AI's Book Hunger: When Tech Takes a Step Too Far

You might have heard a bit of a fuss recently about AI and books. It seems a company, ISBNdb, was trying to get into the AI game by helping create large language models — the kind of AI that powers chatbots like ChatGPT — by feeding them massive amounts of text. The problem? They were apparently using physical books, and in some cases, very rare ones, which sparked a huge outcry.

Now, this company claimed they were simply digitising books, which sounds innocent enough. But reports suggested that some suppliers were being asked for huge numbers of books, far more than usual, and that these books were being 'pulped' or destroyed after their content was scanned for the AI. Imagine cherished first editions or historically significant documents being turned into digital data and then discarded. For many, this felt like a step too far, an act of cultural destruction for the sake of AI development.

ISBNdb has since backtracked, shutting down their controversial AI service and calling it a 'test of market interest'. They've also stated they condemn the destruction of books. This incident highlights a growing tension: the hunger AI has for data, and the ethical lines we need to draw. AI models need enormous amounts of information to learn and improve, but at what cost?

For everyday Australians, this isn't just about old books. It's about how AI companies collect the information that shapes these powerful tools. If AI learns from unethical or destructive practices, what kind of world are we building? It raises important questions about the value we place on physical objects, cultural heritage, and responsible technology development. This event is a wake-up call for the industry to think more carefully about their data sources and the broader impact of their methods.