Big tech is training AI on junk data: Intuition

The post Big tech is training AI on junk data: Intuition appeared on BitcoinEthereumNews.com.

AI models are getting more powerful, but the data they’re trained on is getting worse, says Intuition founder Billy Luedtke. Summary AI is only as good as the data we feed it, says Billy Luedtke, founder of Intuition We’re in a “slop-in, slop-out” era, as AI becomes recursive Decentralized models have the edge with tech and user experience As AI systems grow more pervasive, users are increasingly running into limitations that are hard to fix. While the models improve, the underlying data these models are trained on remains the same. What is more, recursion, or AI models training on data generated by other AI, might actually make it worse. To talk about the future of AI, crypto.news spoke to Billy Luedtke, founder of Intuition, a decentralized protocol focused on bringing verifiable attribution, reputation, and data ownership to AI. Luedtke explains why the current data sets for AI are fundamentally flawed and what can be done to fix it. Crypto.news: Everyone right now is focused on AI infrastructure — GPUs, energy, data centers. Are people underestimating the importance of the trust layer in AI? Why is it important? Billy Luedtke: 100%. People are definitely underestimating it — and it matters for several reasons. First, we’re entering what I call a “slop-in, slop-out” era. AI is only as good as the data it consumes. But that data — especially from the open web — is largely polluted. It’s not clean. It’s not reflective of human intention. Much of it comes from gamified behavior online: likes, reviews, engagement hacks — all filtered through attention-optimized algorithms. So when AI scrapes the internet, what it sees isn’t a holistic picture of who we are. It’s seeing people playing the platform. I don’t behave the same way on Twitter as I do in real life. None…