#generativeML

Donald Hoberndhobern@scicomm.xyz
2025-05-31

@adamsteer

A lot of work on #FAIR has been rather cosmetic and focused on the easy F and A parts. For universities, it can often be a box-ticking exercise. As the excellent #WorldFAIR outputs indicate, it's often been about placing datasets on shelves rather than making the data reusable and accessible.

But I would push back and argue that, when #FAIR is used as a lens to see data from the standpoint of future (re-)use, it is a massive leap forward. Of course, it's then really a call to adopt robust data engineering at all stages in the data pipeline.

This is even more important in the age of #generativeML. Proper documentation of data provenance, licensing, transformations, structure and semantics is essential if we are going to keep track of what ostensible data comes from actual sensors or human observers or has been generated in predictable well-understood ways from such observations.

The CARE principles are also really important as they go into the ethical considerations for how we collect, manage, use and share data.

My only hesitation is not with the CARE principles or with the wording or with the fact that is immensely important for us to decolonialise our approach to information gathering. It's with the fact that they've become a reason to compartmentalise management of Indigenous knowledge and data and to treat the ethics of using those data as a narrowly defined issue. Focus on Indigenous data will help those already inclined to consider these issues to do so, but it risks making an exclusionary culture-war-adjacent issue.

Without in any way wanting to reduce our focus on the excellent reasons for #GIDA and others to foreground these principles, the challenges they address are the same ones that we all face in a capitalist and authoritarian #panopticon.

The CARE principles are part of what should be a much broader rallying cry for consent in data management everywhere. Of course, my suggesting this may just contribute to devaluing the proper concerns of Indigenous communities in this area, but I can only really see foresee the necessary IT underpinnings and practices getting mainstream adoption if they are seen as a central issue.

Donald Hoberndhobern@scicomm.xyz
2025-04-28

So, is #DuckDuckGo now using AI to rewrite preview text for #Wikipedia pages?

If I search e.g. for "Bob Dylan", the preview for WIkipedia on #Google and #Bing is the start of the Wikipedia page.

DDG has replaced it with generic "Learn about ..." text that does not appear in the HTML. The same happens with other searches for prominent people. I haven't explored further.

#GenerativeML #enshittification

Item from DuckDuckGo search for Bob Dylan:

Wikipedia 
https://en.wikipedia.org > wiki > Bob_Dylan
Bob Dylan - Wikipedia 
Learn about the life and career of Bob Dylan, one of the greatest songwriters of all time and a Nobel laureate. Explore his musical influences, genres, albums, awards, books, and controversies.
Donald Hoberndhobern@scicomm.xyz
2024-07-09

Another excellent piece from #EdZitron on #generativeML hype, based on a #GoldmanSachs report:

wheresyoured.at/pop-culture/

"One of the fundamental misunderstandings of the bosses replacing these workers with generative AI is that you are not just asking for a thing, but outsourcing the risk and responsibility. When I hire an artist to make a logo, my expectation is that they'll listen to me, then add their own flair, then we'll go back and forth with drafts until we have something I like."

Andrei Kucharavyandrei_chiffa
2024-04-22

This Friday (26.04.2024), I will be giving a keynote on the the evolving threats landscape in the age of Generative ML.

Looking forwards to the feedback of the cyber-sec community on the work we have been doing with @cydcampus for the last 4 years; and looking forwards to meeting you if you are attending!

Announcement of keynote speaker for Insomni'hacks, a Swiss Cybersecurity conference. The keynote is scheduled for Friday 26 April 2024 9:20-10:10, will be held at the Swisstech convention center, Lausanne. The title is "Threats and Mitigations Landscape in the Age of Generative AI".
Donald Hoberndhobern@scicomm.xyz
2024-03-04

@Snoro

What fresh hell is this? This photo seems to be of a completely made-up bird. At very least, they're not #tropicbirds. Even young ones have much stouter bills than these, the colouration is all wrong, and the crests are not appropriate. If you look at their feet, they connect oddly with the two twigs. This seems clearly (and needlessly) to be an AI-generated photo.

The text is dubious as well. In what universe would you call them "apex predators"?

#fake #generativeML #aaaargh

Andrei Kucharavyandrei_chiffa
2024-01-31

ThioJoe is a pretty good cybersecurity youtube channel for non-professionals.

Out of top 10 new 2024 scams, 2 has been made possible by Generative ML, warranting him to issue a special warning at the end that things were going to get worse.

I agree with him - things are likely to get much, much worse before they get better.

youtu.be/JKNtt_nPmnU?si=wWvAOb

Humans take a stand in the early robot wars. https://www.technologyreview.com/2023/10/23/1082189/data-poisoning-artists-fight-generative-ai/ #generativeml #ml #machinelearning

https://social.vaughnhannon.com/2909-2/

Client Info

Server: https://mastodon.social
Version: 2025.04
Repository: https://github.com/cyevgeniy/lmst