Vic "Loxley" Collins

Vic "Loxley" Collins

@satyricorner.live

Voice Actor & Vtuber | Queer Neurodiverse Londoner | he/him | 37 | 🔞 18+, Minors DNI https://www.satyricorner.com/ https://twitch.tv/satyricorner Design: Francesca Dare Model Artist: @superhumanfoods.bsky.social PFP: @lokiphantom.bsky.social‬

3.7K followers1.3K following4.8K posts

A bug would be a flaw in software producing unintentional behaviour What's shown below is a change log in which they added a bunch of adult sites to a list called `BLOCKED_DOMAINS` They either did this deliberately and are backtracking or there was no oversight when this change was published

ZippyVtuber🔞@zippyvtuber.bsky.social

Good news! Apparently it’s a bug. We’ll have to see if it gets fixed later on

From an outside perspective, it may seem like the UK getting extreme, repressive rules pushed on it by foreign influence is justice for most of the last thousand years Sadly, we've just become a Petri dish, testing how to build support for these views, before they're sent back out into the world.

Artwomble@artwomble.bsky.social

I can't help but feel this is where we're headed sometimes.

Unironically the Direct I'm most excited for this year!

Mystery Wrestling@mysterywrestling.bsky.social

What's coming up for #MysteryWrestling in 2027? Find out on Thursday, November 5th for #MysteryWrestlingDirect at 4 PM PT | 7 PM ET on Twitch and YouTube.

Zach goes on from here to say that most artists, even the greats, are mid (subjective), but I think he fails to acknowledge that someone's "mid" art is practice and progression, even if you're phoning something in to make sure you can have a roof over your head

Zach Weinersmith@zachweinersmith.bsky.social

I think it's not a great argument against AI eating a lot of the arts to say that AI can't be creative/original. Setting aside the question of whether they can... does *most* art strike you as extremely original? The vast majority of people, including great artists most of the time are mid.

View full thread

AI meanwhile, has a limited context window, which doesn't really benefit the iterative nature of art, nor the sprawling amounts of information an artist can gather and develop, which informs the work while never being written down or imagined - and this is before accounting for hallucinations

Ultimately, I understand the point OP is making, if the art is of equal/mid level quality, does it matter what created it My counter argument would be that between AI and a human, only one is going to be able to use that "mid" creation to have artistic innovation in the future

Zach goes on from here to say that most artists, even the greats, are mid (subjective), but I think he fails to acknowledge that someone's "mid" art is practice and progression, even if you're phoning something in to make sure you can have a roof over your head

Zach Weinersmith@zachweinersmith.bsky.social

I think it's not a great argument against AI eating a lot of the arts to say that AI can't be creative/original. Setting aside the question of whether they can... does *most* art strike you as extremely original? The vast majority of people, including great artists most of the time are mid.

I know this reading is not intended, but Wodehouse didn't crack out "the three greatest comic novels" at the outset of his career and then toss off "generic Edwardian light comedy" for his subsequent 70 books Artists evolve over time, they iterate, revisit and build on ideas

Zach Weinersmith@zachweinersmith.bsky.social

And it's worth noting, the great artists were often sustained by making or helping in the making of mid stuff! Wodehouse wrote the three greatest comic novels in English and about 70 novels that are generic Edwardian light comedy.

We had a dishwasher error, and while I was cleaning the filters, I noticed other bits in the kitchen that needed doing, so I cleaned the sink, put on a load of laundry, swept and mopped the floor Now I'm done, content and I have a clean kitchen Is this what it feels like when ADHD is useful?

I've been playing Two Point Museum a lot, and expanding the concept into aquariums, zoos and theme parks is a stretch BUT putting them all in the same game means you can blend all these interconnected themes in one museum! A monorail around your safari park, a ghost train next to the actual ghosts!

Two Point Museum 🦕@twopointstudios.bsky.social

Whoever said history can't be THRILLING? 🎢 Restore the County's old fairground and bring the past to life with high-octane engineering, immersive experiences, and... era-accurate queue times? Two Point Museum: Rides & Relics arrives 8th October 2026 🎡 youtu.be/XPB1CKK2adI

It probably helps that Riquetti Gardens puts me in mind of Alton Towers, which is built in and around the ruins and gardens of a stately home. Maybe it means this blend doesn't seem quite as unusual to folks in the UK.

Three MONTHS? That's a deeply worrying loss

Elf M. Sternberg@elfsternberg.bsky.social

In a study of doctors in Poland, using an AI for three months, then ceasing to use it, resulted in a 20% loss of skill and accuracy. Cognitive systems are metabolically expensive. When your brain knows there's a substitute, it resents not having access to it.

Vic "Loxley" Collins reposted

I remain very skeptical about AI, but I recently learned about OLMo (Open Language Model) from @ai2.bsky.social, which seems to avoid a lot of my biggest objections The training data for most models is derived from human work that has been unethically sourced and deliberately obscured

That worked incredibly well I think my strategy going forward may be to stream directly to @stream.place, with a multistream broadcast to Twitch, and leave a pinned post directing folks to the "real stream", thereby slowly moving myself away from the platform

Vic "Loxley" Collins@satyricorner.live

I just learned that @stream.place lets you forward your stream to other streaming services That's super useful and really something they should point out more I'm going to try a little experiment in half an hour...

I remain very skeptical about AI, but I recently learned about OLMo (Open Language Model) from @ai2.bsky.social, which seems to avoid a lot of my biggest objections The training data for most models is derived from human work that has been unethically sourced and deliberately obscured

View full thread

There are still issues: hallucinations will still occur and this model doesn't address social issues of AI, such as AI psychosis, individual isolation, validating erroneous beliefs, and Nor does it prevent the atrophying of skills reported by some software developers when used as a coding tool

I remain very skeptical about AI, but I recently learned about OLMo (Open Language Model) from @ai2.bsky.social, which seems to avoid a lot of my biggest objections The training data for most models is derived from human work that has been unethically sourced and deliberately obscured

View full thread

On top of this, as an open source model, OLMo is intended to run locally or with a cloud deployment, so would not cause the environmental or privacy concerns nor tech shortages of the major corporate models This does address one of my major objections to local models I made six months ago

Vic "Loxley" Collins@satyricorner.live

I also looked into the "open source" LLMs available and each one of them is ultimately being created and trained by large corporations (even the smallest has over 100 employees) Those companies making some variants of their models "open source" does not stop them being corporate products

I remain very skeptical about AI, but I recently learned about OLMo (Open Language Model) from @ai2.bsky.social, which seems to avoid a lot of my biggest objections The training data for most models is derived from human work that has been unethically sourced and deliberately obscured

By comparison, OLMo's training data is publicly visible. That still doesn't necessarily mean they're not scraping information without consent, but it does make it harder to hide it Research by Rachel Hong, et al. found OLMo derived some training data from other datasets, so it's not entirely clear

Rachel Hong@rachelhong.bsky.social

For instance, we investigate AI2’s Olmo 3 LLM and trace the models and datasets involved in its development. We find that Olmo 3 is actually dependent upon many other AI datasets and models. These interactions illustrate an example of what we call AI entanglement. [3/N]

I've been off work sick due to burnout for a while, but I'm getting ready to go back in about a month During that time, it would benefit me to come up with a solid list of reasonable adjustments to accommodate my ADHD

Both Firefox and Waterfox are taking Ls. This is my current browser of choice, it takes a little getting used to, but the complete absence of AI is totally worth it.

Zen@zen-browser.app

We noticed a rising interest regarding the use of AI in browsers. To be clear about our position on this: Zen currently is disabling all AI features included in Firefox and will continue to do so.

Sushi Cat meets Bloons... Hey @rinpenrose.bsky.social, might be one for the queue

Armor Games Studios@armorgamesstudios.com

🚨 OUT NOW (Meow Available) 🚨 🍣 Sushi Cat – Tower Defense is LIVE! 😼 The sushi battle begins. ⚔️ Build defenses. Place towers. Survive angry sushi. Buy & Play today 🖥️ Steam 👉 armor.ag/SC-TD 🎮 Switch 👉 armor.ag/SCTD-Switch #SushiCat #TowerDefense #IndieGames #Gaming

Woah, @itsquinns.bsky.social on dropout! You love to see it!

Dropout@dropout.tv

This week on Dropout... 🃏 Monday: Game Changer (with Lily Du, Brian David Gilbert, Vic Michaelis, Jacquis Neal, and Ele Woods) 🎨 Tuesday: Toonout 🗣️ Wednesday: Adventuring Academy (with Quinns Smith) 🐷 Friday: Smartyshorts 👇👇👇

The thread below is explaining a thought experiment, so no shade intended to OP, but assuming a metaverse as envisioned by Meta seems to be doing a lot of heavy lifting, considering the project was shelved in March of this year Beyond that, we can just reject the paradigm of a "digital afterlife"

This post is unavailable.

Even taking the base assumption as given, the question being asked is would you prefer a digital avatar of yourself that exists beyond your death to be as accurate as possible, or only have limited data? Part of me doesn't care - I'll be dead! But I'd prefer a model like that to be inaccurate

My loved ones should be allowed to grieve, mourn and celebrate me in their own fashion, not be beholden to some weird computer recreation that never allows them to accept my passing

Are there any good Lego games? Genuine question; I've played a few with my kid and they're quite frustrating for us both because of a combination of unclear puzzles and bad controls

Watching a kids movie in the cinema, so I just learned about Armoursaurs, aka we have Power Rangers at home I'd be less dismissive, but when your T-Rex is red and your triceratops is blue comparisons will be made...