Hacker News new | past | comments | ask | show | jobs | submit | rwmj's comments login

> It shows up as well in modern parenting guidance, including long term studies claiming that parents who prioritize validation over correction produce children who end up not just more mature, confident, and self-assured, but also with much better adult relationships to those parents.

Self-reported "studies" probably. It's highly unlikely this could be tested in any rigorous way. (Not to mention the problem with what "mature, confident, and self-assured" actually means)


The BBC is a weird corporation created by royal charter. But it's not a part of the UK government, and nor are works created by the BBC copyright-free (as is the case in the US for something like NASA).

UK government publications aren't copyright-free either. In fact they manage to be worse than copyrighted, at least for works created before 1988 (some of which are perpetually copyrighted, others until 2040, others for 125 years, it's a big mess). https://en.wikipedia.org/wiki/Crown_copyright#United_Kingdom


I'm not sure about that. In these fields there are plenty of places where you need to ingest or process masses of data (eg. from a sensor in a medical device), and you're only going to sell 5 of these machines a month for 100K each, so $3000+ bill of materials for an FPGA to solve the problem makes sense.

The problem (for Intel) is that you don't sell billions of dollars of FPGAs into a mass market this way.


That's pretty wild. Is that a German employment law or something specific to your company?

It's probably an employment law. My wife's from Tunisia and over there all employment is basically contractual. If your employer lets you go they owe you some kind of fee, and I actually believe the amount might be the remainder of the contact, I'm not sure.

It's employment law.

Germans are so expensive to hire and maintain that companies have offshored German manufacturing to the United States.

(... And God bless Germany for it. Trickle-down theory doesn't work in general in capitalism but it does work in labor negotiations: every right Germans secure for themselves is a right an American company employing Germans and other countries has to abide by when doing business, and it incentivizes the company to minimize their paperwork by treating everyone to the German standard).


An employee who is serving their notice period is still an employee. Unless you mean truly ex-employees who still have access, in which case the company has a big problem if it cannot revoke credentials.

Please post the name of your company so we can be sure to avoid it.

done

I take it that this was posted in the bottom of a locked filing cabinet stuck in a disused lavatory with a sign on the door saying "Beware of the Leopard"?

My parents had a defined benefit (final salary) pension organized by their employer, of the kind that no longer exists. They didn't have to choose how much to save. My father knows how to fix 1980s era cars, but 2020s cars have software that makes them impossible for anyone except a dealer to fix (or even for the dealer in some cases).

The world changes so parental advice may need some adaptation, but the general idea is that they show their child when they make decisions. This may have been more about what job they took for what pension it came with and what car they bought for which labor price and now it is which car to never buy.

This kind of thread bothers me with it's hand wringing. Parents and teachers often spoke of their lives in casual conversation when I was growing up and it was pretty clear to me that, aside from it being a response to having no adult around to make a more useful conversation with for them, there was an element of educational intent.


A bit thin on detail, but will this require confidential VMs with encrypted GPUs? (And I wonder how long before someone cracks SEV-SNP and TDX and pirate copies escape into the wild.)

At the pace models improve, the advantage of going the dark route shouldn't really hold for long, unless I'm missing something.

Access to proprietary training data: Search, YouTube, Google Books might give some moat.

We have Common Crawl, which is also scraped web data for training LLMs, provided for free by a non-profit.

The Common Crawl is going to become increasingly contaminated with LLM output and training data that is more likely to have less LLM output will become more valuable.

I see this misconception all the time. Filtering out LLM slop is not much different than filtering out human slop. If anything, LLM generated output is of higher quality that a lot of human written text you'd randomly find on the internet. It's no coincidence that state-of-art LLMs increasingly use more and more synthetic data generated by LLMs themselves. So, no, just because training data was produced by a human doesn't make it inherently more valuable; the only thing that matters is the quality of the data, and the Internet is full of garbage which you need to filter out one way or another.

But the signals used to filter out human garbage are not the same the signals that would be needed to filter LLM garbage. LLMs generate texts that look high-quality at a glance, but might be factually inaccurate. For example, an LLM can generate a codebase that is well-formatted, contains docstrings, comments, maybe even tests; but it will use a non-existent library or be logically incorrect.

LLM output is uniquely harmful because LLMs trained on LLM output are subject to model collapse

https://www.nature.com/articles/s41586-024-07566-y


Problem with filtering is that LLMs can generate few orders of magnitude more slop than humans.

Are the differences between Google Books and LibGen documented anywhere? I believe most models outside of Google are trained on the latter.

The number of folks that have the hardware at home to run it is going to be very low and the risk of companies for leaking it is gonna make it unlikely IMHO.

I think home users would be the least of their concerns.

It only takes one company to leak it

Or one company to get hacked and the hackers leak it

Realistically the only people able to run models of this size are large enterprises.

Those enterprises won’t take the risk of being sued for using a model without proper permission.


I don't know – if there's still dumb money being thrown towards AI in non-tech and non-privacy-heavy industries, especially ones traditionally targeted by ransomware, there'll always be a chance of datasets getting leaked. I'm thinking retail and consumer product-oriented companies. (There's always non-Western governments without strong security orgs, too.)

Nations.

or large government sponsored entities like Mossad. Air gapping won't protect against spying. Good luck trying to sue them

They can get "hacked" and wooops.

> I wonder how long before someone cracks SEV-SNP

https://bughunters.google.com/blog/5424842357473280/zen-and-...


I'd expect watermarked model weights plus a lot of liability to distinctivise leaking the model.

Steve Vickers was my university lecturer circa 1995. His course was legendarily difficult, in contrast to his lucid tutorials in both the Jupiter Ace and ZX81 manuals! The list of his papers on Wikipedia should give you a flavour of what the course was about: https://en.wikipedia.org/wiki/Steve_Vickers_(computer_scient...

You could get an Orient Kamasu which has day/date complication and is probably 1/100th the price of the Rolex.

Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: