@prologic@twtxt.net idk, where I work everyone wants to use PowerPoint and they want my slide decks. If you don’t have that problem, good!

@prologic If you ever want to convert revealjs slides to PowerPoint, try my handy guide!

If I had more free time right now I’d write another blog post about this. For now, I just wanted to register how infuriating, tiring, and lousy this firehose of AI this/AI that is.

A lot of people in the US don’t seem to know that cars were crammed down our collective throats in much the same way, over enormous protests. Cars killed tons of people, and building roads destroyed communities on a massive scale. Huge numbers of people protested all of this and more, but cars were rammed through as something we just had to bear anyway.

Many people, including me, have raised alarm bells about this AI technology, and yet here we are having it rammed through in much the same way. It’s a pattern in the United States for sure, if not in the Western world generally. The powers that be don’t seem inclined to slow this process down or regulate it in anyway. I suspect they won’t start until the harms it can cause and are already causing become so great they can’t be ignored anymore.

I posted this on LinkedIn:

ACM, Association for Computing Machinery recently circulated a survey about their authorship policies. I strongly agree with their stance that AI text generators should not be listed as authors. I strongly disagree with their stance that research articles could contain generated text if it is disclosed and meets some other reasonable critiera. I believe the inclusion of such text in research articles fundamentally reduces their quality relative to texts authored entirely by human beings. I also believe, given how AI text generators are trained, that their use is a form of plagiarism. I very much hope the ACM reverses course on that particular aspect of their policy.

@phoronix@feeds.twtxt.net Mozilla receives a significant fraction of its funding from Google. There’s no way in hell they are making “trustworthy” AI.

@lyse@lyse.isobeef.org quite a birb!

Chatbots, deepfakes, and voice clones: AI deception for sale | Federal Trade Commission

It’s good to see at least one US agency taking this stuff seriously.

Peter Thiel is a very bad man. It’s amazing to me that anyone has anything to do with him.

@prologic I never got into Docker to begin with and their quasi-corporate structure always put me off, so I’m glad I don’t need to do anything in response to this mess.

Glaze: Protecting Artists from Style Mimicry

Nice. An artist can run their visual art image through this tool. The tool produces a new version of the image that is almost identical to the human eye, but will prevent unethical, extractive AI like Stable Diffusion or Midjourney from learning the artist’s style, so that their style can’t be stolen and copied. The artist can thus freely post images online without having to worry that some asshole company will co-opt their art style.

They do warn that AI advances quickly and this particular tool will most likely not always be effective. However, I think the effort is commendable, and this tool or some future variant could put enough of a barrier in place that it is no longer cost-effective for lousy AI companies to steal from artists.

@prologic @marado I wouldn’t trust docker anymore if you did before and I’d migrate away ASAP. This kind of thing happens constantly: an actually hostile policy meets backlash, company puts out PR for damage control, and then when the fervor dies down they move ahead with the hostile policy.

@prologic But like @marado said, the RSS feed is clean. And the fact that it’s always 13 repetitions and not an arbitrary number suggests a systematic bug.

@marado @prologic hmmm

@prologic hash the content?

@prologic hmm good to know thanks

LibreTranslate - Free and Open Source Machine Translation API

Nice. Self-hostable even!

@carsten@yarn.zn80.net wow!

If you look at the awesome scala weekly twtxt feed, https://feeds.twtxt.net/awesome-scala-weekly/twtxt.txt , it’s wild. “Issue 356”, the recent one I’m referring to, is repeated 13 times. Everything looks fine back to 2022-11-03T21:42:00Z, when “Issue 337” is repeated 13 times. “Issue 336” is repeated 13 times. “Issue 335” is repeated 13 times. Finally, I got bored and stopped counting.

The only conclusion is that this feed is cursed.

I see 13 in Goryon too. I only noticed 4 yesterday, but now that I’m looking again I see 13.

I’ve noticed this behavior before with other feeds.

@prologic 😆 ugh I know

@prologic yes, it turns out I see 13 copies on the web.

@prologic any idea why I’d be seeing four copies of the same post?

wut

@darch@neotxt.dk Damn.

Plugin proposal for yarn: If a user’s first post contains the string “NFT”, they are auto-banned.