Skip Navigation

Posts
16
Comments
223
Joined
3 yr. ago

  • As you write up the alternatives (especially RSS and adblockers! These can be low barrier to entry), it would be great to link them from this part 1! I feel like the ask and text for part 1 is roughly the right size, but the action feels very big and scary to folks (and giving a reassurance early in the text that you can replace stuff and meet your needs might be helpful).

  • The topic is important, the timing is good (just in time for some new-years resolutions), and the writing is effective. Thank you for taking on the project.

    I had hoped that the first suggestion in part 1 would be more accessible than 'delete the accounts and create burner accounts' - we've chosen the most effective but biggest ask, and I don't think this post quite provides the infrastructure required for many people to make the change. FB is used by many folks as social media; the keeping track of friends, events, and family can't really be done from a burner account (your messages alone will identify you entirely to meta).

    And I have a personal pet peeve on this topic that's triggered by the last section: I believe that mindfulness is a good way to improve internet use, but I think we've proven as a society that most people can't implement this sort of self-reflection and intentionality without more structure. Where's the tooling to remove dark patterns, automatically ask these questions after an app use, etc. ?

  • Let's include the whole paragraph at least.

    Apertus was developed with due consideration to Swiss data protection laws, Swiss copyright laws, and the transparency obligations under the EU AI Act. Particular attention has been paid to data integrity and ethical standards: the training corpus builds only on data which is publicly available. It is filtered to respect machine-readable opt-out requests from websites, even retroactively, and to remove personal data, and other undesired content before training begins.

  • This is wonderful! Less of everything depending just on Microsoft goodwill please.

  • Fascinating. I think this is my first time seeing someone celebrate the death of stack overflow?

    The major downside to community walled garden documentation and discussion is that it is no longer searchable. I must learn to sort through 11 different UI for 11 different languages, if I can access them at all, and this information is much more likely to get lost piecemeal (while SO will disappear either never or all at once; and there is a single effort for preservation, instead of smaller independent ones).

    But I do like community building. And I am entirely confused by Joe.

  • I think you are right about what causes things to go bad irl, but I actually most dislike what these reactions do to media. And there I think emotional regulation would not be sufficient for better results. Disliking would need to not be engaging in the way it is.

  • I deeply hate the notion that hating things is cool, edgy, or more interesting than liking things.

  • Thank you for fleshing out your world model and theory. I think that this model falls short of a source (and contradict some other AI-pessimistic economics predictions; namely a crash in computing cost and in crypto), but could be developed into something I'd find compelling.

    Let me brainstorm aloud about what I think this world model predicts that we might have data on...

    Did we see a crash in ISP prices, home and industry internet use, domain hosting, or other computing services in the dotcom bubble? That situation seems extremely analagous; but my vibe was that several of these did not drop (ISP price I suspect was stable), and some of these saw a dip but stayed well above early-internet rates (domain hosting)? I feel like there'd be a good analogy here, but I'm struggling with a way to operationalize.

    I mentioned a use for compute that your reply didn't cover: crypto mining. Do we have evidence that the floor on crypto is well below datacenter operating costs (across exploitative coins as well)? I vaguely remember a headline in this direction. Another use case I don't see drying up: cheating on essay assignments.

    More broadly, this model predicts that all compute avenues are much lower payoff than datacenter operating costs. I think I'd need to see this checked against an exhaustive HPC application list. I know that weather forecasting uses up about as much compute as AI for some supercomputing clusters.

    Governments have already issued rather large grants to AI-driven academic projects. I suspect many of these are orders of magnitude larger than the size of academic AI 6 years ago. (I'll also quickly note that libraries are better than google search has ever been for finding true facts; yet google search has remained above library use throughout its existence.)

  • I know at least a few HPC folks talking about building their datacenters far north, where cooling can be done by opening a window.

  • ? Massive GPU server racks are relatively easy to repurpose for several things. The most likely (if sad) is crypto mining, but there's also expensive weather simulations, cloud gaming, video hosting, etc.

    Requesting a source that these centers are hard to repurpose. I find myself pretty skeptical. Computers are generally multipurpose and easy to swap tasks on.

  • I do feel obligated to at least expose folks to numeracy, even if they won't listen. Hope it's not giving them a repeated trauma somehow.

  • Which feels to me like terrible policies still. Make big projects pay their costs please!

  • Growing inefficient cattle crops in a desert to preserve water rights: not necessary.

    Flying Coast to coast for a business meeting that could be an email: not necessary.

    Manufacturing those cheap scissors that break after 2 uses: should be a crime (not necessary).

    All of these subcases have comparable emissions and externalities to the data centers (at least by my fermi estimates).

  • Not my experience? Search and community have both gotten worse though, and these were great for finding good stuff.

  • (I know it's not the point, but a reminder that data center climate impact, including heat, is nowhere near flight, agricultural waste, or construction. Hate it for its own reasons, not for fake ones.)

  • Mentioned in the article: this supports many african languages, including african sign languages. (There were plenty of apps that already did this for, say, ASL.)

    Relatedly: it's possible to be deaf and illiterate. Similar to being able to speak but not read, these folks are much better served by sign language translation over text. (I can't find data on rates quickly, but see here for discussion of the general phenomena. It's harder to teach the deaf to read, so easier for them to fall between the cracks.)

  • Though we'll need to focus on different damnation for both, as this OP is at best a misunderstanding. See top replies.

  • Reminder to look at top comments, ideally at a bit of a delay. See this reply.

  • Thanks!