indieterminacy

To reference Stack Overflow moderator Machavity, AI chatbots are like parrots. ChatGPT, for example, doesn’t understand the responses it gives you; it simply associates a given prompt with information it has access to and regurgitates plausible-sounding sentences. It has no way to verify that the responses it’s providing you with are accurate. ChatGPT is not a writer, a programmer, a scientist, a physicist, or any other kind of expert our network of sites is dependent upon for high-value content. When prompted, it’s just stringing together words based upon the information it was trained with. It does not understand what it’s saying. That lack of understanding yields unverified information presented in a way that sounds smart or citations that may not support the claims, if the citations aren’t wholly fictitious. Furthermore, the ease with which a user can simply copy and paste an AI-generated response simply moves the metaphorical “parrot” from the chatbot to the user. They don’t really understand what they’ve just copied and presented as an answer to a question.

Content posted without innate domain understanding, but written in a “smart” way, is dangerous to the integrity of the Stack Exchange network’s goal: To be a repository of high-quality question and answer content.

AI-generated responses also represent a serious honesty issue. Submitting AI-generated content without attribution to the source of the content, as is common in such a scenario, is plagiarism. This makes AI-generated content eligible for deletion per the Stack Exchange Code of Conduct and rules on referencing. However, in order for moderators to act upon that, they must identify it as AI generated content, which the private AI generated content policy limits to extremely narrow circumstances which happen in only a very low percentage of AI generated content that is posted to the sites.

c/o https://social.coop/@[email protected]/110490440953441074

This person has nailed the hypocricy:

The enforcement of copyright law is really simple.

If you were a kid who used Napster in the early 2000s to download the latest album by The Offspring or Destiny's Child, because you couldn't afford the CD, then you need to go to court! And potentially face criminal sanctions or punitive damages to the RIAA for each song you download, because you're an evil pirate! You wouldn't steal a car! Creators must be paid!

If you created educational videos on YouTube in the 2010s, and featured a video or audio clip, then even if it's fair use, and even if it's used to make a legitimate point, you're getting demonetised. That's assuming your videos don't disappear or get shadow banned or your account isn't shut entirely. Oh, and good luck finding your way through YouTube's convoluted DMCA process! All creators are equal in deserving pay, but some are more equal than others!

And if you're a corporation with a market capitalisation of US$1.5 trillion (Google/Alphabet) or US$2.3 billion (Microsoft), then you can freely use everyone's intellectual property to train your generative AI bots. Suddenly creators don't deserve to be paid a cent.

Apparently, an individual downloading a single file is like stealing a car. But a trillion-dollar corporation stealing every car is just good business.

@[email protected] @technology #technology #tech #economics #copyright #ArtificialIntelligence #capitalism #IntellectualProperty @[email protected] #law #legal #economics

Writing is a skill and like any other skill the only way to get good at it is to practice. Blogging is ideal for this. Blogging is hardly ever long-form writing so writing posts is a good way to practice in bite-sized chunks. A side benefit of writing is that it’s a necessary component to being a good thinker and writing, as Paul Graham says [1b], is a precursor to thinking. Guo agrees with this and also notes that writing helps him clarify his thinking.

He gives several other reasons to blog including

• Sharing knowledge
• Learning things
• Learning how he’s wrong

[1b1] https://twitter.com/paulg/status/1639954042372730881 [1b2] https://nitter.net/paulg/status/1639954042372730881#m

This is a small curated list of servers to help new users joining Mastodon and the wider Fediverse. All the servers listed have public sign-ups open, have promised to obey certain standards of reliability and responsible moderation, and have opted in to being listed here.

a curated list of servers that comply with certain standards of moderation and reliability

c/o https://social.coop/@[email protected]/110327847898986508

Criteria:

p.s. Some differences between FediGarden and JoinMastodon:

-Servers with over 50k total users will not be listed

-Both Mastodon and non-Mastodon Fedi servers can be listed (though there are still only a few of these at the moment)

-Lists have an ever-changing randomised order and do not prioritise large servers or servers with immediate signups

-Forks are indicated too, if people want a Glitch or Hometown server for example

-Founding dates are shown

https://mstdn.social/@feditips/110327890887600075

What you're missing is the comforting certainty that there's Somebody In Charge. And that this Somebody cares about you, cares about the money they can make off of you, and certainly cares about the 'content' you're producing for them so that they can both mine you and others for precious, precious moneybucks and sell you to their advertisers.

Well, sorry to break it to you, but there are Many Someones In Charge and no one--not anyone--literally no one at all--can tell them all what to do.

Ever hear the expression about herding cats?

That's the decentralized anarchy at the heart of all of this, and what makes it work is communities talking with each other openly, and being able to disassociate from communities that do not suit them--or that might be actively destructive to them.

In many ways, this shouldn’t be a surprise to anyone. The current renaissance in open source LLMs comes hot on the heels of a renaissance in image generation. The similarities are not lost on the community, with many calling this the “Stable Diffusion moment” for LLMs.

In both cases, low-cost public involvement was enabled by a vastly cheaper mechanism for fine tuning called low rank adaptation, or LoRA, combined with a significant breakthrough in scale (latent diffusion for image synthesis, Chinchilla for LLMs). In both cases, access to a sufficiently high-quality model kicked off a flurry of ideas and iteration from individuals and institutions around the world. In both cases, this quickly outpaced the large players.

These contributions were pivotal in the image generation space, setting Stable Diffusion on a different path from Dall-E. Having an open model led to product integrations, marketplaces, user interfaces, and innovations that didn’t happen for Dall-E.

The effect was palpable: rapid domination in terms of cultural impact vs the OpenAI solution, which became increasingly irrelevant. Whether the same thing will happen for LLMs remains to be seen, but the broad structural elements are the same.

Discussion thread for this topic found here:

https://news.ycombinator.com/item?id=35813322

Also, the AGPL isn’t something that commercial operations are afraid of. There are plenty of examples of industry using AGPL software.

The problem (problems, in reality, are actually just opportunities for solutions) is that scheming “software and appliance” companies wish to close source their products (whether they created them or not) under their proprietary brands. They can, and have, launched litigation against those infringing upon or co-opting their trade and service marks, even patents based on their own products - that sometimes was developed and sold without a single scrap of code authored by them. The only thing a permissive license requires is a Copyright notice, somewhere obscure.

“Ohh yeah, we did include a bit of some open source code” is all they gotta admit to when pressed - they never can be forced to divulge that they didn’t write diddly squat!

Maybe… Just maybe… It’s… Your code.

With respect to the patent, all they need to do is change one little thing about the process in which the existing software worked and patent that. Whole 'nother thing though and I’m not here to teach how to engage in patent trolling, like we did at IBM, lolz.

When they do add to that code, their code is under a completely arbitrarily determined proprietary license, locking down the entire product (written by someone else), as closed source under the terms of a license they decide upon - nothing at all remains of the permissively licensed product… Except that obscure little Copyright notice somewhere.

AGPL merely enforces that anything you add to an initially AGPL licensed work must be made available on demand by the people using your product…

Classic educational video.

Nice use of demonstrations and archive material.

We are going to need an Activity Pub 2.0 which fixes the major pain points in the eco system. It must be backward compatible with 1.x, but will introduce new, more scalable, extensible, interoperable ways of doing things for a larger federation and ecosystem. Systems using 1.x can signal it, and those using 2.0 should accept 1.x activities. Those using 2.x can run an upgraded protocol, and when signaling has reached a certain threshold, the old 1.x API can be sunset.

I think the first task on this track is to broadly outline the high level areas that need an upgrade, before drilling into the details. Split AP into its logical functions e.g. identity, payload, discovery, serializations (including parsing), namespacing, ontologies, authentication, signatures, c2s/s2s, large media types, version signalling, extensibility, authorization etc.

Different people will want to work on different things, so maybe the first step is to poll people to see what they want, what are the pain points, and then propose a new spec that gets most of it in.

I understand the reason for this - organization and coordination are difficult tasks. And the team responsible for resolving conflicts should not itself be a source of misunderstanding and conflict.

For this purpose #discord is a very useful, but flawed tool. And alternatives (e.g. #Matrix, #IRC, #Email and #Jitsi) have their own issues - some are just down to network effects, others actually lack features making them less useful for coordination.

https://social.coop/@[email protected]/110294521149402911

But one thing I do want to talk about is the recent idea, I saw probably during a discussion of @atomicpoet:

#Fediverse #autonomous #moderation - the idea that both moderation actions and the coordination of moderation teams should be completely run through the #Fediverse - thus removing a disconnect between #userspace and #moderatorspace on a technological level.

https://social.coop/@[email protected]/110294531671678800

The advantages to organizing #moderation on the fediverse include in my opinion:

• better #privacy, as even when using unencrypted DMs, as long as they stay on the instance being moderated, the instance retains #digitalsovereignty.
• better interactions between moderators and members. #Mastodon, especially, creates an illusion that only one person/account is the face of the instance. That this false for most instances, is a well-known fact. But this is not reflected by the #Mastodon software.

https://social.coop/@[email protected]/110294839498602666

Ive been using Trello for more than 5 years to collect my thoughts, excerpts, and notes to finally realize that I'm pretty locked in. It took me one weekend and some Python to create a workflow to migrate all my boards, including attachments to Obsidian … If you're stuck too, let me know via pm, and I'm glad to share my experiences …

#obsidian #trello #research