ByteBrief
We're a portrait publication through and through. Turn your phone back and your briefing picks up right where you left it.
(We tried widescreen once. It wasn't us.)
Researchers tested four open model families including Qwen2.5 and SmolLM2 on random coin flip sequences. After ten heads in a row, the models predicted an 11th head with probabilities reaching up to 82%. This finding suggests small language models exhibit a gambler's fallacy bias even when explicitly told the coin is fair, contradicting ideal learning expectations for truly random data.
Tap to vote and see what everyone thinks.
Summary by ByteBrief