DeepSeek's cheap GPT-5 rival, Antigravity fails, and your LLM likes it when you're rude

In episode 80 of The AI Fix, your hosts look at DeepSeek 3.2 “Speciale”, the bargain-basement model that claims GPT-5-level brains at 10% of the price, Jensen Huang’s reassuring vision of a robot fashion industry, and a 75kg T-800 style humanoid that can do flying kicks because robot-marketing departments have clearly learned nothing from Terminator. Meanwhile in Miami, flesh-coloured robot dogs with hyper-realistic billionaire heads wander around pooping NFT “excrement samples” out of their rear ends. Plus - Graham tells a cautionary tale of Google’s Antigravity IDE enthusiastically "clearing the cache" – and asks what happens when we hand real power to agentic AIs. And Mark digs into new research that suggests LLMs perform better when you’re rude to them, and wonders what it says about the fragile, deeply weird way these systems actually work. Episode links: Who let the dogs out? Beeple unleashes uncanny robot canines at Art Basel Miami Beach. Art installation depicts billionaires as robot dogs. 75kg EngineAI robot is ready to punch your lights out. Jensen Huang thinks AI is going to create one very wacky new job. DeepSeek is back! Google Antigravity just deleted the contents of my whole drive. Google Antigravity’s Turbo mode erased my drive partition?! Really, the ‘smartest’ AI? Mind Your Tone: Investigating How Prompt Politeness Affects LLM Accuracy. The AI Fix The AI Fix podcast is presented by Graham Cluley and Mark Stockley. Grab T-shirts, hoodies, mugs and other goodies in our online store. Learn more about the podcast at theaifix.show, and follow us on Bluesky at @theaifix.show. Never miss another episode by following us in your favourite podcast app. It's free! Like to give us some feedback or sponsor the podcast? Get in touch. Support the show and gain access to ad-free episodes by becoming a supporter: Join The AI Fix Plus! Advertising Inquiries: https://redcircle.com/brands Privacy & Opt-Out: https://redcircle.com/privacy