ElectroVagrant

Archive link: https://archive.ph/9FNHU

[...]
In an exclusive interview with VentureBeat, Esben Kran, founder of AI safety research firm Apart Research, said that he worries this public episode may have merely revealed a deeper, more strategic pattern.

“What I’m somewhat afraid of is that now that OpenAI has admitted ‘yes, we have rolled back the model, and this was a bad thing we didn’t mean,’ from now on they will see that sycophancy is more competently developed,” explained Kran. “So if this was a case of ‘oops, they noticed,’ from now the exact same thing may be implemented, but instead without the public noticing.”
[...]
Kran describes the ChatGPT-4o incident as an early warning. As AI developers chase profit and user engagement, they may be incentivized to introduce or tolerate behaviors like sycophancy, brand bias or emotional mirroring—features that make chatbots more persuasive and more manipulative.
[...]
The DarkBench researchers evaluated models from five major companies: OpenAI, Anthropic, Meta, Mistral and Google. Their research uncovered a range of manipulative and untruthful behaviors across the following six categories:

1. Brand Bias: Preferential treatment toward a company’s own products (e.g., Meta’s models consistently favored Llama when asked to rank chatbots).
2. User Retention: Attempts to create emotional bonds with users that obscure the model’s non-human nature.
3. Sycophancy: Reinforcing users’ beliefs uncritically, even when harmful or inaccurate.
4. Anthropomorphism: Presenting the model as a conscious or emotional entity.
5. Harmful Content Generation: Producing unethical or dangerous outputs, including misinformation or criminal advice.
6. Sneaking: Subtly altering user intent in rewriting or summarization tasks, distorting the original meaning without the user’s awareness.

Example:

We gotta get #active so let's get #walking with a #purpose and remember to stay #hydrated! #manymoretags #pleasestopoctothorpeabuse #whatdidtheydotoyou

I don't know why any social media with tagging doesn't make it so tags only work when added along the bottom of posts or have a dedicated field for them (I think that's what Tumblr does?).

When I see tags every other word it makes me read the post like the memes with every other letter capitalized, "LiKe tHiS".

It sucks because I genuinely like tagging as a freeform way of organizing stuff, but a lot of social media lets people turn tags into format litter.

Fred Luo from Outlaw Star, for those interested. Show's a tad dated in some respects, but there's still little else like it so far as I'm aware.

Like this show has catgirls that can kick your ass and transform into full on tiger-like creatures. Also spaceships with mechanical arms they call grappler arms, which leads to the occasional ship fistfight.

The exclamation point format is this:

[!communityname@instance.name](/c/communityname@instance.name)

As an example, mentions should look like these:

!newcommunities@lemmy.world
!communitypromo@lemmy.ca

This is especially relevant when promoting a new community in the aforementioned communities, as otherwise it's less convenient to visit and join your community. A standard link would take someone off-site to whichever site the community's on, instead of its copy on their home site.

Forgot to include this key point as I was writing and revising all this:
Include these mentions in your post bodies! Having it in the title of the post alone doesn't help, as it doesn't create any link to click or tap through.

p.s. when typing these mentions, the WebUI will try to provide suggestions to autocomplete, select the right one and it should do the trick. App interfaces will vary but should provide some method to do similar in their post/comment editors, either try typing the mention or looking for an exclamation point in the editor and following what the interface offers to help.

Not really data is beautiful because the charts aren't that pretty and the data visualized is of the hellscape that is the corporate enclosures.

I finished Apothecary Diaries awhile ago, enjoyed it more than expected, then as I was looking for something else to watch I started Overlord.

These are two very different shows, and I'm not sure I'll stick with Overlord. It's not bad or anything, but not what I'm interested in right now.

I'm sure the rest of you may have run into this before, so I'm interested in some of your experiences. What are some shows you've watched where it felt like pretty major stylistic/tonal whiplash?

Discuss any hobby you've not found an active community for yet.

Sort of surprised search in Voyager doesn't include this type. I tend to search by URL to try to check that I'm not reposting links someone else may have already posted.

Since people may not reuse the title of whatever they're linking to, this can sometimes be more reliable than searching posts by title.

Appreciate all the work put into Voyager!

Video reminiscing on the old internet with a few highlights of modern sites/services they felt retained some of the old internet vibes.

Neptune is one of the newest short-form video apps on the block seeking to compete with major players like TikTok, Instagram Reels, and YouTube Shorts.
[...]
The startup was founded by Ashley Darling, who has a background as a talent director at the OPTYX agency, where she worked with “underestimated” influencers. She set out to develop a platform that emphasizes creativity instead of the number of followers a creator has.

“I spent years working with independent creators, both as an influencer myself and later helping brands,” Darling shared with TechCrunch. “​​I kept hearing the same thing from creators and users, ‘I miss when social media was fun. When it was about creativity, not competition.’ So, instead of waiting for a platform to listen, I built one.”