Notices by Paul Kedrosky (paul@social.paulkedrosky.com)

Fouling the nest, as visible from space:

Satellite Image Confirms Massive "Fast Fashion" Clothes Pile in Chile's Atacama Desert https://www.skyfi.com/blog/skyfis-confirms-massive-clothes-pile-in-chile

One of the more common arguments you get from people who imagine themselves broad-mindedly neutral about AI is, Won't it just make current tools a little better—better word processors, spreadsheets, etc., while adding some new tools to the mix, like for creating images that we can use in the preceding tools?

That is a remarkable failure of imagination, like the people early in the PC revolution who thought PCs would never catch on because people weren't likely to all become software engineers writing code. People get trapped in thinking about paving existing cowpaths, rather than re-imagining the existence of such paths altogether.

Heretical idea: There is, however briefly, an arbitrage to do on the hellsite, where you swallow hard and ignore the technorati, and, if a quality source, sign up and exploit the better reach of Twitter Blue, if for no other reason than most quality sources aren't doing that.

So, am I doing that? No. But ... arbitrages are arbitrages.

Seems so obvious that companies should revert to their own domains for authoritative information issuance. Spin up their own Mastodon instance, use RSS, whatever. But relying on a third-party domain for authoritative information under their name was always a material risk, and now there is no turning back given the ongoing authentication debacle.

Even hardcore Twitter types noticing the decline in timeline activity there. Social networks, like all social activities, can flip rapidly from FOMO (fear of missing out) to FOBS (fear of being seen). Twitter is very close to doing that FOMO→FOBS engagement flip.

Escaping the Malthusian trap: We are so trapped in recency bias that we have no idea how anomalous our current experience is, in historical terms.

Source:
• Escaping the Malthusian Trap - kieranhealy.org https://kieranhealy.org/blog/archives/2023/01/08/escaping-the-malthusian-trap/

Spoke to some students today, and it's depressing seeing them discover that there is zero upside to not passing off CGPT work as your own. Idealism washed away by algorithmic text generation. We are creating a generation of stone-cold sociopaths who must cheat to survive.

More seriously, if you have a material amount of training-ready audio, video, or text in the public domain, this is, sadly, something that you should consider. And it still may not help much, given how many ways there are to route around such non-scaling protections.

That this is not a joke and is probably a good idea, and it's only late March, makes me worry about, you know, early April.
---
RT @nearcyan
it may be useful to establish a "proof of humanity" word, which your trusted contacts can ask you for, in case they get a strange and urgent voice or video call from you

this can help assure them they are actually speaking with you, and not a deepfaked/deepcloned version of you
https://twitter.com/nearcyan/status/1640447061035307008

I say this often, but, with generative AI, scamming has found scalable, product-market fit—and people and companies haven't even begun to realize the consequences. Banks, brokers, insurance cos, etc. should be all over it, and they're way, way behind. Losses and lawsuits ahead.

This is fascinating, and more than slightly ghoulish/barbarous:

AI and the American Smile
https://medium.com/@socialcreature/ai-and-the-american-smile-76d23a0fbfaf

Terrific episode of @BBCInOurTime:

In Our Time, Paul Erdős #xp https://www.bbc.co.uk/programmes/m001jc68

If this sort of thing interests you, well … three times a week I highlight three intriguing graphs, with my analysis. You can subscribe here. Or not. Your call, honestly. https://kedrosky.org/subscribe #xp

I am in the camp of "rates stay higher for longer", and inflation is bouncing back, so this Bridgewater graphic (discussed in my mailer tomorrow) reinforces that take. #xp

One day we may look back at this legalistic footnote in Microsoft's 10-Q—our work in AI could "cause harm to ... society"—and wonder at a blithe, tech-deterministic ethos that put quarterly earnings growth ahead of the risk of blowing up the world. https://microsoft.gcs-web.com/node/31271/html #xp

My 'hood back in the home country is turning in some impressive low-temperature records today, with Mount Washington onto the list with an altitude cheat code. Kudos to all. #xp