Howard Marks just dropped a new memo called "AI Hurtles Ahead" a few days ago. He's 78 years old managing $200B+ at Oaktree. Let’s see what this OG thinks about A.I. Here are the key takeaways: • He had Claude write him a 10,000 word AI tutorial. It referenced his past memos, anticipated his counterarguments, and used humor. Howard’s in AWE of A.I. • Claude argued back at him: "Howard, everything you know came from other people. Graham taught you margin of safety. Buffett taught you quality. Munger taught you mental models. How is what I do structurally different from what you did?" • The economic question isn't "does AI truly think?" It's "does AI do the work?" And right now, the answer is yes, • AI has almost every quality of a great investor. Processes more data, no fear, no greed, no anchoring bias. Where it's weak are truly novel situations with no historical pattern to draw from. I’d argue it can be better because they can process more data and aren’t subject to as much emotional biases as we are. • Your edge can't be quantitative information anymore. Everyone has it. AI processes it better than all of us. The only edge left is qualitative judgment. • AI now operates at 3 levels. Level 1: chat (saves you thinking time). Level 2: tools (saves you execution time). Level 3: autonomous agents (does the entire job). We just hit Level 3 It makes me wonder what level 4 is? I think it’s when A.I. agents are proactively setting the goals. • Stop thinking of AI as a search engine. It doesn't just retrieve information. It synthesizes and reasons. That's a fundamentally different thing • AI hallucinates and doesn't always know what it doesn't know. But neither do most people. The bar isn't perfection. It's better than the average knowledge worker. And it already is Whenever people complain about A.I., I tell them to think about the Will Smith Spaghetti video. • No technology in history has been adopted this fast. 400 million users and 75 to 80% of companies in under 2 years. Society literally cannot retrain fast enough to keep up with the jobs AI displaces • On whether AI is a bubble: the technology is real and probably underestimated. But real technology doesn't mean fair prices. Even Claude couldn't make a case that current AI valuations are appropriate • His advice: don't go all in and risk ruin if things go badly. Don't stay all out and miss one of the greatest technological shifts ever. Be selective. Be prudent. Think the most practical way to capture upside is to have a basket of A.I. stocks or QQQ. • He closed with this: a friend told him he'd rather be an optimist and wrong than a pessimist and right. Marks said "me too. I wish I could be confident that my worrying is unwarranted" This memo matters because Marks isn't a tech bro trying to pump his portfolio. He's a 78 year old value investor with nothing to prove.(Edgy - The DeFi Edge 🗡️)