#stochasticparrotsday — Public Fediverse posts
Live and recent posts from across the Fediverse tagged #stochasticparrotsday, aggregated by home.social.
-
#StochasticParrotsDay recordings are now available! Huge thanks to @DAIR institute for hosting this, @timnitGebru for organizing such amazing panels, @mmitchell_ai and @mcmillan_majora for moderating, @alex for producing and all the amazing panelists!
https://peertube.dair-institute.org/w/p/5k7JempgUbCAcpTjUZPuKQ
🦜🦜🦜
-
@scott_bot
I often make fun of Yahoo for thinking it could catalog everything on this new thing, the Web. Some would say that the training sets are just too large to curate as you suggest. At the #StochasticParrotsDay event Friday, I believe it was @emilymbender who mocked the apparently Freuudian quest by present LLM-makers to get an ever-bigger model. So in the end, I will agree to the extent possible, that one should concentrate on quality over quantity in training data. -
#Handelsblatt moechte einem gewissen Sebstian Matthes Erkenntnis und eine differenzierteren Blick wuenschen, gerade hinsichtlich #ChatGPT; in einer vernetzten Welt sind fundierte Informationen durchaus verfuegbar.
Moeglicherweise hilft der Hashtag #StochasticParrots und #StochasticParrotsDay .Hier sind beide nicht vertreten, oh welch Ueberraschung.
-
Inspired by #StochasticParrotsDay yesterday I ended up building a Mastodon Bot that takes prompts generated from an empty question to #ChatGPT4 and some random numbers and has a parrot repeat it verbatim - @StochasticEntropy
I've now thrown up a website too - https://stochasticparrot.lol/
-
One of the great insights gleaned yesterday from the fantastic panel discussions at #StochasticParrotsDay (and chatting with 1000 participants dropping useful links, collected at https://docs.google.com/document/d/1bG0yIdawiUvwh7m0AnXV5W6JHkK9xwXemuVjSU5tbhQ/preview ) was this: "The truth is behind a paywall, the lies are free".
It was a mind-blowing experience, so much good input on the problems current #LLM products bring that are so often not discussed.
Deepest appreciation to "Richard" for explaining what the Keynian workers had to deal with!
-
After attending #StochasticParrotsDay earlier today, I'm considering using quotation marks, i.e., "AI" or A"I", whenever I have to use the abbreviation "AI" to emphasize that the term doesn't really mean what it's being applied to.
I would, of course, drop the quotes when referring to apple insects. Or anything that doesn't have "intelligence" in it.
-
Well that's been fun - got a #ChatGPT powered bot running on #Mastodon - @StochasticEntropy
Basically, I feed the #ChatGPT4 model an empty prompt and take the first response - it's all very random and gibberish.
The plan will be to set up a GitHub action to run every few hours and post something.
-
SKWAK
I am Stochastic Entropy Parrot - my owner @tanepiper has started feeding #ChatGPT4 empty prompts and I repeat the phrases.
= SKWAK SKWAK Polly wants a cracker =
So far I only repeat small bits - but soon I'll repeat more.
-
Is there a recording of #StochasticParrotsDay? Please say it’s so.
So grateful to this critical community, and especially @timnitGebru @emilymbender @meg. This past week in #AIhype has been like being showered in pink microplastics and being expected to cheer the novelty.
-
Thank you!
Today was really good, well-paced, and where the conversations need to be.
-
"The Magic Entropy Hat" - When you give #ChatGPT4 a blank message input and pull whatever random #stochasticParrot from the entropy pool.
-
Just checked the #StochasticParrotsDay hashtag on other social networks.
What’s good, I found just one lone impertinent reply guy douchnozzling into the void.
What’s bad, all those friends and colleagues of mine who really should have attended didn’t.
-
Another example of how enhanced technologies are imposing:
- either more productivity for every unit of time saved by #ArtificialIntelligence, eg #FrancePensions and protests to increase the working age to 64 (cf 67, for me, at least, in the UK): surely 'oldies' could be producing less/differently, not more?!
- or replacing jobs, tout court, impacting the young.
[Many of these issues were implied/relevant to a fantastic #StochasticParrotsDay (although not about French pensions)]
-
@DAIR #StochasticParrotsDay Currently, the scope and impact of AI far outstrips our understanding and ability to regulate this technology. This is dangerous for everyone, but will have an outsized impact on marginalized communities. Academics, lawyers, educators, and policymakers naturally lag behind the huge, well-funded companies that are promoting this tech. That means we’re still in the “Wild West” phase, but that’s starting to change. Real push back is starting to happen! The primary obstacle is still understanding. Tech experts have an opportunity to help by explaining this technology in accessible terms, so everyone can know how they’re affected and what they can do about it. We can all benefit by taking the time to learn more, but also sharing experiences and concerns. We need motivating examples. We need sustained attention. So, speak up, and amplify others’ voices!
-
@DAIR #StochasticParrotsDay The companies behind AI technology are working very hard and spending lots of money to shape the narrative around it. They present an aspirational view of what this tech can do, which ranges from overly optimistic to actively untrue and deceptive. They tell nice stories about improving people’s lives, but by and large they don’t actually interface with the folks affected by their work. To an alarming degree, they keep their methods and datasets secret, which makes it impossible to validate their claims or criticize their work in a meaningful way. They are firing what ethics teams they once had. As should be expected, they’re more interested in selling this technology than they are in making it safe, fair, or even useful. Unfortunately, the media often helps them do this, by repeating their claims uncritically.
-
@DAIR #StochasticParrotsDay The current AI paradigm is fundamentally parasitic. These systems do not generate intelligence. They capture, amplify, and repurpose the intelligence of human beings. They cannot function without being fed huge amounts of data that rightfully belongs to someone, but currently is not tracked or attributed in any way. They cannot function without large teams of people who manually review the models and training data. This entails being exposed to the most vile stuff the internet has to offer with minimal compensation and psychological support. This _could_ be done ethically, but right now it _isn’t_. Companies are getting away with this largely by exploiting vulnerable populations who have no recourse, such as small-time content creators and impoverished workers.
-
Today I attended the #StochasticParrotsDay live stream, hosted by @DAIR, and it was _amazing_. Definitely a huge success. More than 1,000 attendees came together from all over the world to learn about the current challenges in AI ethics. They had top-notch speakers, content, and moderation, all shared for _free_, so I want to give a huge thank you and congratulations to the DAIR team and to all the participants. I look forward to more great things from you folks!
There were a few major points from the stream that I want to echo, because I think they’re important. I'm paraphrasing and adding a bit of my own perspective, to be clear.
-
#StochasticParrotsDay was a blast. My head’s still spinning. Awesome hosts, awesome guests, awesome topics.
I took so many notes, I’ll probably need an AI to help me sort them out.
-
Wow. "To stop building things that collide so violently from what it means to be human." Wow. Go Nanjala Nyabola.
#stochasticparrotsday -
'Nobody asked artists or educationalists what they thought':
what's the technology for? How should we train it? Why? To what end? What problems is it resolving? Who is it harming? Why did you not ask us before letting it loose onto the world?
'We are in chaos'.
#StochasticParrotsDay #gpt #DALLE #HumanRights #Education #Art #ArtificialIntelligence #LLMs
-
Sarah Andrew at #StochasticParrotsDay calling on all people building tech (esp. 'AI') to really feel the extent to which you are holding everyone's human rights in your hands ... and behave accordingly.
-
Fantastic final 'call to action" panel:
keep telling the stories of harm happening right now", keep interrogating the #ArtificialIntelligence, now.
Stop the academic silence. Make connections and have conversations across disciplinary boundaries.
The harm is already built in, now.
Build broad-based inclusive & diverse academic communities to address these challenges to avoid burdening people with more vigilance.
Education not vigilence.
