This New AI Got Stuck Repeating Itself for 2 Straight Hours

WHAT HAPPENED: A person online tried out a newly released AI model called Ternary Bonsai 2 27B, which they were excited about. They gave it a fun, fairly simple request: build a small 3D globe game where a boy scout character walks around exploring different environments. Instead of producing working code, the AI got stuck repeating the exact same thing over and over for more than two hours straight. The user compared it to a headless chicken and posted about the disappointment on Reddit.

WHY YOU CARE: This matters because AI tools are increasingly sold to everyday people as productivity helpers — write my email, build my app, plan my trip. When they work, they save hours. When they fail like this, they can quietly waste your afternoon while you wait for output that never arrives. If you've ever paid for an AI subscription expecting magic, this is the flip side: the tool can look confident and produce nothing useful at all.

THE TECHNICAL PART, SIMPLIFIED: AI models generate text one word at a time, picking what seems most likely to come next. Sometimes they get trapped choosing the same word pattern again and again, with no built-in signal telling them to stop. Smaller or newer models — this one has 27 billion settings, which is mid-sized by today's standards — are more prone to this kind of "looping." Some systems have guardrails to catch it, but clearly this one didn't.

THE HONEST CAVEAT: This is one person's story on a public forum, not a lab test. Other users may have better luck, settings may differ, and the model might work fine for other tasks. Still, it's a useful reality check: don't hand a new AI your whole afternoon without a backup plan, and always skim its output instead of trusting it blindly.