.

From Martin d’Halluin, deputy counsel, News Corp: Attention Allocation in Information-Rich Environments: The Case of News Aggregators by Mihai Calin, Chrysanthos Dellarocas, Elia Palme, Juliana Sutanto (Boston University School of Management). Martin says: “I have been fascinated by the lack of data and research on click-through rates, which is the key metric for digital news publishers. This paper proposes a methodology to study the impact of the length of snippets on clicks.”

From Christopher Reher, data general manager, Axel Springer All Media: Snow Crash by Neil Stephenson. Christopher says: “Great, deep and very funny read. Coined ‘Metaverse’ as a thing and established ‘Avatar’ as a digital replacement for oneself. Way ahead of its time when published and now in parts reality.”

From Anna Vissens, lead data scientist, The Guardian: Invisible Women: Data Bias in a World Designed for Men by Caroline Criado Perez. Anna says: “This book not only illuminates a tremendous gender gap which still exists in our modern society and its historical roots but also teaches us to look for hidden biases when you handle data.”

From Chris Wiggins, chief data scientist, The New York Times: How Data Happened: A History from the Age of Reason to the Age of Algorithms. Chris’ book looks at the social and historical circumstances for how data came to be used and how it influenced everything from economics, society, to, yes, engineering. It’s a bit like Guns, Germs, and Steel but for data. (Chris didn’t write this, Ariane did.)

If you’d like to subscribe to my bi-weekly newsletter, INMA members can do so here.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .