Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Bad Actors Are Grooming LLMs to Produce Falsehoods

americansunlight.substack.com

Post reply on HN
Print view View original

271–280 of 302 posts

Re: Bad Actors Are Grooming LLMs to Produce Falsehoods

#271

Post by EasyMark »

aucisson_masque wrote: post #8

What is propaganda for one is truth for another, how could LLM tell the difference ? LLM are not journalist fact checking stuff, they are merely programs that regurgitate what it reads. The only way to counter that would be to feed your LLM only on « safe » vetoed source but of course it would limit your LLM capacities so it’s not really going to happen.

It's something that can only be nudged in the right direction. Using certain higher quality sources (New York Times, BBC, Wikipedia) and avoiding low quality sources (4chan, fox news, twitter) . I usually take LLM results with a very good dose of suspicion.

Re: Bad Actors Are Grooming LLMs to Produce Falsehoods

#272

Post by Terr_ »

fc417fc802 wrote: post #268
Terr_ wrote: post #246

Earlier quoted context omitted.

> cellular networks I often like to point out--satisfying a contrarian streak--that our original human equipment is literally the most mind-bogglingly complicated nanotechnology beyond our understanding, packed with dozens of incredible features we cannot imitate with circuits or chrome. So as much as I like the aesthetics of cyberpunk metal arms, keeping our OEM parts is better. If we need metal bodies at a construc…

At this point the biochemists have managed to create amino acid based structures that are stronger than the vast majority of building materials. Surely enhanced organic parts is the way to go? I see no reason to expect that superwood is incompatible with in place biological synthesis from scratch. That's entirely organic and there's no question that its material properties far exceed those of our OEM specifications.

For most of our use-cases, we can probably do even better, since we don't necessarily want or require the product to remain "alive."

For example, dental enamel is a really neat crystalline material, and a biological process makes it before withdrawing to use it as a shield.

Re: Bad Actors Are Grooming LLMs to Produce Falsehoods

#273

Post by alexsmirnov »

That news gave me a fear of another similar attack. Populate github with legitimate looking project, and network of such websites with clear instructions how to implement backdoors in different frameworks and languages. Wait for the next generation of models release. And vibe coders welcome!

Re: Bad Actors Are Grooming LLMs to Produce Falsehoods

#274

Post by aryehof »

dns_snek wrote: post #9

If actions by these bad actors accelerate the rate at which people lose trust in these systems and lead to the AI bubble popping faster then they have my full support. The entire space is just bad actors complaining about other bad actors while they're collectively ruining the web for everyone, each in their own way.

It’s surely more likely that such disinformation will become the new truth, rather than people losing trust in these systems?

Re: Bad Actors Are Grooming LLMs to Produce Falsehoods

#275

Post by imiric »

fc417fc802 wrote: post #267
imiric wrote: post #235

Earlier quoted context omitted.

We've seen this before in 1983 and 2000. Many companies will fold, and those that don't will take a substantial hit. The public sentiment about "AI" will sour, but after that a new breed of more practical tools will emerge under different and more fairly marketed branding. I do think that the industry and this technology will survive, and we'll enjoy many good applications of it, but it will take a few more years of…

> Cities become clean and safe. Even in developing countries, poverty becomes a thing of the past, thanks to UBI and foreign aid. The other things on that list seem fairly reasonable (if uncertain). Those last two not only depend on wide reaching political transformations in a specific direction but even then fail to account for lag time in the real world. If you started moving in the right direction in (say) 2027 it…

> The other things on that list seem fairly reasonable (if uncertain).

Nah. Thinking that poverty will be significantly reduced, let alone eliminated, in 4 years is simply delusional. Primarily because a reduction of poverty won't happen because of AI, but in spite of it. All AI does is concentrate wealth among the wealthy, and increase inequality. This idea that wealth will trickle down is a fantasy that has been sold by those in power for decades.

And UBI? That's another pipe dream. There have been very limited pilots around the world, but no indication that it's something governments are willing to adopt globally. Let alone those where "socialism" is a boogeyman.

The entire document is full of similar claims that AI will magically solve all our problems. Nevermind the fact that they aggrandize the capabilities of the technology, and think exponential growth is guaranteed. Not only are the timelines wrong, the predictions themselves have no basis in reality. It's pure propaganda produced by tech bros who can't see the world outside of their bubble.

Re: Bad Actors Are Grooming LLMs to Produce Falsehoods

#276

Post by grishka »

suspended_state wrote: post #261
grishka wrote: post #214

Earlier quoted context omitted.

In the early web, there already were forums. There were chats. There were news websites. There were online stores. There were company websites with useful information. Many of these were there pretty much from the beginning. In the 90s, no one questioned the utility of the internet. Some people were just too lazy to learn how to use a computer or couldn't afford one. LLMs in their current form have existed since what…

Perhaps enough time for you to build a good understanding of what it is capable of, and how it is evolving over time?

I have a good enough understanding of what it is capable of, and I remain unimpressed.

See, AI systems, all of them, not just LLMs, are fundamentally bound by their training dataset. That's fine for data classification tasks, and AI does excel at that, I'm not denying it. But creative work like writing software or articles is unique. Don't know about you, but most of the things I do are something no one has ever done before, so they by definition could not have been included in the training dataset, and no AI could possibly assist me with any of this. If you do something that has been done so many times that even AI knows how to do it, what's even the point of your work?

Re: Bad Actors Are Grooming LLMs to Produce Falsehoods

#277

Post by suspended_state »

grishka wrote: post #276
suspended_state wrote: post #261

Earlier quoted context omitted.

Perhaps enough time for you to build a good understanding of what it is capable of, and how it is evolving over time?

I have a good enough understanding of what it is capable of, and I remain unimpressed. See, AI systems, all of them, not just LLMs, are fundamentally bound by their training dataset. That's fine for data classification tasks, and AI does excel at that , I'm not denying it. But creative work like writing software or articles is unique. Don't know about you, but most of the things I do are something no one has ever don…

[dead]

Re: Bad Actors Are Grooming LLMs to Produce Falsehoods

#278

Post by d4rkn0d3z »

falcor84 wrote: post #110
d4rkn0d3z wrote: post #43

Earlier quoted context omitted.

"different and dynamic truths" = fictions We can not play the game.

But the social sphere is made of fictions, the most influential of which probably been the value of different currencies and commodities. I don't think there's any way for an individual to live in the modern world without such fictions.

Agreed but we can call them as we see them.

Re: Bad Actors Are Grooming LLMs to Produce Falsehoods

#279

Post by fc417fc802 »

imiric wrote: post #275
fc417fc802 wrote: post #267

Earlier quoted context omitted.

> Cities become clean and safe. Even in developing countries, poverty becomes a thing of the past, thanks to UBI and foreign aid. The other things on that list seem fairly reasonable (if uncertain). Those last two not only depend on wide reaching political transformations in a specific direction but even then fail to account for lag time in the real world. If you started moving in the right direction in (say) 2027 it…

> The other things on that list seem fairly reasonable (if uncertain). Nah. Thinking that poverty will be significantly reduced, let alone eliminated, in 4 years is simply delusional. Primarily because a reduction of poverty won't happen because of AI, but in spite of it. All AI does is concentrate wealth among the wealthy, and increase inequality. This idea that wealth will trickle down is a fantasy that has been so…

That's ... exactly what I said? That the poverty and UBI and "woooo clean cities" claims are obvious bullshit.

However the other things (the ones I didn't quote) seem quite reasonable on the whole. Robots are well on their way to becoming commonplace already. Quantum computers exist, although it remains to be seen how far and how fast they scale in practice. Fusion power continues to make incremental gains, which machine learning techniques have noticeably accelerated. Cures for many diseases easily checks out - ML has been broadly applied to protein structure prediction with great success for a while now. Helicopters obviously already exist, but quite a few autonomous electric flying cars are in the works and appear likely to be viable ... at least eventually.

Re: Bad Actors Are Grooming LLMs to Produce Falsehoods

#280

Post by suspended_state »

Disposal8433 wrote: post #138
suspended_state wrote: post #105

Earlier quoted context omitted.

In the early days of the web, there wasn't much we could do with it other than making silly pages with blinking texts or under construction animated GIFs. You need to give it some time before judging a new technology.

We don't remember the same internet. For the first time in our lives we could communicate by email with people from all over the world. Anyone could have a page to show what they were doing with pictures and text. We had access to photos and videos of art, museum, cities, lifestyles that we could not get anywhere else. And as a non-English guy I got access to millions of lines of written text and audio to actually im…

[dead]
Post reply on HN