Counteroffer; how about a 10 hour day for 8 hours pay, plus 2 hours of overtime so you can pay off the loan on the robot. It’s basically the same only less joy and more profit.
Counteroffer; how about a 10 hour day for 8 hours pay, plus 2 hours of overtime so you can pay off the loan on the robot. It’s basically the same only less joy and more profit.
PR isn’t Labour policy (the members have sometimes supported it, but it’s not in the manifesto).
It’s fine to have hard lines, but if they exclude all your potential partners you’re just excluding yourself from government and handing power to more pragmatic parties.
Then he’s not serious about wanting proportional representation. Coalitions are inevitable with PR (and probably a good thing IMO), and you’re never going to have the perfect coalition partner.
100% How dare they suggest his hypnosis business was all an act.
It’s exactly like it sounds. He’s a former hypnotist (given those results maybe not so former) who advertised breast enlargement as one of his services. According the Sun it worked! www.thesun.co.uk/fabulous/798...
I knew the broad story, but I had no idea it had been officially verified by “32B Sun girl KASIE DAVIES”. This changes everything!
That’s assuming the Corbana party: a) Can decide on a name and policy platform b) Can hold together more than 2 MPs who agree with that platform
15,000 students claim asylum per year. That’s a tiny number, and by definition they’re relatively high skill, relatively wealthy and they’ve already poured £££ into the British economy to study. It’s like the gov are running a competition to find the most harmful single policy decision.
There is, and it’s pretty easy. They’re also only “fines”. You basically get charged for the longest journey you could make from where you enter the network. In most cases it’s about £10.
For anything that involves hot water, the energy cost is mostly heating up the water. 4-hr eco cycles use cooler water. More spraying, but compared to heating up the water it’s still less energy. The “quick” cycle (and commercial dishwashers) uses much hotter water, which cleans and dries quicker.
He runs a US think tank now, and the bar is much much higher over there (depending on the topic)
Wow. We were seriously considering using nx in an upcoming project. Absolutely no fucking way we’re touching it now. I’ll try and make a case to outright block the package.
They know that their jobs are currently dependent on 100% toadying loyalty to Trump. Quite a lot of them were chosen by Vance and Thiel though, they’ll switch as soon as they need to.
I agree. He’s far more dangerous ideologically. But, maybe it’s inevitable that him (or someone like him) becomes president next anyway. And maybe his weakness is that he doesn’t have Trump’s magic with the base, so having to take over suddenly ends up better than a planned transfer?
1. I wish we spent half this effort writing documentation for humans 2. The hard part of the job isn’t completing one task. It’s learning from that task until you can apply those lessons to a completely new task, often with no obvious link beyond “this seems like the right direction”.
Like a junior developer you fire and rehire at the end of every day. My team is going through cycles of writing “instructions.md” files detailed enough to complete the tasks we’re trying to give to AI. Every new mistake the AI makes is a new section added to the file. It’s pissing me off because:
Would I have to pay 2011 prices on the outgoing train ticket, but 2025 prices for the return? Or would I need a very long expiry season ticket in order for it to be valid for my entire journey?
He’ll provide an entertaining and provocative way to fill a slot, but is disciplined enough not cause you problems. He guarantees News, but not OfCom. You could put the LibDem immigration on, but viewers won’t care. You could put Tommy on, but you’ll be in crisis meetings the rest of the week.
If Reform’s supporters want to deport the “bad people” then they need to be willing to prove to the state that they’re “good people”. Over and over again. It makes the conspiracy theories about Covid-passes and “15 minute cities” look utopian.
It’s more than just “could”. They’re talking about deporting 1% of the population. That’s not possible without doing some very authoritarian things that will hurt everyone.
* If a future Reform gov does this, then it’s all they’ll get done in their term. Just the process of leaving the ECHR and renegotiating everything that depends on that would take up all the energy of parliament. It’ll be like the worst days of the Brexit quagmire.
* A lot of his targets currently work in the NHS. Where’s the plan for hiring and training British citizens to replace them?
* It’ll tank the economy. Most immigrants work, pay taxes and spend in the economy. It’s easier to deport law abiding and active people (they interact more with the state and have more options), so they’ll be the first to leave. That’s tax revenue, consumer spending, growth walking out the door.
* It’ll require ID cards and intrusive surveillance. If you thought Covid-passes were bad you’re not going to like what Farage has planned. Deporting all the “bad people” means repeatedly proving to the state that you’re a “good person”. Good luck if you can’t do that.
People need to understand: * This won’t just involve deporting the “bad people” that Farage talks about. There aren’t 600k of them anyway, and they’re hard to find. It’ll be people you know, like, and rely on in order to meet the quota. Look at the US to see how that goes.
Farage needs to be clear with his supporters what would actually be required for his plan. You can’t deport 1% of the population without any effect on the 99%.
The best anyone has done so far (IMO) is Dead Ringers: “no no no, let me finish”. The hypocrisy is the target. The silenced outsider who’s on the news every day, the commodities trader champion of the working class, the public school educated anti-establishment hero.
The bits of Trump that Newsome has been able to satirise are where Trump has some separation from his base. None of the MAGA base talk like that, that’s pure Trump. Most of Farage’s shtick is him imitating his base. Hard to satirise that without alienating the people you want to win over.
There are wealth taxes I’d support, but it’s *for* their distorting effects not despite them. A land value tax would encourage better use of land and raise a bit of cash. That’s a wealth tax, I would probably end up paying more because it, but it would have a positive effect on society.
But is still convinced that there’s untold billions to be claimed off the truly wealthy. See also council tax rebalancing.
But, 1. Only if it’s a one-off, so doesn’t help with day-day spending. If we make it permanent it would raise dramatically less. 2. Raising that much takes a 1% annual tax on all wealth over £500k (including property and pensions).
A wealth tax raising £25-50bn would have to go far far beyond what people think of as wealthy. The Wealth Tax Commission thought that they could raise £260bn over 5 years, so that would do the job.
I bet that decision costs you several weeks somewhere down the line.
MS takes that promise of future revenue and spends $80bn on data centres in 2025, including on nvidia GPUs. nvidia becomes a $4tn company on the promise of future GPU purchases. That’s taken as evidence of the future of AI, which justifies pumping >$40bn a year into OpenAI, and round we go again
Me too. I think there’s something messed up in the financial system where OpenAI is a loop of investment opportunities, but AlphaFold is just valuable in its own right. OpenAI needs $40bn per year to survive, which is spends with MS on compute and promises to spend more in the future.
You’d think that would make it less valuable. A car with mileage of 10mpg should be less valuable than a car that does 100mpg. Something’s broken in the system though, in this case the high compute cost of LLMs is creating their value because it creates other investment opportunities around them.
ChatGPT condenses knowledge to something manageable and talk to us about it. Which is also great, I’d argue less important, but it’s still an amazing idea. Significantly, ChatGPT can only work with enormous amounts of investment. To grow it needs more data centres than we’ve ever built before.
AlphaFold can invent new medicines from compounds that have never been seen in nature, but by predicting how they’ll fold and interact with the body. That’s phenomenal, but weirdly not an investment opportunity so not seen as particularly valuable.
It says a lot about society that the reaction to AlphaFold was a couple of excited news articles and a tiny amount of funding. Another company invented a chatbot that can talk like us and we reorganised the entire tech industry around it.
If the roles were flipped he’d have a £mil GoFundMe already and be planning his legal case from beside a pool in Spain.
You’d probably need the peacekeeping troops either way? Depends how far the “defeat & humiliate” goes I guess, but most likely it doesn’t change the need for security guarantees along the border.
For once I was glad to hear it. I’m fine if he talks out the clock about the media, mail-in voting and the DC wasteland.
Are a lot of these journalists called Brian or something?
You can take the comedian out of the club, but…
It’s self-indulgent, but if it means he talks out the clock about whoever he had dinner with and his dystopian fantasy about DC then it’s good for my blood pressure.
No criticism of the reporters, they’re doing their jobs (most of them at least). Such a high stakes way to stage it though. Get the press in afterwards, have them ask questions about what was agreed instead of trying to find ways to start a fight.
Why the fuck do they start these things with questions from the press? The most delicate diplomatic and ethical balance and they set the scene with a shouting crowd of people financially incentivised to light the fireworks.
I can believe there’s a genuine $50bn just in coding assistants, document summaries and low level customer support. The bubble turned that into a multi-trillion industry, and the only path to profitability for an industry that large is genuine AGI, which they don’t have and won’t get from LLMs.
Something like “The LibDems want a better electoral system, and that PR system often involves coalitions. We want a government that best represents the values of our voters. Ideally that’s a LibDem government, but we’ll do whatever we need to do to bring LibDem values to this country.”
@eddavey.libdems.org.uk - please consider a more detailed answer to the question about potential coalition. Exactly because “coming second place is not good enough”. If the LibDems won’t go into coalition with Labour to avoid a Reform gov then I can’t risk voting for you.
Outside of my immediate family there is only one person where I’m confident about their voting intentions. They’re the local MP. I guess probably two if you include their partner, but you never know.
There is some value there. We use it at work, and if I stick to the use cases where I find it useful then I think it makes me about 5% more productive. A 5% productivity boost to some jobs is HUGE totalled up over the world economy. But apparently not huge enough, or exciting enough.
I’ll be very happy if next time I can use the infra that porn built to provide the estate agent with proof that I exist, am UK resident and can (just) afford the rent, but without having to give them any actual information they need to keep secure. See also: phone companies, banks, employers…
Last time I rented a flat I had to give the estate agent (who, being blunt, I trust less than a pornographer) bank statements, address history, passport scans, immigration status… I’m pretty sure they just dumped it all in a shared folder for the whole branch.
The site gets proof of age, but doesn’t know anything about you. Gov uk knows everything about you (already) but doesn’t know what sites you’ve been visiting. Age verification is just the start, I’m actually more interested in other aspects of my identity.
Gov uk does the identify checks and gives you a token saying “Over 18”. It can only be used by you and for the request that started the process. Crucially, it doesn’t reveal your name, address, age. Just “Over 18”.
If it’s done right then that “token” concept goes both ways. You present the gov uk service with a token saying “I need to prove I’m over 18”. It doesn’t reveal to the government what you want to do, just what you need to prove. It could be porn, it could be buying alcohol, Bluesky DMs, etc.
> I should be able to use my gov uk account to provide a simple and easy token of my age, without giving up any information other than “yes, he really is that old”. This is the value that the private sector can’t provide, because they’ll always have the incentive to collect as much data as possible
One of the things that annoys me AI in 2025 is that a $50bn industry is great. If I’d started a company and built a $50bn industry around it in 3 years I’d be pretty fucking happy with that. That is a success story. It’s not enough for this lot though. They have to hype it up into a bubble.
No of course not. Why would you assume that?
And they accepted the risk of that arrest in order to protest the decision to proscribe PA. The question isn’t whether it’s illegal or legal. It’s whether or not it was morally right, and that’s not the same thing.
There are no sensible solutions to this while PA and ISIS are in the same category. Take PA off the list and prosecute them for criminal damage, assault, sabotage, whatever we can make a case for.
Who is claiming Palestine Action are non-violent? They’re very clearly a violent direct action group. The people arrested at the weekend were clearly non-violent though. The guy carrying the Private Eye cartoon wasn’t violent.
So for now it’s easier to fund an AI pilot than a new hire. If the pilot doesn’t work out then we can always hire next year instead.
Where I’ve seen damage to the hiring rates (and the job market is rough at the moment) that’s from uncertainty. It’s not so much that AI is replacing all software devs, it’s that it might replace some of them. No one’s sure how many, or which ones.
I think it’s more an established human mistake that skilled work looks simple from a distance. For example, policy making looks pretty simple to me. I’ve got ideas, I could probably write those up in a doc, link some sources, publish it somewhere. I bet it’s not that easy in reality though?
In practice the actual code is a fairly small part of either job (and the AI isn’t even *that* good at the code part).
Backend devs say it’ll do the UI stuff because that’s just boilerplate and standard colours, but never their work because they need to really understand the architecture. UI devs say it can’t do their work because they have to really understand the users, but it’ll do the plumbing work fine.
Funnily I see the same argument made within coding circles. AI will fully automate *that* sort of coding, but not it’s not good enough for my specialism. I’m sure the AI devs are thinking that too, but I don’t think that makes it true.
Maybe there’s a path to profitability there, but it’ll be a big adjustment from the current approach.
My guess is that the “deeper” thinking (particularly the “Research” style where it seems to be asking itself about the task, then multiple request steps) are spectacularly expensive. Taking that choice away from you is a way of limiting their losses and upselling a more expensive package.
Starmer can describe immigration as having caused “incalculable damage”, or not. I’m not 100% sure that Jenrick’s comments would see him expelled from the Labour Party.
Our internal copilot training included “reformat code” as an example task. When I pointed out that was built into the IDE people complained they couldn’t remember the command name and they like the “✅ code formatted” feedback from the chat…
Yeah, but now we have the option of an agent reproducing that right-click action with a 20% failure rate, BUT describing what it’s doing as it goes.
What if you’re a Spanish national convicted of treason against Spain?
Peter Thiel (PayPal, Facebook, Palantir) and Marc Andreessen (Netscape, general Silicon Valley VC). So the bad news is that they have effectively infinite money.
Whether we want to give them brain space or not, they’ve got into the brains of powerful people in the US and UK governments.
There’s something worrying about the combination of the Thiel/Andreessen money world, British Lords, “warm can, and one in the pocket” thugs and Porsche driving chads. Feels like there would have been more firewalls between those groups in the past.
It’s the drinks table that I find most disturbing. Hard to put my finger on why.
If it’s option 2 then that’s still amazing. The man who made every software dev in the world a little bit more efficient would still be a very successful man. But he wouldn’t be culturally significant in the way AI is at the moment, and it wouldn’t justify the the current crazy valuations.
If it’s option 1 then they need to get on with it. OpenAI is loosing $40bn a year. If they’re sitting on the most valuable technology in the world then they need to put up or shut up.
Altman is admitting that he’s reached the end of the LLM road. If LLMs aren’t going to be enough then either: 1. They’ve got the next thing under the table, a genuine scientific breakthrough. Or, 2. The industry needs to rescale their ambition to “making some knowledge workers 10% more efficient”
OpenAI’s valuation (and ability to raise more cash to cover their losses) depends on their potential to build a human-replacing AGI. What we’ve got so far is useful in a lot of cases, but nowhere near good enough to justify the trillions being poured into it (1% of US GDP!)
That’s vital, because that gap is a fundamental part of how LLMs work. They’re trained, but they can’t learn. It’s a one-time process. You can add context, you can have them talk to themselves, but that’s not learning. It’s just applying the same knowledge to a more complex (and expensive) problem.
Really important admission from Altman here: > I think the way that most of us define AGI, we’re still missing something quite important,... one big one is … this is not a model that continuously learns as it’s deployed from things it finds www.theguardian.com/technology/2...
US government contracts. If Trump issues a royal decree that ChatGPT isn’t sufficiently anti-woke then the only way gov workers will be able to use Copilot is if there’s a Grok option in the drop down.
We need an awkward, neurotic AI wracked with self doubt. > Oh god, I don’t know. It might be Roman numerals? D is 500 right? I’m pretty sure X is 10. So maybe 767? I’m terrible at maths though, definitely check that. Sorry.
The worst possible AI for the human psyche, made by the worst possible people. Confidently stating answers without any knowledge or understanding is the hack people like Musk and Altman used to be successful in the first place. No surprise that they don’t see anything wrong with it
And a military themed “gardening belt” for it. I sympathise a bit, I sometimes carry around some pretty scary looking tools, but I put them in a bag. Not in a holster designed to make me look like a (being generous) hunter.
Part of the point is to tax the value of the land instead of the buildings. So an acre of empty land in a town centre pays the same (quite high) LVT regardless of whether it’s a golf course or a couple of houses. In theory it encourages development instead of land-banking.
It’s the automated version of the Day Today Peter O’Hanrahanaran sketch, Grok is quoting the German finance minister as saying “Trenter percenter”. Except somehow because it’s an AI and not an employee of X we’ve decided it’s ok for it to just make stuff up.
One of my big frustrations with LLMs (all of them, although it’s more dangerous with a “no-filters” AI like Grok) is when they misrepresent their capabilities. Grok clearly says it has reviewed the audio multiple times, then admits it can’t do any such thing.
I’m not a religious man, but can we get some sort of ban on this from a blasphemy point of view?
Which is just a crazy way round to approach the question. It’s what you get from an industry where “engagement” is the only thing that has mattered for years. People are engaging with the bot about X, so we must be doing X well.
Last time I looked into one of these lists it turned out they took a list of questions people had asked ChatGPT and categorised them by profession. So someone asking ChatGPT about the past got tagged “Historian”. Someone asking about a holiday got tagged “Travel Agent”.
AI can replace Historians in that it can answer (some) questions about history. “When was the Battle of Hastings?” That’s not really what a Historian does. In the same way that a Passenger Attendant isn’t there to answer “When is the last train to London?”, or most of those professions.
Not quite as scary as it looked then. Still weird that there’s more support for deporting the primary visa holders than their families. Shows that it’s an emotional reaction rather than anything logical.
What I’m taking from that chart is that the UK wants to deport people who came here on work visas (53% support) but let their families stay (50% support). Which is weird… but it’s the will of the people I guess.
I agree that PA caused damages to military equipment. That’s an extremely serious crime (potentially even sabotage) and goes well beyond protest. It’s not terrorism though, and the consequence of the home sec. choosing to declare it terrorism has been to criminalise and suppress actual protest.