Abbe @Abbe@qoto.org

"It might seem that the obvious course is not to make multiple models but rather to grow a network. Instead of developing two networks for recognizing cats and horses respectively, for instance, it might appear easier to teach the cat-savvy network to also recognize horses. This approach, however, forces AI designers to confront one of the main issues in lifelong learning, a phenomenon known as catastrophic forgetting. A network trained to recognize cats will develop a set of weights across its artificial neurons that are specific to that task. If it is then asked to start identifying horses, it will start readjusting the weights to make it more accurate for horses. The model will no longer contain the right weights for cats, causing it to essentially forget what a cat looks like. “The memory is in the weights. When you train it with new information, you write on the same weights,” says Siegelmann"

https://www.nature.com/articles/d41586-022-01962-y

Fountain in Italy, on sunset gets highlighted by sun, looks like pouring fire

What if extraterrestrial civilizations exist, but they are not so utopian as some SF writers are trying to sell them to us, but they struggle as we, humans on Earth do. And if they finally reach the technological level barely allowing them the interstellar travel and communication with sub-light speed, the civilizations crumble because of social inequality and environmental damage.

Tiny rant: US-/North America-centrism is so pervasive on the English-language Internet. People are quick to describe a plant or animal in a post as "invasive" without knowing where the photo was taken. I'm like, *where do you think it is native to?* Mars? ⁦ಠ⁠_⁠__ಠ⁩

I just saw a user assuming a planarian they found was invasive because that's what North Americans automatically parrot. They were in its native range.

The mighty @pluralistic@mamot.fr has a theory of inflation that works for late-stage capitalism.
Pluralistic: Excuseflation (11 Mar 2023) – Pluralistic: Daily links from Cory Doctorow
https://pluralistic.net/2023/03/11/price-over-volume/

@atomicpoet

Czy to jest coś czego należy się obawiać? Pierwsza próba kiedy BigTech próbuje położyć rękę na tym projekcie i później go ubić tak jak kiedy google sforkowało diasporę po to aby po przejęciu uższykodników zrobić to co google robi ze wszystkimi fajnymi projektami?

Czy może to tylko zwykłe objawy mojej paranoi? :D

Łotwa od 2023 roku zaczęła konfiskować samochody pijanych kierowców. Państwowe parkingi zaczęły się przepełniać, dlatego postanowiono, by wysłać pojazdy do ukraińskiego wojska i tamtejszych szpitali. Pierwszy transport wyjechał w środę.
https://tvn24.pl/biznes/ze-swiata/wojna-w-ukrainie-lotwa-wysyla-skonfiskowane-auta-pijanych-kierowcow-na-ukraine-pierwszy-transport-6812855

Genialne! Brawo!👍

@Radical_EgoCom
Child labor was never fully abolished. Exemptions, particularly in agriculture, have allowed the continued legal exploitation of children, incl long hours and dangerous conditions

Remember in 2008 when millions were losing their houses and millionaires and billionaires swooped in to buy up all the foreclosures at bargain basement prices? Same thing is going to happen with all of the start ups that are going to go under because of SVB. All the intellectual property will be purchased and exploited by fuckers like Elon Musk, Peter Theil and the whole parade of sociopaths.

Look everyone! Squirrel!

If you access corporate email on a personal device that can be unlocked with FaceID, you must change your face at least once every sixty days.

You may not reuse any of your most recent 12 faces.

#infosec #PasswordExpiration #BYOD

Hot take/rant time for #ActivityPub with the news that #Meta plans to dip their toe in the water.

I would view Meta as an existential threat to a non-corporate, AP-based #fediverse in large part *because* of the immaturity of and lack of interoperability in the AP spec itself.

Why?

Because when you have that big of a fish involved in your space made up of small players, the big fish gets to set the rules unless you have some pretty strong protections in place.

Microsoft’s official @dotnet sent its first Mastodon post today.

Pay attention to the domain: dotnet.social.

Which means Microsoft is operating their own Mastodon community server.

@dotnet account was registered in November, but it became active today.

Clearly, Microsoft sees the Fediverse as core to marketing to developers. I wonder what other brands they’ll bring to the Fediverse as well.

https://dotnet.social/@dotnet/109990127969245035

@fediversenews

@billseitz I feel like a lot of #NoCode is just marketing. Tech is in bed with capitalism. I feel like the revolution is the divorce.

The thing driving the idea that #AI can replace human software #developers is the same fantasy that drives #LowCode products: that every automatable business problem is a collection of templates to be stitched together by a “non-techie” who understands what they want enough to express it in the terms of a watered-down vocabulary or interface.

Like #ChatGPT stringing together words from the Internet and books, it assumes everything worth writing has already been written and just needs remixing.

BBC will not broadcast Attenborough episode over fear of rightwing backlash

The BBC has decided not to broadcast an episode of #David #Attenborough’s flagship new series on #British #wildlife because of fears its themes of the destruction of nature would risk a #backlash from Tory politicians and the #rightwing #press, the Guardian has been told.

The decision has angered the programme-makers and some insiders at the BBC, who fear the corporation has bowed to pressure from lobbying groups with “dinosaurian ways”.

The revelation about Attenborough’s programme comes at the end of a week in which another broadcaster, Gary Lineker, has been condemned by Conservative ministers for his criticism of the government’s immigration bill.

https://www.theguardian.com/media/2023/mar/10/david-attenborough-bbc-wild-isles-episode-rightwing-backlash-fears?CMP=Share_iOSApp_Other

A man files a wrongful death suit against the women who helped his ex-wife obtain medication for an abortion. He is seeking $1M in damages.

We are entering the modern day equivalent of the Salem Witch Trials.

#reproductiverights #women #abortion #democracy #politics #law #lawfedi

https://www.texastribune.org/2023/03/10/texas-abortion-lawsuit/

@cdarwin I have been interested in #Longtermism for a while, but hadn’t fully picked up on the central role of #Oxford #Philosophy in it. Just adding here an article referred to in the one you posted. Seems to me there should be far more knowledge and condemnation of all this.

https://www.vice.com/en/article/bvmanv/ok-wtf-is-longtermism-the-tech-elite-ideology-that-led-to-the-ftx-collapse

@strangeseawolf I think you are spot on, here.

Of course, it gets to the subject of varying copyright law around the world and how the Internet works, and what is assumed free or leveraged. I may be mistaken, but I think I've heard that Disney is fighting this fight but I fear their goal will be to make the tool their own or make it prohibitively expensive, not to protect people who could arguably benefit (and there are benefits). History provides a clue; just look at the fight over streaming music.

Again, I'll elaborate.

Programmers will balk, saying how can we know what the AI is producing? Plus more balderdash.

The fact is the sources the AI learned from can be tagged with extra meta data if programmed to do so. Certain requests could have switches to turn them on and off (the Rembrandt switch). Etc., etc. Opt-in/opt-out is possible. Programmers are ingenious.

The programmers will whine, "But we have a piece of work that we've spent so much time and money producing! We can't fix it to do that.

Tough s**t. You've created a knowledge bomb and it will be your fault if it destroys civilization. Fix the program and teach the AI correctly.

Sadly, people cheat.

This discuss completely by-passed another ticking time bomb. If it is ruled unlawful to learn from prior art and for an AI to produce in-the-style-of artwork, this will be applied to human artists.

We live in interesting times.

#writer #writing #author #ai #writingcommunity #coverart #ai #chatgpt #basicincome #labor #automation #programming #coding #streaming

#BoostingIsSharing

"It is easy now to imagine a setup wherein machines could prompt other machines to put out text ad infinitum, flooding the internet with synthetic text devoid of human agency or intent: gray goo, but for the written word." In the Atlantic, @mkirschenbaum wonders whether we might be headed for a "textpocalypse."

https://www.theatlantic.com/technology/archive/2023/03/ai-chatgpt-writing-language-models/673318/?utm_source=copy-link&utm_medium=social&utm_campaign=share

#LLM #ChatGPT #textpocalypse #AI #AIHype