The Representation Is Not the Scholarship
The problem with AI may not be that machines are growing more intelligent. It may be that we are becoming increasingly skilled at amplifying a highly capable, but limited form of cognition.
The problem with AI may not be that machines are growing more intelligent. It may be that we are becoming increasingly skilled at amplifying a highly capable, but limited form of cognition.
Building robust citation and attribution into generative AI systems are foundational to usage, credit and trust. We need to expect more from AI.
Faced with technological shifts not seen since the advent of the internet, Todd Toler and Angela Cochran posit that the biggest challenges for organizations building an AI strategy are human, not technology.
As AI systems increasingly reason from the scientific literature, the integrity signals that make research trustworthy — open data, structured metadata, robust retraction processes — matter more than ever. PLOS CEO Alison Mudditt on why open access publishers have a different set of obligations in an AI world.
Current AI disclosure guidelines are failing and driving AI use underground rather than making it transparent. In this follow-up post, I turn to the more challenging question: what publishers should do about it.
Robert Harington talks to Matt Kissner, CEO of Wiley, in this series of perspectives from some of Publishing’s leaders across the non-profit and for-profit sectors of our industry.
Todd Carpenter describes the new 2029 STM Trends report, which provides a vision and a bridge to the future for the community.
Model licenses simplified library licenses in the 2000s. The same approach can streamline licensing scholarly content for AI training today.
A compilation of links and a video to incisive analyses of ChatGPT and what it means for the future.
On July 4, 1971 Michael Hart posted the first ebook file on the ARPANET and transformed content distribution.