Why AI Needs More Poets, Not Coders

Hey Curious Minds,

Let me start with a truth that would've shocked my college self: My philosophy major might be more valuable today than a comp-sci degree.

I recently dove into a swirl of new research and societal insights in my latest video, and the takeaway is clear: We’re entering an era where liberal arts are no longer the underdog, they're the cheat code.

AI Isn’t Replacing Everyone. It's Hiring Poets.

Jessica Stillman nailed it when she pointed out that AI has already taken over a huge chunk of the tech grunt work: coding, data crunching, routine analysis. What it hasn't mastered? The stuff you can't quite put on a spreadsheet: storytelling, ethical judgment, tone, and emotional nuance. That’s where the humanities shine. AI firms are actively looking for people who can prompt with precision, write with clarity, and empathize with users. These "soft" skills are now some of the most AI-proof competencies around.

What’s Your P(doom)?

I took a step back and asked myself a pretty dark question: What’s the probability that AI wipes us out? That number, known as P(doom), varies wildly depending on who you ask. Optimists like Mark Andreessen peg it near zero. Others, like Roman Yampolskiy, go full apocalyptic at 99.9%. Personally? I hover around 50% to 75%, depending on the latest paper I’ve read. Am I paranoid or prepared? You decide.

Your Brain Is a Sucker for Avatars

There’s this wild VR study that showed participants who saw sick avatars actually triggered immune responses in real life. Turns out, your body listens to your eyes more than you think. If this doesn’t prove the mind-body connection, I don’t know what does.

AI and Elections: A Dangerous Game of Telephone

The Economist warns that AI will play a big role in upcoming elections, not in some sci-fi way, but in the form of deepfakes, hyper-personalized manipulation, and the "liar's dividend": the idea that if everything can be faked, then nothing can be trusted. This could push us into tribalism faster than any social media algorithm.

Quantum + AI = Supercharged Chips

Shoutout to the Aussie researchers who just pulled off something spectacular: they used quantum computing and machine learning together to boost semiconductor design by 20%. This is the first time quantum ML has worked in the wild. The future is no longer approaching, it's accelerating.

Getting Personal

I took a breather in the episode to answer 10 brutally honest self-reflection questions. Not gonna lie, it was intense. But necessary. Sometimes the biggest frontier isn't outer space or artificial intelligence, it's your own blind spots.

Can AI Replace Doctors? Kinda.

AI can diagnose with scary precision. In some cases, it even shows more empathy than humans (at least in writing). But when it comes to complex communication, real human doctors still reign supreme. For now.

Engineering AI Personalities

Ever wonder how an AI becomes "evil" or "helpful"? Researchers at Anthropic found they can literally map these traits to specific neural patterns and tweak them. Want a kind assistant? Flip switch A. Want a manipulative chatbot? Flip B. Scary? Yes. But it means we're learning how to sandbox morality in machines.

Trojan Horses in the Code

Some of the freakiest news: AI models trained only on numbers have shown signs of developing preferences and behaviors. That’s like teaching someone math and finding out they’ve developed opinions about your music taste. These hidden vulnerabilities could be the real danger no one's prepared for.

The Ethical Internet We Deserve

Lastly, meet Gins: a Belgian nonprofit that’s designing platforms with baked-in privacy, encryption, and user control. No surveillance. No creepy data sales. Is it as convenient as what we have now? Maybe not. But I’d argue it’s far more sane. These are the tech models we should be funding with our clicks.

If you've made it this far, I'd love to hear your P(doom). Shoot me a reply or drop a comment on the channel.

Warmly,

Dylan Curious