Jacob Aron

Jacob Aron

@jjaron.bsky.social

Senior editor at New Scientist

5.8K followers3.5K following5.2K posts

Pinned

I used to share book recommendations on the other place but got out of the habit this year as engagement massively dropped off. Thinking about putting together my top 10 list for this year though, so as a test, like this, and for each like I'll recommend something great I read in the past 5 years

As a child I had this, it was incredible, I would read it cover to cover. Modern equivalents seem to be a single volume and picture heavy. Obviously all the information is on the internet, but what am I meant to do, hand a child Wikipedia? No thanks.

My latest AI Explained column argues that concerns over vibecoded copycat apps and AI agents are part of a larger battle around who gets to control what happens on your computer. The browser wars, ad blockers and Napster point at what might happen next www.newscientist.com/article/2595...

newscientist.comPeople are vibe coding copyrighted apps with AI – will tech firms sue? | New ScientistEvery time a new technology hits the internet, there is a battle between the rights of users and major companies. To see how this will play out with AI, columnist Jacob Aron looks to examples from the...

"What's a VHS?!" my daughter said, with a tone of disgust I couldn't possibly replicate in text form. I busted this out in response, first published 1988

Is there anything less accessible than an academic paper published as a book chapter? Either you have to pay a ridiculous amount to get the chapter by itself, or an even more ridiculous amount to buy the whole book that you probably don't want.

Jacob Aron reposted

Replying to @gralefrit.bsky.social

If a show isn’t available on Sounds or iPlayer, it’s because a deal hasn’t been done to pay the creators. They can’t just jack you into the Big Hard Drive Of Everything Your Parents Paid For.

Jacob Aron reposted

Are we gonna do this, Bluesky? Are we all just gonna not understand accounting? 72B$ of the 80B$ is convertible notes. They became greater liabilities because the company's valuation skyrocketed. They don't have an 80B$ loss in the sense that almost everyone here is imagining.

This is 20 years of books. At some point I'm going to have to admit that this is too much. Probably once the list exceeds my expected remaining lifespan.

I used Claude and my actual reading data to project the likely point at which I will no longer be able to read all of my to-read list before I die

Jacob Aron reposted

Experts are never going to be able to validate all AI solutions, whether we're talking maths, cryptography or anything, really. I'm not really sure what the solution is.

Margot Finn@eicathomefinn.bsky.social

'Cryptography specialists are thrilled at the surge in interest in codebreaking — but they are also being overwhelmed by a deluge of requests to validate AI solutions, which can often be incorrect.'

Cybersecurity relies on people choosing large, hard to guess numbers as private keys. They are mathematically transformed in a way that is impossible to reverse engineer. This does not work if you accidentally choose 4 www.theregister.com/security/202...

theregister.comIndian certification authority eMudhra’s private keys compromised – one of them was the number ‘4’Massive clean-up job under way because this mess makes it possible to impersonate any website or service

Jacob Aron reposted

Replying to @colin-fraser.net

Even without the specific weird terminology, this is just such a bizarre thing to say, a super weird way to express yourself, and can’t be properly understood outside of the very unusual epistemological / ethical framework that these guys all have.

Scanning through Steam Next Fest demos and I reckon I have a 90% hit rate on guessing whether a game is AI generated based on the store text alone

Going to go against the grain and say that I actually think the Amol Rajan Connection is very sensibly priced - if you look at the full tier list it goes from free to bonkers. This lets you capture value on all levels - you don't want to make it hard for superfans to give you *more* money

Now, I will question whether Amol Rajan actually has any superfans. But in general, media outlets could do with learning from mobile game business models, where you make it easy for your most engaged customers to give you stupid amounts of money while also having a free option at top of the funnel

Jacob Aron reposted

The box for this game is EXACTLY 12-inches long. You know what that means, don't you? The game is a foot.

Experts are never going to be able to validate all AI solutions, whether we're talking maths, cryptography or anything, really. I'm not really sure what the solution is.

Margot Finn@eicathomefinn.bsky.social

'Cryptography specialists are thrilled at the surge in interest in codebreaking — but they are also being overwhelmed by a deluge of requests to validate AI solutions, which can often be incorrect.'

I wonder if we are in an awkward middle period, though. These solutions are the low-hanging fruit that no one got around to finding. You can imagine eventually being able to ask an AI to solve all such problems at once, and thus no longer relying on random people who just happened to prompt

That would allow for more systematic study ("Here is your morning digest of solutions to unsolved problems, ordered by importance") but human attention is still a bottleneck. The only solution is to let a machine rubber-stamp problems as solved, but then you can rightly ask what is the point?

It's fine, I'm sure it's not urgent

Adam Vaughan@adamvaughan.bsky.social

The IPCC, the UN's climate science panel, has failed to agree a timeline for its next next big report. Was due in 2028, but some countries have been pushing for it to be delayed until 2029.

Just read Diamond Dogs by Alastair Reynolds, an extremely cool novella about a team exploring a strange and deadly alien tower. Highly recommended!

"I'm not a Frankenmodel, I'm not a Frankenmodel," I continue to insist as I shrink into a corn cob

Colin@colin-fraser.net

Frankenmodel refuses to believe it's not just Claude

Jacob Aron reposted

Replying to @hern.bsky.social

English journalism: ROBOTAXI BAN BACK ON American journalism: in Wisconsin town, robotaxis not not not banned, council says mayor says council says

This is how we get the Borg

Colin@colin-fraser.net

the frontier labs don't make doing this with frontier models easy but the open models are working great together. Great job!

For the morning crowd, my take on the self-replicating AI story that has been spreading across the internet at alarming speed