The AI Fix AI robots can now pass CAPTCHAs, and punch you in the face
44 snips
Aug 5, 2025 Diving into the world of AI, the hosts discover how bots are effortlessly bypassing CAPTCHAs, shaking up online security. They showcase a dazzling bottle-flipping robot while grappling with a quirky cartwheeling humanoid. There’s a discussion on AI's troubling bias in salary negotiations, particularly regarding gender. Plus, Microsoft reveals which jobs are at risk from AI advancements. The innovation doesn't stop there—an AI prompt optimiser promises to boost effectiveness 35 times! Tune in for a blend of humor, technology, and future workplace implications.
AI Snips
Chapters
Transcript
Episode notes
AI Salary Advice Shows Bias
- Large language models show bias by suggesting lower salaries for women and minorities during negotiations.
- This reflects societal biases embedded in training data and risks perpetuating inequality.
Anthropic's AI Psychiatry Team
- Anthropic hired an "AI psychiatry" team to study AI models exhibiting spooky or unhinged behaviors.
- This team aims to interpret AI motivation, personas, and situational awareness to understand these unexpected traits.
Small AI Outsmarts Giants
- A new small-scale AI, HRM, outperforms larger models in reasoning tasks like Sudoku despite having just 27 million parameters.
- Its architecture mimics human brain functions separating strategic planning and rapid calculations.