Facilitating Human-LLM Collaboration through Factuality Scores and Source Attributions