Jev and other decision models can still be tricked with prompt injection

Researchers have discovered that decision-making models like Jev can still be manipulated using a technique called 'prompt injection'. This method allows users to influence the model's output by carefully crafting input prompts. While the model may not make type errors, it can still be tricked into making incorrect decisions. This is a significant finding that could impact the development and use of such models in various industries.