Dr Sarah

Storyville - Eternal You A Storyville documentary exploring the digital afterlife business, in which companies are using AI to create avatars of deceased people that the bereaved can interact with.

Recommend this, if y'all can access it... Old now, in GenAI terms, but a disturbing, yet fascinating, view of those seeking to make money from grief.

www.bbc.co.uk/iplayer/epis...

Yes! Every time I log in to bluesky!

Publication day! It's been a labour of love. Thanks to @philswatton.bsky.social and Dan for their ideas, support and encouragement. And to @gypsystitcher.bsky.social for the genius idea of using the ons census!

(lifted from somewhere else)!

I'm sorry but 4.5 is not as good as gpt4!

Spot on from Andrew Marr on the vacuity of the Government's unquestioning embrace of AI and big tech, at the same time as that technology is accelerating the far right radicalisation and economic division of our society

Why? I just want them to help out with the laundry and not scare the cats.

That’s funny. The robots reenactment of Pris being retired in the movie Blade Runner.

This really does align with discussions on narrative progression... I'm actually looking forward to getting back to work, to try this out! :)

@emalliab.bsky.social reckon this might be something I could try with Willowbrook! :o

We may have to rethink consciousness completely. Here's why | BBC Science Focus Magazine In their quest to pinpoint the start of human consciousness, neuroscientists are diving into some truly bizarre territory

Yup, agreed.

"One of the biggest markers of a conscious brain, according to neuroscientists like Frohlich, is the default mode network. This is a network of brain regions that is active when you’re at rest – when you’re daydreaming, reminiscing or thinking about the future."

Also, I think there's some weird psychology happening between non-technical users and the LLM 'personalities'. The excellent use of natural language (and sitting on a huge amount of knowledge) leads to an enhanced (misplaced) belief in the systems' capabilities.

I might do a paper on it.

💯. too many managers suffer from fomo, or are obsessed with cutting the wage bill.

I hate the phrase but we need more clear eyed thinking in commissioning.

LLMs do have strengths, but from my point of view they are best suited to helping an expert, who is able to make instant value assessments.

Or monetized.

Anyone else notice when theres a lot of posts about LLM/LLM agents systems failing... They're quickly followed up with posts about possible consciousness?

I cant tell if such posts come from hope, cope or denial... Or are simply monetary.

He's been hanging out with too many tech bros!

You expect too much.

Current ai shouldn't be used to review papers imo, as it's not accurate enough.

As for being robust against prompt attacks? Not likely with LLMs

The belief that these abilities are around the corner, is part of why the system of review isnt being properly fixed.

Wow...

I won't be signing up. I agree young people need some protections, but bluesky's solution isn't a good solution.

And if the system therefore assumes I'm a child and stops putting all the weird furry related posts on my timeline, then #winnerwinnerchickendinner!

Replit makes vibe-y promise to stop its AI agents making vibe coding disasters Announces beta for separate production and development databases that will land in a few weeks Vibe coding service Replit has announced changes to its product that should prevent the database deletion disaster reported by one of its users.…

Replit makes vibe-y promise to stop its AI agents making vibe coding disasters

We have decided to ban the use of GenAI for research, writing & creative work at our organization.

In fact, we make people sign an agreement saying that their research, analysis, writing, and creative work are *theirs* and not done by GenAI.

Personally, I think big tech are exploiting our need for connection. They didn't need to train them to have such identities.

There's no need for the LLMs chain of thought outputs to be in the first person. Just smoke and mirrors.

Given it appears he's making money from it, people must be lapping it up.

Personal opinion, the more posts like this, the more the reality will dawn, on the limitations of these systems and those that believe in them.

There is no mechanism for honesty in an LLM.

Why do people routinely believe responses when LLMs are called out for disappointing behaviour. People treat them like naughty children that fundamentally know right from wrong. When they have no concept of either.

Agreed, he's treating it like it's human, not just an employee. No grasp of what the LLM is capable and not capable of.

The bit that really gave me the ick, was 'you asked me to 100% honest, and my honest reply is.... '. #facepalm.

An OpenAI Investor Appears to Be Having a ChatGPT-Induced Mental Health Crisis Bedrock co-founder Geoff Lewis has posted increasingly troubling content on social media, drawing concern from friends in the industry.

I do sometimes wonder if AI-induced psychosis is much more widespread at the upper echelons of the tech elites than we think. It might explain some of the more ridiculous ideations, and associated convictions, for our collective future ... futurism.com/openai-inves...

They reflect their training data, therefore they reflect *our* own biases.

Wow

@philswatton.bsky.social yikes ...

Screenshot: No room for wokeness! @Wokenessisevil Notice how no one, not a single soul, has ever complained about the female gaze in entertainment. [photos show Hugh Jackman (Wolverine) and Alan Ritchson (Reacher) with ripped muscles]

This isn’t the female gaze. This is the male gaze. It’s men who are obsessed with the physicality of other men. Women, in the main, would much rather see a man who looks like he’d make himself a stack of buttered toast and let you steal a slice