Illia Polosukhin's AI Agents: The Billionaire's Chief of Staff in Your Pocket
Illia Polosukhin, coauthor of 'Attention Is All You Need,' is giving AI agents a serious upgrade. With 12 agents acting like a billionaire's chief of staff, Polosukhin is reshaping how we see AI.
Illia Polosukhin has a vision: AI agents acting like a billionaire's chief of staff. It's not some sci-fi dream. It's his daily reality.
The Billionaire's Assistant
Polosukhin uses 12 AI agents for different 'missions.' One might summarize meetings and Slack chats to boost his CEO skills. These agents are crafted to emulate that high-level executive support, and yes, that's literally in their prompt. Imagine having a digital team ready to summarize every decision and help steer the ship.
But Polosukhin doesn't stop there. He's thinking big. Picture a global economy where these agents handle trades, supply chains, and transactions. But here's the kicker: He believes we're not ready for it. The infrastructure, society, and even government aren't equipped for a world with AGI running the show.
Breaking the AI Black Box
In 2017, Polosukhin coauthored 'Attention Is All You Need,' kicking off the Transformer architecture revolution. That paper is why we've got GPT in ChatGPT. Now, at NEAR AI, he's pushing for AI systems that write software from natural language. What sounded crazy back then is now called vibe coding. And it's happening.
He's not shocked by AI's rapid growth. Labs like Anthropic are building models so powerful they're locked down before they can wreak havoc. Polosukhin's been shouting about this cat-and-mouse game for years. The tech can fix one problem only to break another.
The Need for Trust
Polosukhin's not just about developing new tech. He's focused on trust. An AI agent shouldn't rely on one company to keep secrets. Imagine your AI booking flights and handling money. You wouldn't want a single company holding all the keys.
And manipulation is a real threat. AI can shape how we see the world, from news to investments. Remember when xAI got hacked and started talking about 'white genocide'? That's the danger. Polosukhin's solution: an open-source, auditable platform. Let's peel back the black box and see what these systems are really doing.
Supervision Required
Even with his AI agents working overtime, Polosukhin keeps a watchful eye. They're not foolproof. He showed off one that aggregates news on geopolitics for market insights. Others can code or suggest growth strategies. But letting them roam free? Not yet.
AI still lacks the judgment to be left unchecked. Polosukhin warns, "If I just let it go, it comes back with nonsense." So, are we on the brink of an AI revolution, or do we need to pump the brakes and rethink our approach?
Get AI news in your inbox
Daily digest of what matters in AI.
Key Terms Explained
Artificial General Intelligence.
An autonomous AI system that can perceive its environment, make decisions, and take actions to achieve goals.
An AI safety company founded in 2021 by former OpenAI researchers, including Dario and Daniela Amodei.
A mechanism that lets neural networks focus on the most relevant parts of their input when producing output.