Instagram Story Save

How can a Nano Banana prompt expose a much bigger issue in AI bias? Share the similar patterns you’ve encountered.

Icaro Lab just discovered that POETRY can jailbreak LLMs… exposing a blindspot in AI safety research.

Comment “PAPER” and I’ll send you the study 😉

I wanted to weigh in on the AI Monet Water Lilies situation.

Those unfamiliar… the anonymous artist SHL0MS trolled everyone to prove a point about AI commentary.

But what I think it actually goes to show is how we need to maintain credibility more than ever. This has really shone a light on how difficult it is to know what’s up from down in this wild world of AI and misinformation.

Digital rights are human rights.

Many of the most powerful AI systems are being built by a handful of countries and companies, while the people who’ll be most affected aren’t part of the conversation.

I was so grateful to be invited by @foreignanddevelopmentoffice to attend the Global Partnerships Conference this week, where conversations about violence against women and girls and how it intersects with AI came up again and again.

I started posting on here because I wanted to inspire more women to be in the rooms where the decisions about our future are being made. Having a window into the partnerships being built around AI as it relates to these pressing issues was a real privilege.

Human rights apply online and offline, especially as AI continues to shape both.

Comment PATH for ways you can get involved right now ⬇️

I fundamentally believe that there is a way to make tech work for us, but sadly, that is not the trajectory we’ve been heading down with AI… at all. In fact, much of our technology has facilitated the concentration of power into the hands of a very few tech companies and a very small number of countries, and if we continue down this path, things will very likely get so much worse.

A Better Path for AI, created in part by @futureoflifeinstitute , introduces a framework for what a different path could actually look like, and how we get there through policy, governance, and technological development.

Not sponsored, just excited about solutions-driven approaches to what is going on with AI, though I do partner with Future of Life Institute on other projects.

Link to betterpathfor.ai can also be found in my bio.

Comment “HACK” to apply.

On 6 June I’ll be co-hosting the AI Risk Content Hackathon with @futuretense.tv and @bluedotimpact

It’s no secret that there’s a ton of noise in the AI space right now. From talk of a “race with China” to AGI being a “cure for cancer”…

I’m looking for content creators, journalists, social media marketers… basically anyone who is concerned with the current trajectory of AI and wants to help people actually understand it.

Full day at LISA in London with food, drinks, and prizes.

Link to apply is also in my bio.

ChatGPT is obsessed with goblins… and it’s actually a really solid… and concerning example of an underlying problem with AI.

7th June Event is LIVE☕️🎤👀

Panel & Pastries is happening on June 7th in London and these three amazing women are going to discuss AI risks & safety for beginners and those who are interested!

Nina Wolff-Ingham - Operations Lead at the London Initiative for Safe AI, working to make conversations about AI risk more accessible to people outside the field.

Caitlin Yardley - AI safety communicator and content creator behind FutureBriefing, turning complex AI research into content that actually makes sense for everyone.

Jenny Huzell - Founders Associate at AI Whistleblower Initiative, working at the heart of the AI insider community and ecosystem.

A morning of pastries, matcha, honest conversation and a goody bag to take home

LINK IN BIO

A huge thank you to @bluedotimpact for sponsoring this event and helping make it happen
And thank you @fableandfalcon for allowing us to host it at your book shop/cafe!

#theaigirlsclub #panelandpastries #AIrisk #AIsafety #womenintech

These poor dogs…

@rspca_official RSPCA had to respond to comments after an image of theirs was accused of being AI-generated.

It was, in fact, a very real photo taken by an RSPCA rescuer of 250 dogs crammed inside one home.

While the dogs have since been rescued, the situation speaks to a massive epistemic problem.

Stevie and Sandy are still looking for their forever home at Southridge Animal Centre north of London.

Link to their adoption profile in my bio ⬆️

Is a sandwich more regulated than AI?

Over the past year, we’ve seen people like Yoshua Bengio and Tristan Harris repeat
that there is little to no regulation of AI in the US...

And at the federal level this is absolutely true!

Link in bio to take action ⬆️

#vegan #AI #aiSafety

Comment “PAPER” and I’ll send you the link.

AI companies want us to believe superintelligent AI will cure cancer.

Cancer has touched every single one of us, and that shared grief is being used to justify building bigger and bigger AI models while billions are pulled away from the research that could actually save lives.

This is just one of many insights physician and researcher Emilia Javorsky makes in AI vs Cancer.

Full essay and roadmap linked in bio.

The leading AI models vs the trolley problem…

The responses show how the “moral”inconsistency of LLMs.

Try out the tool for yourself by clicking the link in my bio or by commenting “QUIZ” below.

#AI #trolleyproblem