Recent searches
Search options
"As an ethically acting organisation, we should no longer promote commercial Gen AI or try to integrate
it into our products."
Quote from my slides about the ethics of Generative AI:
https://blog.johanneslink.net/2024/07/18/ethics-of-generative-ai/
BTW, the whole presentation was heavily influenced by recent talks from @tante
@jlink This is really, really, really, really, really good. Short , to the point, well-structured. All the arguments hit the topic on the nail. Everyone in every company should read this.
Since I agree and applaud 99% of this, I'll highlight the one thing that didn't click for me: I don't see the link between "ok for individual => exploitation at corporation level" (true!) and the tragedy of the commons (which for me is sth that can apply with just individuals, no corps).
Anyway, bravo!
@jlink also, I should highlight that this wasn't 100% things I knew. Thanks to this I learned that the common intuition that training is the most energy-intensive task is wrong. This came as a big surprise and changes things a lot!
@PierricD This holds for the popular models. It’s probably not true for the specialised ones that see much lower usage.
@jlink Good stuff, especially leaning on the ethical organization! I wonder how this will have to change once LLM is deeply integrated and unavoidable in desktop environments. i.e., will it be a cost that orgs will just take and open up a slippery slope from there? ("we already use it, can't be that bad actually")
@czr "Ethics are never efficient"
@jlink I would add that no matter whether it is legally intellectual property theft, it is harming communities of small artists everywhere
since now people are afraid to be taken advantage of when they share their work freely, building communities is harder and those that are already built button down the hatches
it's like when NFTs became popular, suddenly there were people who wanted to 'harvest' from these communities, just now it's automated
@mmby Very good point!
@jlink
Do you know Atlas of AI by @Katecrawford
https://katecrawford.net/atlas
It looks on multiple levels at the destructive effect of #ai
Highly recommended
What I also find puzzling is, that a lot people criticising #AI (an related technologies) are not CIS (white) men(*)
- @timnitGebru &
- @emilymbender (On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?
- Cathy O'Neil (Weapons of Math Destruction)
- @Katecrawford (Atlas of AI)
- Sara Wachter-Boettcher (Technically Wrong: Sexist Apps, Biased Algorithms, and Other Threats of Toxic Tech)
- Meredith Broussard (Artificial Unintelligence: How Computers Misunderstand the World)
- Shoshana Zuboff (The Age of Surveillance Capitalism: The Fight for a Human Future at the New Frontier of Power)
(*) Persons present excluded
@realn2s @timnitGebru @emilymbender
Maybe it's not so surprising given that "AI" is a conservative and privileges-enforcing technology.
@jlink Thanks, very useful presentation!
@jlink could you please share the link to version you presented yesterday at JUG Darmstadt?
@pedjak https://johanneslink.net/downloads/GenAI-Ethics-JUGDA.pdf
It should also be somewhere on the JUG DA website, but I couldn’t find it either.