To expound, this just seems like a power grab to me, to "lock in" the lead and keep AI controlled by a small number of corporations that can afford to license and operate the technologies. Obviously, this will create a critical nexus of control for a small number of well connected and well heeled investors and is to be avoided at all costs.
It's also deeply troubling that regulatory capture is such an issue these days as well, so putting a government entity in front of the use and existence of this technology is a double whammy — it's not simply about innovation.
The current generation of AIs are "scary" to the uninitiated because they are uncanny valley material, but beyond impersonation they don't show the novel intelligence of a GPI... yet. It seems like OpenAI/Microsoft is doing a LOT of theater to try to build a regulatory lock in on their short term technology advantage. It's a smart strategy, and I think Congress will fall for it.
But goodness gracious we need to be going in the EXACT OPPOSITE direction — open source "core inspectable" AIs that millions of people can examine and tear apart, including and ESPECIALLY the training data and processes that create them.
And if you think this isn't an issue, I wrote this post an hour or two before I managed to take it live because Comcast went out at my house, and we have no viable alternative competitors in my area. We're about to do the same thing with AI, but instead of Internet access it's future digital brains that can control all aspects of a society.
Although I assume if he’s speaking on AI they actually intend on considering his thoughts more seriously than I suggest.
If this is so, and given the concrete examples of cheap derived models learning from the first movers and rapidly (and did I mention cheaply) closing the gap to this peak, the optimal self-serving corporate play is to invite regulation.
After the legislative moats go up, it is once again about who has the biggest legal team ...
Relevantish: https://arxiv.org/abs/2301.00774
The fact that we can reach those levels of sparseness with pruning also indicates that we're not doing a very good job of generating the initial network conditions.
Being able to come up with trainable initial settings for sparse networks across different topologies is hard, but given that we've had a degree of success with pre-trained networks, pre-training and pre-pruning might also allow for sparse networks with minimally compromised learning capabilities.
If it's possible to pre-train composable network modules, it might also be feasible to define trainable sparse networks with significantly relaxed topological constraints.
Shouldn't we be evaluating ideas on the merits and not categorically rejecting (or endorsing) them based on who said them?
The problem is when only the entrenched industry players & legislators have a voice, there are many ideas & perspectives that are simply not heard or considered. Industrial groups have a long history of using regulations to entrench their positions & to stifle competition...creating a "barrier to entry" as they say. Going beyond that, industrial groups have shaped public perception & the regulatory apparatus to effectively create a company store, where the only solutions to some problem effectively (or sometimes legally) must go through a small set of large companies.
This concern is especially prescient now, as these technologies are unprecedentedly disruptive to many industries & private life. Using worst case scenario fear mongering as a justification to regulate the extreme majority of usage that will not come close to these fears, is disingenuous & almost always an overreach of governance.
Ideas that drive governing decisions should be globally good - meaning there should be more than just @sama espousing them.
You'll be stuck in the muck while they're laughing their ass off all the way to the bank.
AI is beyond-borders, and thus unenforceable in practicality.
The top-minds-of-AI are a group that cannot be regulated.
-
AI isnt about the industries it shall disrupt ; AI is the policy-makers it will expose.
THAT is what they are afraid of.
--
I have been able to do financial lenses into organizations that even with rudimentary BI would have taken me months/weeks - but I have been able to find insights which took me minutes.
AI regulation right now, in this infancy, is about damage control.
---
Its the same as the legal weed market. You think BAIN Capital just all of a sudden decided to jump into the market without setting up their spigot?
Do you think that haliburton under cheney was able to setup their supply chains without cheney as head of KBR/Hali/CIA/etc...
Yeah, this is the same play ; AI is going to be squashed until they can use it to profit over you.
Have you watched ANIME ever? Yeah... its here now.
AI is following the path of Web3
And here we are - Sam trying to lock down his first mover advantage with the boot heel of the state for profit. It's fucking disgusting.
> You either die a hero, or live long enough to become the villain
Sam Altman has made the full character arc
Exactly. Came here to say pretty much the same thing.
This is the antithesis of what we need. As AI develops, it's imperative that AI be something that is open and available to everyone, so all of humanity can benefit from it. The extent to which technology tends to exacerbate concentration of power is bad enough as it is - the last thing we need is more regulation intended to make that effect even stronger.
True open source AI also strikes me as prerequisite for fair use of original works in training data. I hope Congress asks ClosedAI to explain what’s up with all that profiting off copyrighted material first before even considering the answer.
What people also fail to understand is that AI is largely seen by the military industrial complex as a weapon to control culture and influence. The most obvious risk of AI — the risk of manipulating human behavior towards favored ends — has been shown to be quite effective right out the gate. So, the back channel conversation has to be to put it under regulation because of it's weaponization potential, especially considering the difficulty in identifying anyone (which of course is exactly what Elon is doing with X 2.0 — it's a KYC id platform to deal with this exact issue with a 220M user 40B head start).
I mean, the dead internet theory is turning true, and half the traffic on the Web is already bot driven. Imagine when it's 99%, as proliferation of this technology will inevitably generate simply for the economics.
Starting with open source is the only way to get enough people looking at the products to create any meaningful oversight, but I fear the weaponization fears will mean that everything is locked away in license clouds with politically influential regulatory boards simply on the proliferation arguments. Think of all the AI technologists who won't be versed in this technology unless they work at a "licensed company" as well — this is going to make the smaller population of the West much less influential in the AI arms race, which is already underway.
To me, it's clear that nobody in Silicon Valley or the Hill has learned a damn thing from the prosecution of hackers and the subsequent bloodbath of cybersecurity as a result of the exact same kinds of behavior back in the early to mid-2000s. We ended up driving out best and brightest into the grey and black areas of infosec and security, instead of out in the open running companies where they belong. This move would do almost the exact same thing to AI, though I think you have to be a tad of an Asimov or Bradbury fan to see it right now.
I don't know, that's just how I see it, but I'm still forming my opinions. LOVE LOVE LOVE your comment though. Spot on.
Relevant articles:
https://www.independent.co.uk/tech/internet-bots-web-traffic...
https://theconversation.com/ai-can-now-learn-to-manipulate-h....
Could you share the minutes from the Military Industrial Complex strategy meetings this was discussed at. Thanks.
If you actually watch the entire session, Altman does address that and recommend to Congress that regulations 1) not be applied to small startups, individual researchers, or open source, and 2) that they not be done in such a way as to lock in a few big vendors. Some of the Senators on the panel also expressed concern about #2.
how will that work? Isn't OpenAI itself a small startup? I don't see how they can regulate AI at all. Sure, the resources required to push the limits are high right now but hardware is constantly improving and getting cheaper. I can take the GPUs out of my kids computers and start doing fairly serious AI work myself. Do i need a license? The cat is out of the bag, there's no stopping it now.
Compute continues to get cheaper and cheaper. We have not hit the physics wall yet on that.
That and if someone cracks efficient distributed training in a swarm type configuration then you could train models Seti@Home style. Lots of people would be happy to leave a gaming PC on to help create open source LLMs. The data requirements might be big but I just got gigabit fiber installed in my house so that barrier is vanishing too.
- The next most scary uses are by governments against the people of other countries.
- After that, corporate use of ai against their employees and customers is also terrifying;
- next, the potential for individuals or small organizations seeking to use it for something terrorism-related. Eg, 3d printers or a lab + an ai researcher who helps you make dangerous things I suppose
- near the bottom of the noteworthy list is probably crime. eg, hacking, blackmail, gaslighting, etc
These problems will probably all come up in a big way over the next decade; but limiting ai research to the government and their lackeys? That's extremely terrifying. To prevent the least scary problems, we're jumping into the scariest pool with both feet
Look at how China has been using AI for the last 5-10 years: millions of facial recognition cameras, a scary police force, and a social credit system. In 10-20 years, how much more sophisticated will this be? If the people wanted to rebel, how on Earth will they?
Hell, with generative ai, a sophisticated enough future state could actually make the Dead Internet Theory a reality
That's the future of ai: a personal, automated boot stomping on everybody individually and collectively forever, with no ability to resist
He is playing the game, this guy ambition is colossal, I don't blame him, but we should not give him too much power.
Democracy will not withstand AI when it's fully developed. Let me offer a better written explanation of my general views than I could ever muster up for a comment on HN in the form of a quote from an article by Dr. Thorsten Thiel (Head of the Research Group "Democracy and Digitaliziation" at the Weizenbaum Institute for the Networked Society):
> The debate on AI’s impact on the public sphere is currently the one most prominent and familiar to a general audience. It is also directly connected to long-running debates on the structural transformation of the digital public sphere. The digital transformation has already paved the way for the rise of social networks that, among other things, have intensified the personalization of news consumption and broken down barriers between private and public conversations. Such developments are often thought to be responsible for echo-chamber or filter-bubble effects, which in turn are portrayed as root causes of the intensified political polarization in democracies all over the world. Although empirical research on filter bubbles, echo chambers, and societal polarization has convincingly shown that the effects are grossly overestimated and that many non-technology-related reasons better explain the democratic retreat, the spread of AI applications is often expected to revive the direct link between technological developments and democracy-endangering societal fragmentation.
> The assumption here is that AI will massively enhance the possibilities for analyzing and steering public discourses and/or intensify the automated compartmentalizing of will formation. The argument goes that the strengths of today's AI applications lie in the ability to observe and analyze enormous amounts of communication and information in real time, to detect patterns and to allow for instant and often invisible reactions. In a world of communicative abundance, automated content moderation is a necessity, and commercial as well as political pressures further effectuate that digital tools are created to oversee and intervene in communication streams. Control possibilities are distributed between users, moderators, platforms, commercial actors and states, but all these developments push toward automation (although they are highly asymmetrically distributed). Therefore, AI is baked into the backend of all communications and becomes a subtle yet enormously powerful structuring force.
> The risk emerging from this development is twofold. On the one hand, there can be malicious actors who use these new possibilities to manipulate citizens on a massive scale. The Cambridge Analytica scandal comes to mind as an attempt to read and steer political discourses (see next section on electoral interference). The other risk lies in a changing relationship between public and private corporations. Private powers are becoming increasingly involved in political questions and their capacity to exert opaque influences over political processes has been growing for structural and technological reasons. Furthermore, the reshaping of the public sphere via private business models has been catapulted forward by the changing economic rationality of digital societies such as the development of the attention economy. Private entities grow stronger and become less accountable to public authorities; a development that is accelerated by the endorsement of AI applications which create dependencies and allow for opacity at the same time. The ‘politicization’ of surveillance capitalism lies in its tendency, as Shoshana Zuboff has argued, to not only be ever more invasive and encompassing but also to use the data gathered to predict, modify, and control the behavior of individuals. AI technologies are an integral part in this ‘politicization’ of surveillance capitalism, since they allow for the fulfilment of these aspirations. Yet at the same time, AI also insulates the companies developing and deploying it from public scrutiny through network effects on the one hand and opacity on the other. AI relies on massive amounts of data and has high upfront costs (for example, the talent required to develop it, and the energy consumed by the giant platforms on which it operates), but once established, it is very hard to tame through competitive markets. Although applications can be developed by many sides and for many purposes, the underlying AI infrastructure is rather centralized and hard to reproduce. As in other platform markets, the dominant players are those able to keep a tight grip on the most important resources (models and data) and to benefit from every individual or corporate user. Therefore, we can already see that AI development tightens the grip of today’s internet giants even further. Public powers are expected to make increasing use of AI applications and therefore become ever more dependent on the actors that are able to provide the best infrastructure, although this infrastructure, for commercial and technical reasons, is largely opaque.
> The developments sketched out above – the heightened manipulability of public discourse and the fortification of private powers – feed into each other, with the likely result that many of the deficiencies already visible in today’s digital public spheres will only grow. It is very hard to estimate whether these developments can be counteracted by state action, although a regulatory discourse has kicked in and the assumption that digital matters elude the grasp of state regulation has often been proven wrong in the history of networked communication. Another possibility would be a creative appropriation of AI applications through users whose democratic potential outweighs its democratic risks thus enabling the rise of differently structured, more empowering and inclusive public spaces. This is the hope of many of the more utopian variants of AI and of the public sphere literature, according to which AI-based technologies bear the potential of granting individuals the power to navigate complex, information-rich environments and allowing for coordinated action and effective oversight (e.g. Burgess, Zarkadakis).
Source: https://us.boell.org/en/2022/01/06/artificial-intelligence-a...
Social bots and deep fakes will be so good so quickly — the primary technologies being talked about in terms of how Democracy can survive — I doubt there will be another election without extensive use of these technologies in a true plethora of capacities from influence marketing to outright destabilization campaigns. I'm not sure what Government can deal with a threat like that, but I suspect the recent push to revise tax systems and create a single global standard for multinational taxation recently the subject of an excellent talk at the WEF are more than tangentially related to the AI debate.
So, is it a transformational technology that will liberate mankind of a nuclear bomb? Because ultimately, this is the question in my mind.
Excellent comment, and I agree with your sentiment. I just don't think concentrating control of the technology before it's really developed is wise or prudent.
If you're not at the table, you're on the menu.
> We need to MAKE SURE that AI as a technology ISN'T controlled by a small number of powerful corporations with connections to governments.
This, absolutely this. I am really concerned about his motives in this case. AI has massive potential to improve the world. I find it highly suspicious that an exec at one of the lead companies in AI right now wants to lock it up. (Ever read the intro to Max Tegmarks book?)
A better idea is to regulate around the edges: transparency about the data used to train, regulate the use of copyrighted training data and what that means for the copyright of content produced by the AI, that sort of stuff. (I think the EU is considering that, which makes sense.) But saying some organisations are allowed to work on AI while others aren't, sounds like the worst possible idea.
https://societyofrock.com/in-1985-frank-zappa-is-asked-to-te...
Less fulfilling text version:
If you take seriously any downsides, whether misinformation or surveillance or laundering bias or x-risk, how does AI model weights or training data being open source solve them? Open source is a lot of things, but one thing it's not is misuse-resistant (and the "with many eyes all bugs are shallow" thing hasn't proved true in practice even with high level code, much less giant matrices and terabytes of text). Is there a path forward that doesn't involve either a lot of downside risk (even if mostly for people who aren't on HN and interested in tinkering with frontier models themselves, in the worlds where surveillance or bias is the main problem), or significant regulation?
I don't particularly like or trust Altman but I don't think he'd be obviously less self-serving if he were to oppose any regulation.
The layperson in the middle who has been happily plugging prompts into ChatGPT claiming they are a "prompt expert" are the ones the most excited.
For those that truly understand AI, there is a lot that you should genuinely be worried about. Now, don't confuse that for saying that we shouldn't work on it or should abandon AI work. I truly believe that this is the next greatest revolution. This is 1,000x more transformative than the industrial revolution, and 100x more transformative than the internet revolution. But it is worth a brief consideration of the effects of our work before we start running into these changes that could have drastic effects on everybody's daily life.
Except ... when you look at the problem from a military/national security viewpoint. Do we really want to give this tech away just like that?
Yes. It is. I'm sure hostile, authoritarian states that are willing to wage war with the world like Russia and North Korea will eventually get their hands on military-grade AI. But the free world should always strive to be two steps ahead.
Even having ubiquitous semi-automatic rifles is a huge problem in America. I'm sure Cliven Bundy or Patriot Front would do everything they can to close the gap with intelligent/autonomous weapons, or even just autonomous bots hacking America's infrastructure. If everything is freely available, what would be stopping them?
Work on open source locally hosted AI is important. I keep local clones and iterate as I can.
Let the market develop organically.
A government-controlled… never mind artificial god, a government-controlled story teller can be devastating.
I don't buy Musk's claim ChatGPT is "woke" (or even that the term is coherent enough to be tested), but I can say that each government requiring AI to locally adhere to national mythology, will create self-reinforcing cognitive blind spots, because that already happens at the current smaller scale of manual creation and creators being told not to "talk the country down".
But, unless someone has a technique for structuring an AI such that it can't be evil even when you, for example, are literally specifically trying to train it to support the police no matter how authoritarian the laws are, then a fully open source AGI is almost immediately also a perfectly obedient sociopath of $insert_iq_claim_here.
I don't want to wake up to the news that some doomsday cult has used one to design/make a weapon, nor the news a large religious group target personalised propaganda against me and mine.
Fully open does that by default.
But, you're still right, if we don't grok the AI, the governments can each secretly manipulate the AI and bend it to government goals in opposition to the people.
This is a key point. Every culture and agency and state will want (deserve) their own homespun AGI. But can we all learn how to accommodate to or accept a cultural multiverse when money and resources are zero-sum in many dimensions.
Hanno Rajaniemi’s Quantum Thief trilogy gives you a foretaste of where we could end up.
