The pursuit of everlasting growth that pg describes inevitably results in cheating. At a certain point the market reaches a saturation point and if you're capitalising on every possible opportunity to retain your high growth rate, that will include hoarding resources and circumventing consumer protections.
The hoarding resources and circumventing consumer protections is just the start of it.
Eventually it becomes rational to start buying politicians, and subsequently laws. The next obvious avenue is to then control entire government agencies like the FAA or the FCC and just write favorable laws and regulations they don't even have to circumvent.
But even that isn't the end because they're growing too fast, they actually outgrow the law, so breaking it becomes a rational, profit-driven choice. Huge fines? Regulators breathing down your neck? No worries! Just spend more money then has ever been spent in an election to their favored presidential candidate, and then they get to just shut down investigations into themselves!
But even that isn't enough -- soon it becomes a rational business-forward goal to take over the entire government; or even better become the government. First a city, then a state, then a nation. Guess what folks EVEN THAT won't be enough. Not even everything on this entire planet Earth is enough for them; they also want the Moon and Mars and the entire solar system. They will have to become God at some point for this growth to keep up, and that will still be too little for their egos to bear. Something has to give.
Oh, this is already the case. As they say, the future is already here, it's just not evenly distributed yet.
What's broken is accountability. No one will hold those in power accountable, so instead they mutate into criminals because they can. Then they attract the rest of the criminals and it becomes a critical mass of maniacs, kakistocracy.
> Although a large body of research finds that workers want to work remotely, our findings suggest that workers may not realize the costs of remote work for their well-being
Sentences like this just make my eyes roll. 'Workers' have agency to make judgement on what has a positive and what has a negative effect on their well being.
Personally not having to commute led to me being able to attend meetup groups in the evenings where I formed an amazing group of friends and met my current partner. It had such a massively positive impact on my well-being.
Human review is about learning and there's an implied social contract in that someone is giving you their time to make you better. It isn't necessarily necessary but replacing it with AI shows a fundamental misunderstanding of why it is part of the process.
Watching this from outside the US, what gets me is not that corruption exists (shocker!) but how brazen they are with doing all of this in the open. Somehow it seems to be working in their favour as it appears to be normalising such behaviour.
This administration has certainly committed brazen acts of corruption, but this case looks like a more garden variety style "it's who you know" situation.
Don Jr. had invested two years ago, it says, and could deny he did so knowing his father would have influence on Pentagon loans at the time. The firm can be touted as "protecting national security" so the loan itself is justifiable.
It's all spun as "smart business thinking," and "everyone else does it," etc. And that's also not entirely untrue. If any Democrat politician were to make noise about it, the Republicans will probably go, Well, what about Hunter (Biden)? The libs will have been owned, mission accomplished.
Governments cater to corporations rather than to people. It's an unpopular opinion on here since so many people profit from the industry, however the fact is AI and data centres are seen incredibly negatively by the majority of society and the rhetoric and lobbying practices of the companies pushing for this new age are the reason.
What does this show that we didn't know already? LLMs cannot provide accurate answers to questions where data is not included in their training sets. This doesn't appear to have much substance
LLMs can and will provide inaccurate answers to questions where data is included in their training sets too, that's in the nature of neural networks. It's just less likely that when the data is not in the training set...
I mean that's true but I don't think that's realistically what's going on when one model gives an unqualified "Yes" and the other gives an unqualified "no."
You can argue the study isn't as case-closed-decisive as we'd ideally like, but it's certainly evidence. It's probably hard to design a better study.
What are you talking about? The models were not ALLOWED to have confidence (or the lack thereof). They were explicitly told to give a single label, and in most cases, all of them were correct depending on additional context they would surely have provided, especially with access to the internet (which some didn't have). This is just silly.
This argument of "if we don't do it, someone else will" to justify theft is so tiring. The companies doing the stealing are collectively the same ones that have power to prevent it, if they were incentivised to do so.
Also, the following line is absolutely laughable.
> Fortunately Putin is neither as war hungry nor as politically desperate as Starmer.
reply