Stephen Downes

Knowledge, Learning, Community

I've had a mostly inactive account at iris.ai for a couple of years now (it has recently been warning me it will be deleted if I don't do anything) and used it to access a decentralized data architecture service so I have a sense of the capability the AI company is bringing to this project. The focus of this particular project is 'the AI Chemist', which is a nice hook, but the real story is Iris's access to the CORE library: a metadata dataset (title, author, abstract, publishing year, etc.) for some 210 million articles, and a full text dataset of 29.5 million articles. The AI Chemist, meanwhile, was actually introduced back in June, 2020 (something the Jisc article doesn't mention) when Iris had about 18 million articles. So the big change here is the introduction of all that metadata from the millions of closed-access papers. That doesn;t really seem to me to be a great leap forward.

Today: 1057 Total: 1065 [Direct link] [Share]


Stephen Downes Stephen Downes, Casselman, Canada
stephen@downes.ca

Copyright 2024
Last Updated: Mar 29, 2024 03:49 a.m.

Canadian Flag Creative Commons License.

Force:yes