Olives @olives@qoto.org

https://arxiv.org/abs/2405.05904 Apparently, "AI" models have a hard time learning new knowledge via fine-tuning, instead primarily relying on pre-training.

https://www.eff.org/deeplinks/2024/05/no-country-should-be-making-speech-rules-world I urge you to look past Elon's personality and to look at the dreadful precedent it sets. #auspol #FreeSpeech

Someone builds a bot to play a competitive game geared around negotiation -> Ahh, look, a chat bot is deceiving people. *strongly implies ChatGPT is going to go around deceiving people*

Basically, there are a load of oranges to apples comparisons and tortured stretches to try to make out that there is a story when there really isn't.

We also know not to take the people who wrote this seriously because this is quite the piece of misleading clickbait.

https://nichegamer.com/helldivers-2-is-still-effectively-banned-in-countries-without-psn/ Well, that's not good.

It's not a secret that these bots are unreliable, for instance, they have strong tendency to make things up, and someone shouldn't rely on any of it's outputs.

ChatGPT (and other OpenAI models) in particular also have serious symptoms of being over-trained.

Still, chasing every hypothetical and "scoop" gets really silly.

https://www.theguardian.com/technology/article/2024/may/10/is-ai-lying-to-me-scientists-warn-of-growing-capacity-for-deception Do you think The Guardian is scared of ChatGPT competing with their site?#AI #ukpol

I'm seeing people criticizing the U.S. Government for messing with crypto anonymizers. That does sound like a privacy concern.

I think people underestimate just how nosy institutions can be in general and how little financial privacy there is in the mainstream banking system.

While you might be able to interpret it more narrowly, if you squint at it, I wouldn't count on people reading that document doing so.

https://www.wired.com/story/what-happens-when-a-romance-author-gets-locked-out-of-google-docs/
In the year of 2024, someone should be able to reasonably expect that they can mind their own business without a company digging through their private files (or that shared with a small select group). It's likely that legislation might be passed to curb this practice, and already has been in some jurisdictions. That said, just because someone could theoretically look through someone's files doesn't mean they should.

When it comes to moderation practices, it is very inappropriate for #Google to attempt to moderate "sexual content" here, and it feels like something which could easily trip people over. It is inherently user hostile and there isn't a good reason for it.

The article covers both this specific case (and a few other cases more generally and briefly), the following passage is not about this case:

"To a banhammer, every query looks like a nail: depictions of rape disappeared, but so did posts by rape survivors."

There are problems with this passage. For instance, this person focuses on one specific case "posts by rape survivors", and fails to unpack the more implication of pieces of fiction (with dark themes) being censored, which is an obvious incursion on freedom of expression. By failing to engage with the main problem at hand, it is also easier for concerns of censorship to be ignored entirely.

https://qoto.org/@olives/112362450620045294 This is a large part of why I will just point people to my new porn science piece directly.

I'm not saying someone can't cover that case "posts by rape survivors". In fact, it's a fairly important case. What I'm saying is that they shouldn't cover it exclusively.

In a way, this reminds me of someone writing a piece to argue that "ageplay" shouldn't be censored. Instead of making an argument that it isn't a form of abuse, and someone shouldn't be discriminated against because of the actions of a few criminals (guilt by association), they relied entirely on an argument that it wasn't inherently sexual. This isn't an inaccurate argument, in a number of cases it is not, it also misses the point.

#FreeSpeech

Interesting to see people pointing out the privacy implications of VR apps (which a few are rebranding as XR...?).

If someone would keep privacy in mind when building these things, that would be cool.

This cartoon fits this very well. #FirstAmendment #FreeSpeech

https://thehill.com/opinion/national-security/4651053-a-disturbing-national-security-bill-could-silence-nonprofits-and-college-protests/
"This week, the Senate may pass a bill granting the executive branch extraordinary power to investigate and strip nonprofits of tax-exempt status based on a unilateral accusation of wrongdoing.

The potential for abuse under H.R. 6408 is staggering. If it were to become law, the executive branch would be handed a tool perfectly designed to stifle free speech, target political opponents and punish disfavored groups."
#FirstAmendment #FreeSpeech

Sci-fi enthusiast hit by #Twitter shadow ban. #FreeSpeech