Data Leverage
Archive
Previously published articles, preserved with their original source links.
45 articles
-
The AI "Evaluation Crisis" Is an Opportunity to Get Data Flow Right
Why the AI evaluation crisis could force a reckoning on dataset provenance, attribution, and consent.
-
April 2026 small points
A running collection of short ideas about sovereign AI, augmentation, evaluation labor, data pipelines, and quasi-enclosure.
-
Attestation across the AI Supply Chain
A proposal for interoperable attestation objects that connect training data, evaluation labor, and AI-generated outputs across the AI supply chain.
-
"People First" Policy Ideas that Complement Each Other (through better data flow)
Reacting to a wide-ranging set of policy ideas from OpenAI.
-
AI is driving the cost of polish down; some musings on fancy versus terse artifacts
AI progress means the "polish" of a figure or website no longer proxies for quality. Can we try to turn this into a good thing for curation, attention allocation, and even AI progress itself?
-
Two natural allies of a "Data Transparency" agenda: capabilities forecasters and social simulators
Making an "if you like X, you might want to support Y" argument for data-focused policy
-
A Short Guide to Data Strikes and Conscious Data Contribution in the Context of 2026 Frontier AI
Back to the basics of data leverage.
-
The Paradox of Reuse in 2026: A Case of Quasi-Enclosure, or "Subsidized Club Goods that Sort of Look Like Public Goods"
How we can understand, and react to, the complicated impacts of AI systems on online communities and knowledge commons
-
The Coding Agent Data Deal
On user data control, coding agents as retrievers, and the value of your coding transcripts
-
Coding agents are (1) a big deal, (2) very relevant to data leverage, and (3) able to help build tools that support data leverage!
Sharing an early reaction to recent coding agent discourse and two relevant projects
-
Almost Everybody -- Including Both Data Creators and AI Companies -- Stands to Benefit from Clearer "Data Rules".
In fact, anyone who doesn't think they will be a "big winner" long term benefits from clear rules, even if it means training data costs more in the short term.
-
How collective bargaining for information, public AI, and HCI research all fit together
Another recap post for the Data Leverage newsletter!
-
Which datasets should we assume are "in all the AI models"?
New model releases keep (re)sparking discussions about training data. What can we assume is upstream in the data river, and what do we want to see happen?
-
On AI-driven Job Apocalypses and Collective Bargaining for Information
Reacting to a fresh wave of discussion about AI's impact on the economy and power concentration, and reiterating the potential role of collective bargaining.
-
How do we know our AI output is good? Double checks, bar charts, vibes, and training data.
Connecting evaluation and dataset documentation via the lens of "AI as ranking".
-
Each Instance of "AI Utility" Stems from Some Human Act(s) of Information Recording and Ranking
It's ranking information all the way down.
-
Google and TikTok rank bundles of information; ChatGPT ranks grains.
Google and others solve our attentional problem by ranking discrete bundles of information, whereas ChatGPT ranks more granular chunks. This lens can help us reason about AI policy.
-
[microblog] One book is worth "0.06%" benchmark points to AI; is "no different from noise". What gives?
Commenting on recent coverage of, and discussion about, Meta's arguments about training data value quantification.
-
Public AI, Data Appraisal, and Data Debates
A consortium of Public AI labs can substantially improve data pricing, which may also help to concretize debates about the ethics and legality of training practices.
-
Evaluation Data Leverage: Advances like "Deep Research" Highlight a Looming Opportunity for Bargaining Power
Research agents and increasingly general reasoning models open the door for immense "evaluation data leverage".
-
Tipping Points for Content Ecosystems
Our AI design choices in 2024 could preclude "Powerful AI" in 2030.
-
AI Labs Should Open Source Data Protection Technologies
There's still incredible tension in the current data paradigm, but sharing "data protection" technologies, like those used by OpenAI to accuse DeepSeek of model theft, can help cut a path forward.
-
Live by the free-content-for-training sword, die by the free-content-for-training sword
There's deep tension in the current ask-for-forgiveness-free-for-all approach to acquiring data for model training. Will "open" models cause this tension to reach a breaking point?
-
Selling AGI like AG1: Will Consumers Push Back Against Proprietary Blends of Herbs and of Data?
The race to produce premiere AI products with high price tags might change the standards around data disclosure.
-
Perplexity CEO's Interaction with Striking New York Times Workers Does Not Reflect Well on the AI Industry
The idea that data-dependent AI systems are ready and willing to crush any leverage from knowledge workers is unlikely to make the AI industry look good to the public.
-
Is Zuckerberg right to say that your specific creative work has no value to AI?
Examining the Meta CEO's claim that the "individual work of most creators isn’t valuable enough for it to matter" in the context of AI training.
-
"Many Models" and "Track Changes" for AI: Some Thoughts on LLM Interfaces
Interacting with many models and harnessing the power of
-
Building a Data Pipeworks for Democratic AI: From Human Knowledge to Records to AI Systems
Focusing on feedback loops -- connecting modern AI to early cybernetics-style thinking -- could help solve looming challenges and support democratic inputs to AI.
-
Will the New York Times Data Strike Have a Large Impact on ChatGPT?
How can we start thinking about how opt-out decisions by content-producing organizations will affect LLMs?
-
A Harbinger of the Future of Content? The New York Times Starts a Data Strike
The New York Times is trying to remove its content from OpenAI models, surfacing tensions around copyright, economic harms, privacy, and the distribution of AI benefits.
-
The WGA Strike is a Canary in the Coal Mine for AI Labor Concerns
Could Upcoming Data Legislation Enable a "Right to Data Strike"?
-
Data Leverage Recap: December 2022 - April 2023
The Last Three Months in Review: What's New and What's Next
-
Plural AI Data Alignment
Measuring the Alignment of AI Systems Based on their Data Pipelines
-
ChatGPT is Awesome and Scary: You Deserve Credit for the Good Parts (and Might Help Fix the Bad Parts)
More on why you're an expert language model trainer
-
Don’t give OpenAI all the credit for GPT-3: You might have helped create the latest “astonishing” advance in AI too
The much-celebrated GPT-3 that can answer questions, write poems, and more wouldn’t be possible without content written by millions of people around the world. Shouldn’t they get some credit?