NASA’s new dark energy space telescope can also detect killer asteroids
technologyreview.com·1h ago
AI Research
arXiv CS.AI·28d ago·5 min read
Beyond Accuracy: How Humans Evaluate Legally Correct but Socially Controversial Legal Advice from Machines
TL;DR
Humans often face a dilemma when evaluating legal advice generated by machines that is legally correct but socially controversial. This study explores how individuals assess such advice, focusing on the balance between legal accuracy and social acceptability.
✦ Why It Matters
Engineers developing legal AI systems should incorporate social context analysis to improve user acceptance and trust.
Key Takeaways
- Evaluate how AI-generated advice is perceived by users.
- Incorporate reasoning to improve the acceptance of AI advice.
- Understand the balance between objectivity and context in legal advice.
- Design AI systems that address public concerns about automation.
Full Summary
As AI systems increasingly provide legal advice, understanding public perception is crucial, especially when the advice is legally correct but socially controversial. A survey experiment with 3,348 adults in mainland China examined how legal advice attributed to either an AI or a human lawyer was evaluated.
Surprisingly, attribution to AI did not negatively impact perceived reasonableness. Instead, AI advice was seen as more objective, which improved its perceived reasonableness, but it was also viewed as less comprehensive, which had the opposite effect.
Importantly, providing reasoning alongside the advice significantly boosted perceived reasonableness for both sources. These findings suggest that public responses to AI legal advisors are influenced by a balance of expectations regarding objectivity and contextual sensitivity, challenging the notion of algorithm aversion.
The results have implications for designing AI recommendation systems in sensitive areas like law.