Skip Navigation

InitialsDiceBearhttps://github.com/dicebear/dicebearhttps://creativecommons.org/publicdomain/zero/1.0/„Initials” (https://github.com/dicebear/dicebear) by „DiceBear”, licensed under „CC0 1.0” (https://creativecommons.org/publicdomain/zero/1.0/)U

UnpluggedFridge

@ UnpluggedFridge @lemmy.world

Posts
0
Comments
55
Joined
2 yr. ago

  • FBI Arrests Man For Generating AI Child Sexual Abuse Imagery

    Jump

  • These cases are interesting tests of our first amendment rights. "Real" CP requires abuse of a minor, and I think we can all agree that it should be illegal. But it gets pretty messy when we are talking about depictions of abuse.

    Currently, we do not outlaw written depictions nor drawings of child sexual abuse. In my opinion, we do not ban these things partly because they are obvious fictions. But also I think we recognize that we should not be in the business of criminalizing expression, regardless of how disgusting it is. I can imagine instances where these fictional depictions could be used in a way that is criminal, such as using them to blackmail someone. In the absence of any harm, it is difficult to justify criminalizing fictional depictions of child abuse.

    So how are AI-generated depictions different? First, they are not obvious fictions. Is this enough to cross the line into criminal behavior? I think reasonable minds could disagree. Second, is there harm from these depictions? If the AI models were trained on abusive content, then yes there is harm directly tied to the generation of these images. But what if the training data did not include any abusive content, and these images really are purely depictions of imagination? Then the discussion of harms becomes pretty vague and indirect. Will these images embolden child abusers or increase demand for "real" images of abuse. Is that enough to criminalize them, or should they be treated like other fictional depictions?

    We will have some very interesting case law around AI generated content and the limits of free speech. One could argue that the AI is not a person and has no right of free speech, so any content generated by AI could be regulated in any manner. But this argument fails to acknowledge that AI is a tool for expression, similar to pen and paper.

    A big problem with AI content is that we have become accustomed to viewing photos and videos as trusted forms of truth. As we re-learn what forms of media can be trusted as "real," we will likely change our opinions about fringe forms of AI-generated content and where it is appropriate to regulate them.

  • We have to stop ignoring AI’s hallucination problem

    Jump

  • We do not know how LLMs operate. Similar to our own minds, we understand some primitives, but we have no idea how certain phenomenon emerge from those primitives. Your assertion would be like saying we understand consciousness because we know the structure of a neuron.

  • We have to stop ignoring AI’s hallucination problem

    Jump

  • You seem pretty confident that LLMs cannot have an internal representation simply because you cannot imagine how that capability could emerge from their architecture. Yet we have the same fundamental problem with the human brain and have no problem asserting that humans are capable of internal representation. LLMs adhere to grammar rules, present information with a logical flow, express relationships between different concepts. Is this not evidence of, at the very least, an internal representation of grammar?

    We take in external stimuli and peform billions of operations on them. This is internal representation. An LLM takes in external stimuli and performs billions of operations on them. But the latter is incapable of internal representation?

    And I don't buy the idea that hallucinations are evidence that there is no internal representation. We hallucinate. An internal representation does not need to be "correct" to exist.

  • We have to stop ignoring AI’s hallucination problem

    Jump

  • How do hallucinations preclude an internal representation? Couldn't hallucinations arise from a consistent internal representation that is not fully aligned with reality?

    I think you are misunderstanding the role of tokens in LLMs and conflating them with internal representation. Tokens are used to generate a state, similar to external stimuli. The internal representation, assuming there is one, is the manner in which the tokens are processed. You could say the same thing about human minds, that the representation is not located anywhere like a piece of data; it is the manner in which we process stimuli.

  • We have to stop ignoring AI’s hallucination problem

    Jump

  • My thesis is that we are asserting the lack of human-like qualities in AIs that we cannot define or measure. Assertions should be made on data, not uneasy feelings arising when an LLM falls into the uncanny valley.

  • We have to stop ignoring AI’s hallucination problem

    Jump

  • I think where you are going wrong here is assuming that our internal perception is not also a hallucination by your definition. It absolutely is. But our minds are embodied, thus we are able check these hallucinations against some outside stimulus. Your gripe that current LLMs are unable to do that is really a criticism of the current implementations of AI, which are trained on some data, frozen, then restricted from further learning by design. Imagine if your mind was removed from all stimulus and then tested. That is what current LLMs are, and I doubt we could expect a human mind to behave much better in such a scenario. Just look at what happens to people cut off from social stimulus; their mental capacities degrade rapidly and that is just one type of stimulus.

    Another problem with your analysis is that you expect the AI to do something that humans cannot do: cite sources without an external reference. Go ahead right now and from memory cite some source for something you know. Do not Google search, just remember where you got that knowledge. Now who is the one that cannot cite sources? The way we cite sources generally requires access to the source at that moment. Current LLMs do not have that by design. Once again, this is a gripe with implementation of a very new technology.

    The main problem I have with so many of these "AI isn't really able to..." arguments is that no one is offering a rigorous definition of knowledge, understanding, introspection, etc in a way that can be measured and tested. Further, we just assume that humans are able to do all these things without any tests to see if we can. Don't even get me started on the free will vs illusory free will debate that remains unsettled after centuries. But the crux of many of these arguments is the assumption that humans can do it and are somehow uniquely able to do it. We had these same debates about levels of intelligence in animals long ago, and we found that there really isn't any intelligent capability that is uniquely human.

  • Takeaways from AP report on how the search for the coronavirus origins turned toxic

    Jump

  • Where is the safety report for the Wuhan wetmarket? You know the one that unequivocally started one viral pandemic? Then, while closed down, we enjoyed a period with no new coronavirus pandemics? And then, shortly after reopening, there was another coronavirus pandemic originating in Wuhan? That wetmarket, you have a report on that one?

  • ChatGPT provides false information about people, and OpenAI can’t correct it

    Jump

  • Notice that there are methods, data, and peer reviews that I can freely scrutinize. All things your opinion piece lacks.

  • X now treats the term cisgender as a slur

    Jump

  • It is so strange to say that identity should take a back seat to humanism when every historical example of discrimination and dehumanization is based on identity. Identity in those instances is not imposed on oneself, but is used to define the outgroup that is being dehumanized. Identity politics is simply an honest accounting of groups that being descriminated against. When the discrimination ends, we see the group identity evaporate. We need only look at the early 20th century definitions of Caucasian, and the identity politics of Irish and Italian Americans subsequently evaporating when that definition evolved to include all Americans of European decent, to see that identity politics is a reaction to injustice and not the other way around.

  • Removed

    US Lemmys what could Biden do in the next 6 months to EARN your vote? (other than just not being Trump)

    Jump

  • For real, we've got the first openly pro-union president, we expanded NATO, student loan forgiveness, actual infrastructure funding, the first administration to openly push back against Israel during war time, all of that in only 4 years. He is the most effective president of my lifetime and I am happy to vote for him again.

  • Locked

    TikTok sues the US government over ban

    Jump

  • The real question you are asking is whether inaction is worse than inconsistency. Should we not put out a fire unless we can put out all fires? What you are suggesting is to let something burn for the sake of consistency.

  • Locked

    TikTok sues the US government over ban

    Jump

  • TikTok pushed a notifications to all US users with the phone numbers of their local congressmen to oppose the bill. So many calls came in that the phone lines were jammed.

    Let me distill that for you: China attempted to directly influence legislation with a mass propaganda campaign targeted at its US user base.

    Please explain to me why that isn't a threat and why the US should allow hostile foreign powers to directly influence internal politics?

  • Locked

    TikTok sues the US government over ban

    Jump

  • This is the real question. Is there a loophole that allows foreign governments to freely exercise mass surveillance and psyops if they allow US citizens to post on a blackboard outside their offices?

  • Locked

    TikTok sues the US government over ban

    Jump

  • The government can already access the data with a warrant. The ownership of TikTok has literally 0 effect on the government's ability to access user data. Not being owned by the Chinese government has a huge impact on China's ability to access that data.

  • Locked

    TikTok sues the US government over ban

    Jump

  • Except the most relevant part: it is owned by a hostile foreign government.

  • Locked

    TikTok sues the US government over ban

    Jump

  • Requires a warrant or subpoena. That is the difference.

  • Study reveals "widespread, bipartisan aversion" to neighbors owning AR-15 rifles

    Jump

  • Details like this are really just a distraction. Do you really think the average respondent understands these technical details, or have any good reason to memorize the specs of all rifles? The focus on the AR-15 is not because of any risk associated with that particular gun, but because most people understand that this is a semi-auto rifle. There is no other model of gun that will have that kind of widespread recognition.

    Drawing up these very silly technical arguments is a willful ignorance of the underlying issue: What is the limit of deadly force we should allow one person to lawfully own? We don't let people own tactical nukes. We don't need to argue over thermonuclear or hydrogen nukes. We don't need to understand quantum mechanics to regulate these devices. The technical details do not matter. The potential body count is what matters. And so it is with guns, which happen to occupy that grey area where reasonable people disagree on an acceptable level of lethality. You do not need to know all the different models of gun to be killed by one, so we should not require such technical knowledge when engaging in discourse around their regulation.

  • Biden administration plans to reclassify marijuana, easing restrictions nationwide

    Jump

  • Is no one going to acknowledge that a huge portion of the American electorate actually supports Israel's genocide? Part of living in a democracy is accepting that official policy reflects some mixture of the views of the electorate. If the US electorate is still mixed in its view of Israel then the official US policy should be mixed as well (which it is).

    I will use my vote to push for an end to the genocide, the release of hostages, and a stable 2-state solution. But I will not abandon core democratic values just because I find myself in the minority.

  • ChatGPT provides false information about people, and OpenAI can’t correct it

    Jump

  • You linked a paper on planning in LLMs. Planning is largely in the domain of reinforcement learning. The paper you linked conflates reasoning with planning, alongside the obviously biased prose, so the author really doesn't seem credible. I prefer nuanced and careful evaluations such as: https://www.sciencedirect.com/science/article/pii/S2949719123000298

  • ChatGPT provides false information about people, and OpenAI can’t correct it

    Jump

  • Here I think you are behind on the literature. LLMs can infer and reason, and there are whole series of papers that evaluate LLMs for these properties the exact same way we evaluate humans. So if you can't trust the metrics, then you cannot even assert that humans can reason and infer and understand.