Elco Van B.
Posted
Lying AI
The AI works well for projects of roughly 1,000 lines of code or less. Beyond that point it tends to hallucinate or even tell lies: it claims to have made changes it hasn’t while it chews through tokens fast, and even rewrites or deletes stable functions when you only ask it to change something minor—like a logo or icon. Even when enabling the 'dynamic reasoning'. In its current form it’s really unusable for building production-quality apps. The team behind it is friendly (they once refunded my tokens), but that doesn’t fix the AI’s inaccuracies and erratic behavior.



