The AI Revolution: A Bright New Era or a House Built on Plunder?
Acknowledging the problem with how the latest AI models were trained with stolen data
Your coffee break on a regular Tuesday afternoon. Perusing your favorite online magazine, sipping at your mug, you stumble on an article. You find yourself swept up by a scintillating piece about the future of space travel, filled with sharp phrases and sharper insights on the speculative possibilities for tomorrow. You are so enthralled you donβt notice your break time accidentally double. When you finish reading, though, you are knocked back to awareness by something unusual. The byline gives no mention of a human author. Instead, it reads βWritten by GPT-4β in what you instantly realize is meant to be an act of transparency.
Welcome to the brave new world of artificial intelligence, where AI-powered models like GPT-4 can craft articles, write poems, and even pen scripts for TV shows (say goodbye to writersβ strikes, Hollywood). These machines are meant to be the luminaries of our time, their inner workings as intriguing as the outputs they generate. But behind the sheen of this AI revolution lurks a serious dilemma. The data that fuels these models, their material and their lifeblood, comes from the vast expanses of the internet. And by that I mean not just benignβ¦