🔆 The Art of Fair Exchange: Exploitative AI and the Data Commons

Was this email forwarded to you? Sign up here

🗞️ Issue 11 // ⏱️ Read Time: 9 min

Hello 👋

Data: A resource built by humans

This week we are looking at a growing controversy. As you probably know by now, the value of Generative AI models is, to a large extent, dependent on the training data. This training data is built by humans, in one way or another. Most of the content on the internet, including the images you’ve uploaded on your social media, news articles, chat forums, and artwork, is gathered by crawlers that automate the process of data collection. This data is largely collected without permission, credit, or compensation to the contributors and makers of this crucial resource. 

This is causing tension. Lately, it seems like AI litigation is becoming hotter than generative AI itself, with numerous cases highlighting that credit should be recognised when credit is due: There is a growing concern over the extraction of this data without clear consent, acknowledgment, or financial reparation to those who own it or are featured within it. Rich Skrenta who is the head of CommonCrawl, a tool scraping the internet for data, expresses the view of many AI companies: “You posted your information on the internet, intentionally so that people could come and see it. And robots are people too.” He continues: “By publishing something on the internet without explicitly telling robots to avoid it, you’re consenting to its use by AI.” 

The Consequences

These exploitative ways of collecting training data not only raise questions about privacy and consent but also the fairness of profiting from communal or individual contributions without fair compensation. Some of the consequences of these methods are:

Economic Inequity: AI companies are reaping the benefits, while creators and contributors of data are getting left out of the economic value chain.

Trust Erosion: The unchecked use of data leads to a loss of trust in AI technologies, and the companies developing them. Subsequently, this hinders innovation and the adoption of a technology that could potentially help us solve some of the most pressing problems we face, including climate change, biodiversity loss, and poverty.

Cases from reality

👩🏽‍🎓 A student at the Stockholm School of Economics is developing an AI product aimed at assisting teachers in their work. The objective is to commercialise the product for school use. Utilising the principle of public access to records—a cornerstone of Swedish democracy—the student has obtained historical examination papers from schools to train the AI model. However, the crux of the matter lies in the intended purpose of this democratic principle, which is to ensure transparency and accountability within the government, enabling public scrutiny of government officials' and agencies' actions. In this case, there are concerns regarding the potential misappropriation of resources (data) for the company’s self-interest. 

⚖️ The New York Times, a leading American daily newspaper, is suing OpenAI and Microsoft for allegedly using its copyrighted content without permission. The lawsuit, filed at the end of last year, contends that millions of articles published by the Times were utilized to train ChatGPT and other AI products developed by the defendants. Although the lawsuit does not specify an exact amount in monetary demands, it asserts that OpenAI and Microsoft should be liable for "billions of dollars in statutory and actual damages" due to the "unlawful copying and usage of The Times's uniquely valuable works."

Responsible approaches