That's a fair point. But models like GPT4 do not hallucinate much when it comes to summarizing. So I don't think these applications contribute to anything negative.