In this episode, Illia Polosukhin joins Henrik and Jeremy to trace the origins of transformers and how practical constraints inside Google led to a breakthrough that reshaped modern AI. He explains why recurrent models were hitting limits, how parallel attention opened the door to scale, and why he believed a major jump in capability was imminent long before the rest of the world saw it.The conversation then turns to the risks and responsibilities of today’s AI systems. Illia describes how models can be subtly guided to influence user opinions, why open weights are not the same as truly open models, and how hidden behaviors can be embedded during training. He explains why provenance and verifiable data pipelines matter, especially as AI begins mediating more of the information we rely on.Later in the episode, Illia outlines how blockchain can support trust, identity, and coordination in a future where AI agents act on our behalf. He shares why information is becoming more valuable than money, how ownership of personal AI models will shape user agency, and why domain expertise becomes significantly more powerful when paired with modern generative tools.Key Takeaways:Transformers emerged from practical constraints, not theoryIllia explains that the shift from recurrent networks to attention was driven by speed and parallelization needs at Google, not a desire to invent a new paradigm.AI’s step change was foreseeable to early buildersIllia expected a ChatGPT level breakthrough several years before it arrived, based on clear research signals and accelerating model performance.Provenance and trust will define the next phase of AIAs AI systems can be subtly manipulated, Illia argues that verifiable data pipelines and transparent training processes are essential to prevent large scale misinformation.Ownership and identity matter in an agent driven worldIllia believes individuals will soon rely on AI agents that act autonomously, making it critical that users own their models and that interactions between agents are secured and verified.https://near.ai – NEAR AI Cloud and Private Chat products are now live, try them hereIllia's X: x.com/ilblackdragonIllia's Substack: ilblackdragon.substack.comNEAR X: x.com/nearprotocol00:00 Intro: AI and Information Control00:29 Meet Illia Polosukhin: Co-Author of 'Attention is All You Need'01:03 The Evolution and Impact of AI13:24 The Birth of Near AI and Blockchain Integration15:16 Challenges and Innovations in Blockchain and AI22:17 Privacy and Security in AI Applications26:58 Exploring Sleeper Agents in AI29:19 Practical AI Implementation in Teams30:06 AI's Role in Product Development31:41 Challenges and Future of AI in Development36:35 AI and Economic Alignment41:46 The Future of AI Agents44:14 Debrief📜 Read the transcript for this episode: For more prompts, tips, and AI tools. Check out our website: https://www.beyondtheprompt.ai/ or follow Jeremy or Henrik on Linkedin:Henrik: https://www.linkedin.com/in/werdelinJeremy: https://www.linkedin.com/in/jeremyutley Show edited by Emma Cecilie Jensen.

Ask this episode

“What did the council decide on item 42?” — answered with the speaker and the moment.

Transcript

Related