Video:
16:36:34 From Fabien Benetou : random work on graphs https://twitter.com/search?q=%40utopiah%20graph&src=typed_query
16:37:39 From Fabien Benetou : 3D from wikidata https://twitter.com/utopiah/status/1106238075095654402
16:52:59 From Mark Anderson : !! more tl;dr markers
16:59:29 From Fabien Benetou : https://huggingface.co/spaces/benthecoder/news-summarizer
16:59:45 From Fabien Benetou : summarizer
17:03:18 From Mark Anderson : !! Good examples, absolutely!
17:12:13 From Fabien Benetou : Not sure if that’s been discussed before but system.com could be interesting. Could also invite Lisa Strausfeld to discuss the design decisions behind it.
17:13:23 From Fabien Benetou : (knowing there is also documentation on specific topics before, e.g https://docs.system.com/system/how-system-works/relationships-methodology to go back briefly on category as a kind of relationships)
17:18:15 From Mark Anderson : fasrai book https://docs.fast.ai
17:19:05 From Mark Anderson : !! ‘fastai’ book: https://docs.fast.ai
17:19:07 From Frode Hegland : https://www.amazon.co.uk/New-Media-Reader-MIT-Press/dp/0262232278/ref=sr_1_1?keywords=the+new+media+reader&qid=1663172237&sr=8-1
17:19:17 From Frode Hegland : The New Media Reader the second book
17:19:24 From alanlaidlaw : https://colab.research.google.com/github/fastai/fastbook/blob/master/01_intro.ipynb#scrollTo=4ChNJBLVhRns
17:19:26 From Mark Anderson : ^^^ come with CD 🙂
17:20:05 From alanlaidlaw : https://www.youtube.com/watch?v=8SF_h3xF3cE&t=2894s
17:21:47 From Frode Hegland : is a researcher in machine learning.¶
He work aims to apply the state-of-art in open-source intelligence and reliable interfaces for AI deployment. Exploring robustness of vision models led me to work on a novel tool for anticipating AI model’s failures and invent new investigative workflows. Co-leading the Interactive Reading Group at Cohere for AI.
Supervised by Joseph Gonzalez, Trevor Darrell, Lisa Dunlap, and Neerja Thakkar at Berkeley AI Research.
research https://twitter.com/ucbrise https://twitter.com/berkeley_ai https://twitter.com/forai_ml Person | Timeline: – | research interests: AI interpretability, ML + investigative journalism/OSINT + interfaces | From Kharkiv, Ukraine
17:23:41 From Fabien Benetou : http://www.newmediareader.com
17:25:35 From Mark Anderson : Links in print: ‘see more’. citations, …
17:30:25 From Frode Hegland : have to move inside for power. listening while moving
17:39:51 From Fabien Benetou : containerization and provenance hopefully address this challenge of preservation at least partly, it’s hard but IMHO shouldn’t slow us down
17:39:57 From alanlaidlaw : agreed. its a problem of the future. two points: the chestnut about firing linguists and improving the model. AI already making new episodes with “legacy” format of TV shows.
17:40:57 From Fabien Benetou : example of expertise in the domain https://www.cambridge.org/core/journals/art-libraries-journal/article/abs/preserving-an-emerging-digital-arts-landscape-digital-preservation-at-university-of-the-arts-london/37645F3059FC153199C8A6FFC95D3957
17:47:56 From alanlaidlaw : one thing I’d like to talk about VR in relation to technological revolutions. specifically how path dependencies paint the primary model of the new medium. Then, how the primary mental model often leads to an interface saddlepoint (after a period of high innovation) and that saddlepoint (or valley) persists for a long time. Ex: at some point, we settled on cars to work a certain way. then the focus shifted to deployment, policies, downstream effects on industry.
17:48:13 From alanlaidlaw : fabien: beautiful
17:51:23 From Mark Anderson : !! makes me think of the margin note feature when writing in Author (even if the notes are only within the doc).
17:57:29 From Mark Anderson : !! Interestingly the malleability in VR just described is partly constrained by suitable metadata.
17:58:11 From Mark Anderson : !! sorry, …is partly… (wrong emphasis.
17:58:37 From alanlaidlaw : agreed, Frode. but the VR wave is crashing simultaneously w the AI wave. the only thing similar would be (maybe) web if it happened at the same time as personal computers. (certainly overlapped). these waves seem much larger. I mention that bc it indicated that the future is far less predictable.
17:58:55 From alanlaidlaw : I’m not worried about the metaverse – as I understand it.
17:59:20 From Mark Anderson : wave is larger partly due to more of us being ‘connected’.
18:00:27 From Fabien Benetou : Meta is one of the 2 biggest ad companies worldwide, Google being the other
18:00:43 From Fabien Benetou : so 100% we’ll get “better” (read worst) ads in Vr and AR.
18:00:44 From alanlaidlaw : another topic: in praise of ebooks. bc they settled on a saddlepoint fairly on (impoverishing the potential), it has been preserved from the ruinous effects of advertising.
18:01:09 From alanlaidlaw : fairly early* on
18:01:15 From Mark Anderson : An interesting ‘skill’ I find myself acquiring is learning not not noticeably see adverts on-screen. Unless of course if full-screen which is where most read, on their phones.
18:01:44 From Mark Anderson : not not → to not
18:02:19 From Fabien Benetou : f*ck ads https://ublockorigin.com/
18:03:28 From Fabien Benetou : (single use device https://store.steampowered.com/steamdeck and Linux based so very tinkerable)
18:03:56 From alanlaidlaw : Advertising is the true Moloch. its all on a weird belief that clicks and pause and other behaviors actually leads to more purchases. its a strange religion