Appreciate this review!--especially for the tools at the end that are more purpose -built for research.
A word of warning: Perplexity and Copilot both will frequently give citations for sites that have nothing to do with the claims that they have made. I have had Perplexity and Copilot both hallucinate/fabricate and then give references.
This is so helpful--thank you for offering it! I haven't used ChatPDF or Sci Space yet so I was especially interested in your experience with those. I would love to know what your assessment of the accuracy of the summaries is. I haven't tested extensively recently, but I've been warning students about inaccurate summaries and have seen examples in Elicit where the one-sentence summary did not capture the main finding of the paper but focused on background information.
Thanks, Anna! Your work on critical AI has been so helpful to me, so I'm thrilled you got something out of my post. The table feature in Elicit definitely had some issues on accuracy as you described. It seems have more potential right now than it does utility. For me, Scite offered the best summaries, but I haven't used any of them enough to give a final take. Most exciting to me about all of them was how accessible they make the primary texts.
Appreciate this review!--especially for the tools at the end that are more purpose -built for research.
A word of warning: Perplexity and Copilot both will frequently give citations for sites that have nothing to do with the claims that they have made. I have had Perplexity and Copilot both hallucinate/fabricate and then give references.
This is so helpful--thank you for offering it! I haven't used ChatPDF or Sci Space yet so I was especially interested in your experience with those. I would love to know what your assessment of the accuracy of the summaries is. I haven't tested extensively recently, but I've been warning students about inaccurate summaries and have seen examples in Elicit where the one-sentence summary did not capture the main finding of the paper but focused on background information.
Thanks, Anna! Your work on critical AI has been so helpful to me, so I'm thrilled you got something out of my post. The table feature in Elicit definitely had some issues on accuracy as you described. It seems have more potential right now than it does utility. For me, Scite offered the best summaries, but I haven't used any of them enough to give a final take. Most exciting to me about all of them was how accessible they make the primary texts.
Hi Eric, thanks for your kind words! Good to know that you did see some of the same problems with the summaries. I need to try out Scite!
As we reimagine education with AI, we have to ask better questions.
Gratitude to @Eric Hudson as he offers a compass to navigate these uncharted waters.
Every community is based on trust. What do we value more in any relationship?
So, how do students - and we are all students - learn in an AI world?
How do we foster cultures of integrity, transparency, and responsibility?
How do we learn how to learn?
What do we really value?
Really well said, Kevin.