Having massively over-promised on an immature technology with really quite a limited range of productive use-cases, #AI firms are now contriving to deliver.
However, because there is no tangible product, they are using their financial might and institutional power to bludgeon the Internet and society as a whole into an AI-shaped hole.
It's like a company that makes anti-diarrhoea meds pumping laxatives into the water-supply to engineer enough demand to justify billion of investment.
Interesting data from a new edition of the Foundation Model Transaprency Index - collected six months after the initial index was released.
Overall, there's big improvement, with average score jumping from 37 to 58 point (out of a 100). That's a lot!
The interesting fact is that researchers contacted developers and solicited data - interactions count.
More importantly, there is little improvement, and little overall transparency in a category that researchers describe as "upstream": on data, labour and compute that goes into training. And "data access" gets the lowest score of all the parameters.
Can Generative #AI improve social science? https://www.pnas.org/doi/10.1073/pnas.2314021121?TOC_v121_i21 That's a reasonable discussion of the advantages and risks, including limited reproducibility of using LLMs.
An open AI infrastructure for LLMs seems reasonable. Maybe funding agencies can support first steps in this direction @academicchatter