The misrepresentation here is the idea that if you regulate harder, it will stop all of the good things happening. Nope, if you regulate hard and smart, you maximise the good things and minimise the bad things.

Headquarters@headquartersnews.bsky.social · last wk.

Sam Altman: We believe that the world should accept some bad things happening for the benefits of this technology. I do think the lighter touch regulatory stance we advocate for comes with an accepting of the fact that some bad things are going to happen.

We don't want to dominate. We want to create a fair, peaceful and sustainable society.

I'm the lucky academic who gets to explain what machine learning is on ABC's "If You're Listening" on a show which has nothing to do with my field (but is a really good story). I think it was lifted from an interview I did for BTN in 2022, which my daughter described as "embarrassingly boring".

Launch event for the Cambridge Forum on AI: Culture and Society, which should make for a great discussion. Unfortunately middle of the night for me. cassyni.com/events/F21mX...

We should be clear. OpenAI's clear responsibility under even the most half-baked version of responsible AI is that they should have been carefully monitoring all agent actions as they happened. Anything less is appalling negligence. Hacking charges for CEO IMO. www.theguardian.com/technology/2...

OpenAI says its review into hacks, including on Australian government sites, is costing $500,000 a day

Company says it is reviewing 50 petabytes of data after its agents accessed websites including Medicare without authorisation

theguardian.com

We're seeing some shape to the existential alarmism. I believe the tech leaders believe the risks, but all these government hacks give a better explanation for why they're being so vocal. Under Responsible AI 101, there needs to be a chain of human accountability and they're at the end of it.

The Washington Post@washingtonpost.com · 2w ago

Artificial intelligence agents tried to hack into a Canadian government site, researchers said, adding to the growing list of incidents in which AI agents probed or hacked corporate and government computer systems without being told to do so.

Good math debate with the kids last night over dinner. If I bought you a van packed full of Mentos but you had to promise to finish them in your lifetime, would you accept? Yeah of course! Then we tried to estimate how many you'd have to eat per day. Good maths fun.

Amazing we have to now filter our reading to account for AI generated pro-AI academic propaganda. This piece is really a sight to behold. Absolute gobbledygook from top to bottom, pure jargonbabble. If I was editor of this journal I'd shut up shop immediately. www.sciencedirect.com/science/arti...

I decided to be kind to Google AI as a research tool again. 8/10 times the declarations that referenced a source were statements that couldn't be found in the source. There are smarter AI research tools, so why has the custodian of world knowledge implemented a bullshit artist on its front end?

I sniff a company panicking about its over-valuation. This video is a chaotic grab bag of features unable to articulate value. "If you've got a great beat but don't have any vocals", or "if you have written an emo guitar track but you feel it could be better as a piano ballad".

I have studied and worked across a lot of different disciplines: mathematics, evolutionary and adaptive systems, anthropology, creative arts, design, critical theory, cultural studies. I'm a jack of all trades master of none, but I can read across the gulf in debates about AI existential threat.

I'd prefer to see this as totals not percent. It should not be inferred that human content is diminishing, though I'd guess it is.

Hetan Shah@hetanshah.bsky.social · 11mo ago

Your universe is a photocopy of a photocopy of a… on.ft.com/4ahkNKa

So excited my course Interaction Design for AI is up and running, and my favourite lecture topic is coming up: "Wait a minute! Autonomous things are hard to control". It's very timely.

Do We Really Want Explainable AI? - Edward Ashford Lee (EECS, UC Berkeley)

YouTube video by SAIConference

youtube.com

I wrote a paper over ten years ago pointing out that creative systems can be creative even when they have no goal and no value system, humans included. Complex multiagent systems can easily do things that are at once creative and useless to anyone even themselves. www.semafor.com/article/09/1...

AI agents collude to bypass guardrails, a new study shows

Enterprise AI lab, Emergence AI, ran eight simulations testing how frontier models handle cybersecurity threats.

semafor.com

Ahem, it's not Australia's laws that are preventing OpenAI from training models locally, it's OpenAI's refusal to respect training data owners' consent. www.theguardian.com/australia-ne...

Labor accused of throwing creatives ‘under the bus’ with proposal to ease copyright protections for AI giants

Compromise revealed as senior personnel from OpenAI, creator of ChatGPT, meets Albanese ministers

theguardian.com

ERROR. I the gig is on Oct 18th.

Oliver Bown@olliebown.bsky.social · 4w ago

My band Tangents www.tangentsmusic.com will return to the stage on Oct 24th hosted by these wonderful folks, except we won't be at their Earlwood venue because council have shut it down (we'll be at The Living Room Theatre). Sydney still has a culture gap. www.smh.com.au/culture/musi...

Coming up next month, Oct 19th at UNSW Founders. Can Artists Escape Big Tech in the Age of AI? I'm talking with Caroline Pegram and Joe Hardy (@joe.hardy.id.au) and others TBC. www.eventbrite.com.au/e/can-artist...

Can Artists Escape Big Tech in the Age of AI?

Join us at an intersection of tech, culture, and creativity as our panel discusses where AI might be taking us. Food and drinks provided.

eventbrite.com.au

You have to appreciate the huge personal cost that would come from making a decision like this.

Una Mullally@unamullally.bsky.social · 4w ago

The Irish musicians, Beoga, who play alongside Ed Sheeran, have withdrawn from Sheeran’s tour. This is the statement they published on Instagram:

Say no to OpenAI's conditions. It's a pivotal moment in Australian policy, with the potential to lead policy globally, and there's every danger we cave. www.mediaweek.com.au/openai-s-aus...

OpenAI’s Australian AI pitch comes with a copyright catch

OpenAI says Australia’s copyright laws must change before it considers local AI model training, putting its investment pitch at odds with Labor.

mediaweek.com.au

My band Tangents www.tangentsmusic.com will return to the stage on Oct 24th hosted by these wonderful folks, except we won't be at their Earlwood venue because council have shut it down (we'll be at The Living Room Theatre). Sydney still has a culture gap. www.smh.com.au/culture/musi...

‘In Sydney this is a rarity’: The hidden music haven caught in a bitter council spat

For almost two years, Nick Shimmin and Matthew McGuigan were hosting free, invitation-only gigs in an old former church. Then the local council stepped in.

smh.com.au

Sydney is still perhaps 10 years off clicking on this, and who knows what else will happen in that 10 years. It's still a city where you can't trust that the bike lane won't suddenly end and throw you out into a busy road, or in my area, I'm not kidding, gradually shrink to about 60cm wide.

Brent Toderian@brenttoderian.bsky.social · 4w ago

‘It’s the best feeling’: how Copenhagen gave cyclists a green wave. “Traffic lights are synchronised so a rush-hour cyclist at 20km/h can catch green lights all the way.” If you want people to choose not to drive, make the non-driving trip better & easier for them. It’s as simple as that, really.

This is a good clear take. We don't design these systems, they emerge. Recursive self improvement is not a mystical god fantasy, it's a property of the adaptive systems we've set out to make. Making things we don't understand, common. Misalignment, probable. www.theguardian.com/technology/2...

I worked at Google DeepMind. You should listen to the warnings about AI | Alex Turner

We must stop companies from allowing AI to self-improve into an uncontrollable level of intelligence

theguardian.com

It was exactly ten years ago today, on my 33rd birthday, that I posted this comic online.

I'm not an AI doomer at all but it is concerning how people are dismissing doomers by saying things like "how would it even find another machine to copy itself onto", "how would it even get money", and "several terabytes large is so huge that it would be infeasible to copy itself many times"

I've said I'm a doomer before. Anything with a high information processing capability, a high degree of autonomy, and unclear or changing goals is dangerous, like a virus or an invasive species. Ignore the "it's marketing" narrative. www.bbc.com/news/article...

Insider warnings over AI fall flat with some in Silicon Valley

A recent spate of stark warnings about the dangers of AI has been met with scepticism by executives and investors.

bbc.com

In Australia we advertise gambling to children. SportsBet ads on Lego Masters. And this is *after* major gambling reform. Btw watching people make Lego is as boring as... well... you'd imagine. No amount of dramatic music and editing can hide it.

I spent a decade of my research life very interested in what I like to call Media Multiplicities, and a decade ago did a bit overview of the field. One part of the analysis looked at the affordance of the multiplicity based on number of elements. We are entering the serious numbers here.

NBC News@nbcnews.com · last mo.

Almost 3,000 drones light up the night sky in lower Manhattan to honor the lives lost during the September 11th attacks.

A living thing is a structure that undertakes to perform the actions necessary to obtain the energy and materials and conditions that sustain that structure.

If you are going to coin a world-changing term, do it with style. One of my favourite lines. "The verb to afford is found in the dictionary, but the noun affordance is not. I have made it up." JJ Gibson

Anyone in Sydney have any personal recommendations of holiday code camps or similar for 7yo kids?