Counterspeech for Mitigating the Influence of Media Bias: Comparing Human and LLM-Generated Responses