Post AbTeHD8To1LfH0Npke by webology@mastodon.social
(DIR) More posts by webology@mastodon.social
(DIR) Post #AbTaCpKpxFSuTvokWe by simon@fedi.simonwillison.net
2023-11-04T22:00:55Z
0 likes, 0 repeats
The US Copyright Office asked for public comments on the issue of generative AI training data and got ~9700 comments: https://www.regulations.gov/docket/COLC-2023-0006/commentsMost of them are seeverwl paragraphs long. Anyone with government experience know what the process looks like for analyzing them?Assuming you get staffers to read all 9,700, what do they do next? Do they prepare some kind of summary document for decision makers to read?
(DIR) Post #AbTakBEdYHH7xYozyK by hq1@fosstodon.org
2023-11-04T22:06:58Z
0 likes, 0 repeats
@simon will they use LLM to generate a summary?
(DIR) Post #AbTc35tVjUOEqsWT4q by jiejie@mastodon.social
2023-11-04T22:21:31Z
0 likes, 0 repeats
@simon There is a FAQ that links to a document about comments and the rule-making process.https://www.regulations.gov/faq“Yes. Public participation matters. Democratic, legal, and management principles justify why public comments make a difference in regulatory policy. Public participation is an essential function of good governance. Participation enhances the quality of law and its realization through regulations (e.g. Rules).”Also, there’s an API. :Dhttps://www.reginfo.gov/public/reginfo/Regmap/REG_MAP_2020.pdf
(DIR) Post #AbTcF7Y1PYSdSGCq9Y by MattiSG@maly.io
2023-11-04T22:22:31Z
0 likes, 0 repeats
@simon I’ve seen this sort of consultation unfold in European governments. Interns or low level staff were used to classify comments into buckets, which were summarized by more senior staff and quantified.
(DIR) Post #AbTcF9kVE7dIHY3psu by MattiSG@maly.io
2023-11-04T22:22:56Z
0 likes, 0 repeats
@simon Another, much larger example, could be France’s “Grand débat national”, where automated summarization and classification was used. The French Parliament even hosted a dedicated hackathon: https://www2.assemblee-nationale.fr/15/evenements/2019/restitution-d-analyses-des-donnees-du-grand-debat-national/un-hackathon-inedit
(DIR) Post #AbTcR6QxNXHrTVcBnM by simon@fedi.simonwillison.net
2023-11-04T22:23:56Z
0 likes, 0 repeats
@jiejie I want to hear the behind-the-scenes stories from the actual staffers - what do they do with this stuff? It's a huge amount of content!
(DIR) Post #AbTeHD8To1LfH0Npke by webology@mastodon.social
2023-11-04T22:46:35Z
0 likes, 0 repeats
@simon I wonder what % were generated or submitted by AI?
(DIR) Post #AbTkPDyBBldvtvlnHc by JMMaok@mastodon.online
2023-11-04T23:55:04Z
0 likes, 0 repeats
@simon 9700 doesn’t actually seem like a large number after what the government dealt with for net neutrality. One of several articles about the flood of bot comments: https://www.wired.com/story/bots-form-letters-humans-fcc-net-neutrality-comments/
(DIR) Post #AbTwAYfDAJoDBlfeBk by jasonkoebler@mastodon.social
2023-11-05T02:06:55Z
0 likes, 0 repeats
@simon the sad answer is that public comments are not really considered. I spoke with a few experts about this during the initial net neutrality battle and the comments that are considered are the ones written by corporations by lawyers and lobbyists: https://www.vice.com/en/article/ev5b3k/net-neutrality-fcc-zero-consumer-comments-citedhttps://www.vice.com/en/article/qkvjy7/proof-the-fcc-will-ignore-your-net-neutrality-comment-and-listen-to-isps-instead
(DIR) Post #AbUWaYXbqAtwggoLei by kitten_tech@fosstodon.org
2023-11-05T08:55:10Z
0 likes, 0 repeats
@simon I've seen user research summaries in the UK gov; they tend to have a bunch of statistics ("56% of responses expressed support for the proposed change", "22% cited fear of giant spiders as a concern", etc) and pulling out a few representative quotes ("I can see that replacing teachers with giant spiders would free up school budgets to spend on equipment and could help with classroom discipline, but...")