Posts tagged “research”

• Why reading a long document is not the same as tracking every change

  Ai2’s StateBench separates finding facts from following updates. What should teams check before trusting AI to reconstruct changing work records?

• What Anthropic’s study means for human review of AI-assisted work

  Anthropic’s analyst study separates correct calculations from independent judgment, with practical review proposals and limits on what the findings show.

• Why an AI novelty score is not proof of an original research idea

  Ai2’s study shows how reviewer context and assumed labels shape AI novelty scores—and what researchers should check before acting on them.

• Google’s Patent-Lawyer Study: Measure AI Output and Human Judgment Separately

  Google’s patent-lawyer study separates AI-assisted drafting gains from unaided judgment. Here is how to evaluate output, efficiency and professional learning.

• OpenAI's 100-Plus Math Claim Needs a Review System as Scalable as the Search

  OpenAI says an internal model resolved over 100 open math problems. The hard part now is scaling proof review, attribution, disclosure, and control.

← Back to blog