At 3:14 p.m. on a rain‑slicked Thursday, Boulder‑based startup SynthetiQ released its latest language model, PizzaPredictor‑XL, claiming 97 % accuracy in forecasting when a Domino’s order will arrive. The demo, streamed from a rooftop with a view of the Flatirons, showed the model halting a pizza truck at the exact moment the dough was ready, prompting applause from a crowd that included the mayor’s office—who had a bowl of stale croissants that were still warm—and a group of local meteorologists.
A separate study from the University of Amsterdam’s Cognitive Computing Lab, published Monday, found that the same model can generate a fully personalized bedtime story in under 30 seconds. In a test run, the LLM produced a tale about Sir Whiskers, a tabby cat who becomes a secret agent, complete with a moral about the importance of nap times, and the story was read aloud by a Dutch child wearing a knitted hat.
I tried to ask the model for a coffee recommendation, and it suggested a latte with a side of existential dread.
Because who needs pizza when you can have predictive models?
