Considering the news from the next thread https://www.whitehouse.gov/briefing-room/statements-releases/2023/10/30/fact-sheet-president-biden-issues-executive-order-on-safe-secure-and-trustworthy-artificial-intelligence/, this really puts things into perspective..a rather sad perspective.
this post was submitted on 31 Oct 2023
1 points (100.0% liked)
LocalLLaMA
3 readers
1 users here now
Community to discuss about Llama, the family of large language models created by Meta AI.
founded 1 year ago
MODERATORS
Oldest trick in the book. Once you've established yourself as a market leader, regulations strongly favour you over smaller competitors since you're better prepared to comply with them, especially if you were involved in shaping them.You'll gladly trade slightly reduced profit margins for prolonged market dominance.
Thus, it makes sense to trigger regulations, even if it requires you to embellish the truth.
It won’t work. US regulations that don’t make sense will just shift the interesting product development overseas. As has happened in every industry up to now…
It'll work enough for it to be worth it. People don't like jumping through hoops and often end up just going with what's easily available. Enthusiasts will download Chinese LLMs but a large American corp? Not going to happen. Truth is, people like us are barely a tiny blip on the radar of commercial enterprises.
They're willing to sabotage everyone because they won't be sabotaging themselves that much? That's messed up.
I always thought of Andrew Ng as a more recognizable name than Google Brain.
Agree it's very strange they didn't use his name
Possibly to avoid political biases/aversions.
Yes, every ML practitioner knows him but probably the prompt engineer aren't aware of his massive contributions to the field.
I mean the closest I've ever come to being a prompt engineer is using Sillytavern to jack off and I've heard of Andrew Ng.
Yup, the guy is a legend.
I only read the article because I recognized Andrew Ng in the photo.
Real insiders know that arrogance is measured in nano-Ngs. He has a bit of a sordid reputation with people that have worked with him.
Tell me what you knoww
I'd be curious to hear this as well, only seen him in webinars and stuff
That article is on businessinsider.com, its readers know neither Andrew Ng, nor Google Brain, but they all know Google, and Google Brain has "Google" in it (and Andrew Ng does not).
I assumed that this was the case since Altman started moaning about dangers of AI waay back.
Moaning about it while still developing SOTA models
Does the argument not make sense? Why not first evaluate the arguments for not open sourcing models on the face instead of reaching for people’s personal incentives to lie about it? Seems like people forgot step one and just went to the assumption of mal intent, like you said you did.
Given that we don’t really know how AI can be used for malicious purposes, might it make sense that the org with by-far the most powerful model chooses not to release their secrets, as to slow the pace of malicious use?
Is it possible that Altman believes this, or does his incentive to lie about it so greatly outweigh anything else that you can’t even consider the merits of the argument? I hear way too much about why OAI must be lying about this, not enough considering what they have to say.
That's exactly what an AI would say!
Capitalism = free-market competition
Socialism = public monopoly (the state, or a private monopoly)
These AI companies want to put us under a socialist regime. We must stop them.
Socialism = public monopoly (the state, or a private monopoly)
How is a private monopoly a public monopoly?
A company can achieve such high levels of productivity that it no longer relies on the profit motive, whether it's privately or publicly owned. When I refer to a "private monopoly," I'm describing a private enterprise that has effectively transitioned into a publicly regulated entity with a focus on improving societal ills.
reminds me of the DMV .. which is a flaming pile of dogshit and treats people like shit because they can do whatever they want and still be in business.. you keep your socialism
interesting!
I honestly think that if there is a chance of AI wiping out humanity we should take it seriously.
Companies are right now doing the easy thing and saying there is no danger. What I want is proof. Physicists showed proof for example that the LHC was safe.
If there is even a 0.1% chance of an AI taking over we should be serious about it. Alternatively imagine if I said the following: I am genetically engineering monkeys to be super smart way smarter than humans. I am also going to give them the tools to improve themselves further and to reproduce near instantly then I'm going to release them for 5.99 a month as servants to humans. Hoe many of you would be worried.
Some companies are doing the easy thing and saying there’s no danger, and then there’s others that are being honest about the risks they see over the horizon which causes moronic threads like this. There’s no winning… people are too damn cynical
I'd watch. Movie or episodic format?
People barely take climate change seriously and we're fairly confident it's going to wipe us out.
In the pursuit of getting rich, transparency and responsibility often take a back seat.
👏👏👏
The bad thing is, just because Andrew Ng states this, doesn't make it true or the possibility of dangers less relevant. There are people outside of big businesses like Hinton who also warn, even though he is not part of "big companies" anymore.
Also, what is all of this about? In the end there are multiple scenarios in which ways AI can harm society. It probably won't be about Terminator rising. On the other hand precausions revolve around the fact that we actually don't really know, because this technology is so new.
I also don't think that "big companies" like OpenAI even need to shut down smaller businesses, because - as Sam Altman stated - incoming money really isn't any issue for them at all. They are drowning in money.
While there are certainly people who only care for money and other kinds of status symbols, I still believe that many people working within those companies actually try to be truthful about their work as individuals.
How could they be "lying" about risks? Risks aren't facts, they're statements of probability. Big tech companies may benefit from propagating the existence of risk, but they can't be "lying" about risks unless there's some scientific study showing that the risk doesn't exist.
In other news: the water is wet.
I am not concerned about the AI itself. Rather, it is about who instills "the rules" into that AI. Are they going to be Asimovian or Robocopic?
Knowing human history, I am not optimistic.
Doesn’t basically everyone know who Andrew Ng is? Or at least change google brain to Coursera’s founder maybe..?
I'm so baffled this has not been realised by people before, it's so obvious and it's not the first time in history it happens either.
First of all Max Tegmark, it's not even the slightest suspicious that his "non profit" organisation received millions of donations from Elon Musk? I have not figured out what Elons stake in this is yet but I have absolutely no doubt in my mind it's economical, basically everything he ever did and said has been to manipulate the Stockmarket etc, I doubt that changed recently.
Then you have OpenAI, that first and formost is everything but Open and very ProprietaryAI nowadays, and what seriously annoys me is that OpenAI in particular been "teasing" about "AGI in n days" etc on several occasions for what purpose if not to manipulate expectations and investors, yet they are one of the most driving in this matter- are people really that stupid that they can't put together 1 and 1?
I mean it's not a complete lie, but definitely over exaggerated!
There much more subtle and urgent risks to mitigate, which haven't been caused by AI -- though the AI has the potential to accelerate/widen them.
Governments and authorities need to be very careful with regulations, let alone over regulating -- because the trust of people in governments is very low. Making it really hard for people to follow and for organizations to adhere.
Minding the fact that the world is not aligned on almost anything, other countries will for sure take advantage of the situation, offer almost regulation-free zone for AI development -- so they attract as many experts and businesses as possible.
On a personal note, I'm following the EU AI regulations closely, to see where it goes -- and in case the EU ended up forcing heavy regulations that slow down the development. Then I'm one of the people who will look for other place to move to, where I can get access the latests with minimum-to-no concerns.
No shit, they can loophole around any regulation with their gazillions, it's everyone else that gets fucked.
Young Man! OpenAI and friends
Honestly would make a good skynet origin story for the next terminator