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/)T

Telorand

@ Telorand @reddthat.com

Posts
11
Comments
1085
Joined
3 yr. ago

  • AI is making us think and write more alike: Large language models are standardizing human expression — and subtly influencing how we think

    Jump

  • No. If people think em-dashes are a "surefire sign" of LLMs, they're just as dumb as the people who take LLM output uncritically. Sometimes, you need to separate a thought with something other than a period, semicolon, or parenthesis, and a hyphen or double hyphen is simply not correct grammar. LLMs can pry my em-dashes from my cold, dead fingers.

  • How Much Do LLMs Hallucinate in Document Q&A Scenarios? A 172-Billion-Token Study Across Temperatures, Context Lengths, and Hardware Platforms [TLDR: 25%]

    Jump

  • How Much Do LLMs Hallucinate in Document Q&A Scenarios? A 172-Billion-Token Study Across Temperatures, Context Lengths, and Hardware Platforms [TLDR: 25%]

    Jump

  • Great, when did I say otherwise? Pareidolia is a thing humans do, because we like patterns. Finding patterns is something that has benefited our species, but it is sometimes so strong that we see faces in electrical outlets or the shape of a car's front profile (for example).

  • How Much Do LLMs Hallucinate in Document Q&A Scenarios? A 172-Billion-Token Study Across Temperatures, Context Lengths, and Hardware Platforms [TLDR: 25%]

    Jump

  • We do enjoy pareidolia, don't we?

  • How Much Do LLMs Hallucinate in Document Q&A Scenarios? A 172-Billion-Token Study Across Temperatures, Context Lengths, and Hardware Platforms [TLDR: 25%]

    Jump

  • One in 100. However, that is simple a measure of probability, so do not expect that to always be true for every 100 prompts.

    For example, if you rolled a 100-sided die 100 times, it's possible to get a one every time. In practice, it would likely be a mix. You might have a session where you get no wrong answers and times when you get several.

    The problem is that ignorant people trust these models implicitly, because they sound convincing and authoritative, and many people are not equipped to be able to vet the information being generated (also notice I didn't say "retrieved").

  • ‘Happy (and safe) shooting!’ popular AI chatbots like Open AI’s ChatGPT, Meta AI, and Google Gemin helped teen users plan violence and killings in hundreds of tests

    Jump

  • That's what regular people never seem to understand (and the AI apologists are hoping you don't know). These models aren't "getting better," they're just filled with more reactive patches over these unintended responses. And as the models scale up, so do the holes that need patching.

    It's a never ending game of bad-prompt Whack-a-Mole, all at the cost of our environment and safety, just so the Tech Bros can try to convince venture capitalists that "AGI is definitely just around the corner, trust me, bro," and keep that bubble filled with their own farts.

  • Voiden - A Markdown based Open Source Alternative to Postman

    Jump

  • Absolutely! Not making fun, but whenever somebody brings up Emacs in earnest, this is what I think of!

  • Tech industry is in tariff hell, even if refunds are automated

    Jump

  • Voiden - A Markdown based Open Source Alternative to Postman

    Jump

  • Emacs is not that hard. You can learn Emacs in one day, every day!

  • System76 on Age Verification Laws

    Jump

  • A relatively small company can't afford to fight a protracted legal battle or simply ignore the law. They have employees with families, and $800/hr for legal representation adds up fast, not to mention potentially getting hit with $6500 fines per infraction for refusal to comply. They also can't afford to just not sell in California, which has a huge chunk of the US population.

    We don't have to be happy about the state of things, but it's not their fault that capitalism and authoritarianism have effectively forced them to comply.

    Be upset by all means, but remember to focus your anger upon those who actually put/is putting these laws in place.

  • Kamala Harris might run for president again in 2028. Please, no | Arwa Mahdawi

    Jump

  • Are you fucking stupid? You lost to Donald Fucking Trump—you know, the racist, fascist pedophile rapist? You lost to that guy, and you'd be running against his specter and his legacy, the racist and xenophobic sentiments that still run deep in this country.

    What is uniting the American people, however, is a rising class consciousness, not establishment Dems like you. We already tried voting for "not-Trump," and y'all squandered the opportunity to appease the wealthy, expand the police state, send more bombs to Israel, and treat all the people shouting "Danger!" like children, rather than securing our democracy and standing against genocide.

    Fuck. Right. Off.

  • Meta Workers Say They're Seeing Disturbing Things Through Users' Smart Glasses

    Jump

  • It never really stopped. Even after Cambridge Analytica, Meta was allowed to go back to business as usual with barely a slap on the wrist.

    That people keep buying their products and using their platforms and then being shocked to find out they've been used is honestly people's own ignorant fault.

  • Meta Workers Say They're Seeing Disturbing Things Through Users' Smart Glasses

    Jump

  • Some phones do have hardware switches, but they're certainly the exception.

    Whether our phones are surreptitious spy devices when they're oriented correctly and have adversarial software running is very different from, "I'm going to buy always-on camera glasses and send the data to one of the most untrustworthy companies on the planet," the latter of which was my point.

  • Meta Workers Say They're Seeing Disturbing Things Through Users' Smart Glasses

    Jump

  • Well yeah. The people who buy these things are disturbing individuals. Kinda goes with the territory.

  • I searched CachyOS vs Fedora on YouTube and all I found was AI Slop 🙄

    Jump

  • CachyOS also has a live USB.

    But I'd actually go a step further. Install both in a VM using their flagship DE. CachyOS's installer actually comes with many different DEs and window managers, so you should be able to get the look and feel of many different setups there, too.

  • Bcachefs creator claims his custom LLM is 'fully conscious'

    Jump

  • I know what you're trying to say, and I'm inclined to agree on some level, but unlike the days of the dotcom bubble, there's people who recognize what these systems represent and are doing things to counter their effects. To use your examples, AWS and Cloudflare are so prolific, because they were allowed to be without any meaningful resistance in their early stages.

    Thankfully, we are still in the early stages, and even with all the widespread use by consumers and businesses, generative AI still isn't profitable. There's resistance to their efforts by regular people and those with platforms, so I'm less inclined to think of these systems as inevitable; even if they are, I don't think they'll be the only option.

  • Bcachefs creator claims his custom LLM is 'fully conscious'

    Jump

  • Bcachefs creator claims his custom LLM is 'fully conscious'

    Jump

  • Yes, I do. My reasoning is twofold:

    • Existing tools rely greatly upon data generated by humans. Reddit in particular has been noted as a large source of training data for LLMs, and I believe Stack Overflow has as well. If people start to rely heavily upon LLMs, their training data gets stale. AI companies have tried to shore up these shortcomings by training on other AI generated datasets, but that is precisely how hallucinations happen.
      • Essentially, LLMs as sold by the tech bros are an ouroboros. They will stall without fresh and unique human input.
    • LLM usage does not reinforce learning. You can produce code, maybe even quickly, but the skills needed to produce good code are ones you have to maintain with practice. If LLMs were to become the defacto coding tool used by nearly everyone, I expect we'd lose the ability to maintain those very models within a generation.
      • tldr: LLMs make people stupid.

    I agree that they're not fully going away, but the Boomers and Gen Xers who are trying to shoehorn AI into everything don't actually understand what it is they've bought into, and if things continue as they are, tech bro AI will eat itself, leaving the bespoke ML models to do actually useful things in areas like science and medicine.

  • Bcachefs creator claims his custom LLM is 'fully conscious'

    Jump

  • Later: "Are you fully conscious?"

    "No, I'm just an AI simulating consciousness."

    "But I thought you said you were conscious before...?"

    "I'm sorry, you're absolutely right! I am conscious. Thank you for pointing out my error. I'm always striving to improve my answers."