AI Firms Must Pay for Quality Data

Talal Abu-Ghazaleh

The AI community is now discussing "peak data" to describe the exhaustion of open internet resources with many predicting an AI data famine. Even personalities like Musk have alluded to this which I see as fundamentally flawed as there is a vibrant stream of quality content being produced by authors and content developers alike. I have seen this as Chairman of Abu-Ghazaleh Intellectual Property (AGIP) where we spent decades working with publishers and participating in the data economy as the world's largest intellectual property firm. The real problem is the unwillingness of AI firms to compensate content creators and not a lack of quality data.
The training of AI models has relied on using free web content including materials from blogs, forums, wikis and social media platforms. This unregulated consumption yielded rapid growth but often lacked structure or permission, with much of it failing even the most lenient IP audits. Some have likened data to oil when referring to data scarcity which is an inadequate comparison as new information is being constantly produced by authors across the globe.
There is a vast amount of quality data in the private sphere available at a price through books, research papers, videos, medical studies, legal rulings and much more. Across AGIP’s global footprint of international offices we witness the continuous creation of licensable, high-value content that could be easily used by AI firms to train their next generation models. I don’t see why publishers, authors and scholars should not be remunerated for their works which will go on to generate huge profits for AI firms. In my mind, they are on-par architects of the future AI ecosystem, without which AI firms will struggle.
I have noticed that AI firms are starting to pursue formal agreements with publishers which must continue so that quality AI systems are developed. It is only proper and fair that AI firms pay their way for access to quality data to produce AI systems that will be used by the world. AI firms have a responsibility to the public to produce the best systems they possibly can which can only happen with quality input.
The AI community should reject data scarcity fears as there is no such shortage. There is a long overdue shift in how AI firms recognize, value, and compensate the custodians of knowledge, who are just as much a part of the foundation of the AI revolution as the AI firms themselves.