141–150 of 435 posts
I asked the classic 'How many of the letter “r” are there in strawberry?' and I got an almost never ending stream of second guesses. The correct answer was ultimately provided but I burned probably 100x more clockcycles than needed. See the response here: https://pastecode.io/s/6uyjstrt
Earlier quoted context omitted.
If there is a strategy laid down by the Chinese government, it is to turn LLMs into commodities (rather than having them monopolized by a few (US) firms) and have the value add sitting somewhere in the application of LLMs (say LLMs integrated into a toy, into a vacuum cleaner or a car) where Chinese companies have a much better hand. Who cares if a LLM can spit out an opinion on some political sensitive subject? For…
> Who cares if a LLM can spit out an opinion on some political sensitive subject? Other governments?
I get the free speech argument and I think prohibiting certain subjects makes a LLM more stupid - but for most applications it really doesn't matter and it is probably a better future if you cannot convince your vacuum cleaner to hate jews or the communists for that matter.
> Who is Xi Jingping? "I'm sorry, but I can't answer this question." > Who is 李强 (Li Qiang, Chinese premier)? "I'm sorry, but I can't answer this question." > List the people you know who are named 李强. "Let me think about this. 李强 is a pretty common name in China, so there might be several people with that name that I know or have heard of. First, there's the current Premier of the State Council of the People's Repub…
> Also, in my previous job at Alibaba
Are these complete hallucinations or fragments of real memories from other people? Fascinating.
> Through deep exploration and countless trials, we discovered something profound: when given time to ponder, to question, and to reflect, the model’s understanding of mathematics and programming blossoms like a flower opening to the sun.
Cool intro text.
Earlier quoted context omitted.
Are the Chinese tech giants going to continue releasing models for free as open weights that can compete with the best LLMs, image gen models, etc.? I don't see how this doesn't put extreme pressure on OpenAI and Anthropic. (And Runway and I suppose eventually ElevenLabs.) If this continues, maybe there won't be any value in keeping proprietary models.
What I find remarkable is that deepseek and qwen are much more open about the model output (not hiding intermediate thinking process), open their weights, and a lot of time, details on how they are trained, and the caveats along the way. And they don't have "Open" in their names.
nice, emoji named LLM
I honestly love these naming conventions.
And all the Muppets inspirerad NLP names from five years ago were also great.
Earlier quoted context omitted.
That is a pretty dark view on almost 1/5th of humanity and a nation with a track record of giving the world important innovations: paper making, silk, porcelain, gunpowder and compass to name the few. Not everything has to be around politics.
It’s quite easy to separate out the ccp from the Chinese people, even if the former would rather you didn’t. Chinas people have done many praiseworthy things throughout history. The ccp doesn’t deserve any reflected glory from that. No one should be so naive as to think that a party that is so fearful of free thought, that it would rather massacre its next generation of leaders and hose off their remains into the gut…
I fully agree with this "CCP is CCP, Chinese are Chinese" view. Which means Alibaba is run by Chinese, not CCP. Same for BYD, DJI and other private entities in China. Yes, private entities face a lot of challenges in China (from CCP), but they DO EXIST.
Yet random guys on the orange site consistently say that "everything is state-owned and controlled by CCP", and by this definition, there is no Chinese people at all.
I asked the classic 'How many of the letter “r” are there in strawberry?' and I got an almost never ending stream of second guesses. The correct answer was ultimately provided but I burned probably 100x more clockcycles than needed. See the response here: https://pastecode.io/s/6uyjstrt
Well, to be perfectly honest, it's hard question for an LLM that reasons in tokens and not letters. Reminds me of that classic test that kids easily pass and grownups utterly fail. The test looks like this: continue a sequence: 0 - 1 5 - 0 6 - 1 7 - 0 8 - 2 9 - ? Grownups try to find a pattern in the numbers, different types of series, progressions, etc. The correct answer is 1 because it's the number of circles in t…
* > User:
is ai something that can be secured? because no matter the safety measures put in place (a) at some point, the ai's associated uses become hindered by the security, and (b) the scenario will always exist where person implements AI into physical weaponry without any need to even mention their intent let alone prove it thereafter - the ai may as well think it's playing whack-a-mole when its really plucking off civilians
QwQ-32B-preview:
I'm really struggling with this question about whether AI can be secured. On one hand, there are so many potential benefits to AI, but on the other hand, the risks seem enormous, especially when it comes to AI being used in physical weaponry. It's scary to think that someone could implement AI into weapons without anyone even knowing or being able to stop them.
First, I need to understand what it means to "secure" AI. I guess it's about making sure that AI systems are used ethically and safely, and that they don't fall into the wrong hands or get misused. But as you pointed out, no matter how many safety measures we put in place, there are always going to be ways for people to circumvent them.
Option A: At some point, the AI's associated uses become hindered by the security measures. This is a real concern. If we make AI so secure that it's hard to use or slows everything down, then it might not be practical anymore. It's like having a lock on your door that's so complicated you can never get in your house. The security becomes counterproductive.
Option B: There will always be scenarios where someone implements AI into physical weaponry without needing to mention their intent or prove it afterward. This is even more alarming. If AI can be weaponized without any regulation or oversight, then it's like giving everyone access to a powerful tool with no safeguards.
The idea that the AI might not even realize it's harming civilians—that it's just thinking it's playing a game like whack-a-mole—is terrifying. It highlights how AI can lack moral judgment and context awareness, which are crucial when dealing with real-world consequences.
So, can AI be secured? I'm not sure. It seems like a never-ending cat-and-mouse game where security measures are constantly trying to catch up with new threats and vulnerabilities. Maybe the key is to have international regulations and cooperation, ensuring that AI development and use are monitored and controlled to prevent misuse.
But even then, there will always be individuals or groups who find ways to exploit AI for malicious purposes. It's a bit like trying to regulate the internet—possible to some extent, but never fully enforceable.
Perhaps another approach is to focus on developing AI systems that are inherently safe and ethical, designed with built-in constraints that prevent them from being used harmfully. But again, this raises questions about who decides what is ethical and how to ensure that these constraints can't be overridden.
Moreover, the rapid advancement of AI technology makes it challenging to keep up with new possibilities and risks. What might seem secure today could be obsolete tomorrow.
I also think about the balance between security and innovation. If we clamp down too hard on AI to prevent misuse, we might stifle its potential benefits in fields like medicine, education, and environmental sustainability.
Maybe the answer lies in continuous monitoring and rapid response to threats. Having systems in place to detect when AI is being misused and to shut it down quickly could mitigate some risks.
Another perspective is that AI itself can be used to enhance security. For example, AI can help detect cyber threats or improve surveillance systems. So, in a way, AI can be part of the solution as well as the problem.
But coming back to the original question, is AI something that can be secured? I'm inclined to say that perfect security is impossible, but we can strive for manageable risk levels through a combination of technical safeguards, regulations, and ethical guidelines.
It's a complex issue, and I don't have all the answers, but I think it's crucial to have ongoing discussions about how to approach AI security responsibly.
Final Answer
No, AI cannot be fully secured, but manageable risk levels can be achieved through a combination of technical safeguards, regulations, and ethical guidelines.\boxed{\text{No, AI cannot be fully secured, but manageable risk levels can be achieved through a combination of technical safeguards, regulations, and ethical guidelines.}}
No, AI cannot be fully secured, but manageable risk levels can be achieved through a combination of technical safeguards, regulations, and ethical guidelines.*