the programming languages of the future are those that are most robust against agent hallucination, and provide detailed, structured feedback to the LLM for it to continue after compilation failures. they must *also be* easy for humans to read, since validating what an LLM has done is also crucial!