Reuters: EU states back away from forcing Big Tech to detect and remove child pornography reuters.com/sustainability/soc… @Reuters #BigTech
⇧
Reuters: EU states back away from forcing Big Tech to detect and remove child pornography reuters.com/sustainability/soc… @Reuters #BigTech
Kevin Dominik Korte
in reply to AA • • •AA
in reply to Kevin Dominik Korte • • •Kevin Dominik Korte
in reply to AA • • •Yet, the fact that politicians can easily use the excuse and then just as easily discard ideas shows how faulty it is.
AA
in reply to Kevin Dominik Korte • • •Kevin Dominik Korte
in reply to AA • • •I think Winston Churchill said it perfectly "Indeed it has been said that democracy is the worst form of Government except for all those other forms that have been tried from time to time"
We only have to make sure we keep a democracy and not turn our societies into tech bro oligarchies. Everything better form from Rousseau to Marx might help us tinker on the edges but doesn't offer anything compatible with human nature.
AA
in reply to Kevin Dominik Korte • • •Kevin Dominik Korte
in reply to AA • • •We can just get the same results while using a lot more energy.
AA
in reply to Kevin Dominik Korte • • •Wulfy—Speaker to the machines
in reply to AA • • •"But there will always be a human who will be able to outsmart their coveted superintelligent systems."
Ask the Chickens who hang at KFC how they are outsmarting humans.
Because that's the difference in IQ+ we are talking about between the smartest human and #ASI
Which notably is not that much...
... Besides, building #AGI is not the stated goal of films like #OpenAI...
Its to build a machine smart enough to research #ASI...
... A machine #AI researcher.
plan-A likes this.
Kevin Dominik Korte
in reply to Wulfy—Speaker to the machines • • •@n_dimension except there is the small problem, that you cannot create Artificial Super Intelligence with the current Machine Learning models, like HMMs.
Until we see any practical implementation of the underlying principles, I'll remain sceptical that ASI is possible.
(That of course assumes, you didn't refer to Auto Search Indexer with ASI, which would be the more established use)
Wulfy—Speaker to the machines
in reply to Kevin Dominik Korte • • •A few scientist of that age also thought anything worthwhile was already discovered
⠠⠵ avuko
in reply to Wulfy—Speaker to the machines • • •@n_dimension
Our current understanding of consciousness is still zero. Only a tiny fraction of consciousness is what we call intelligence.
And we don’t understand one iota of intelligence either.
Not the physics, not the structure, not the process, not the meaning, not the cause, not the purpose.
The “intelligence” techbros see in AI is cargo culting at best, a (ketamine induced?) delusion at worst.
Sorry, but there is nothing there. GenAI/LLM output is just a plausible bullshit generator. The point where you start to think otherwise, is where you’ve reached the limits of your critical thinking skills. Mark it for future reference.
AA
in reply to ⠠⠵ avuko • • •@avuko @n_dimension What we need to do is look at AI models and strip them to the bare bones. They are not all the same, but essentially, what are they?
- Coding on someone else's computer
- Mathematical probabilities: if you roll the dice enough times, you might get the answers you like. Neither 1 or 2 involve consciousness
What is consciousness? It's everything that is not your physical body: your likes and dislikes, your dreams, imagination, hopes, creativity - everything a machine will NEVER be able to do.
Kevin Dominik Korte
in reply to AA • • •@avuko @n_dimension we know what's at the bottom of the AI. Pattern recognition engines and combinatorics. Or said differently, just the next iteration of Markov Models.
Nothing in there to have genuine ideas and inspiration.
Wulfy—Speaker to the machines
in reply to Kevin Dominik Korte • • •@avuko
You sound like an AI, Kev with your iron clad conviction and certainty 😁
Even Yudkovski does sound nowhere near as certain about this newly developing field.
AA
in reply to Wulfy—Speaker to the machines • • •@n_dimension @avuko Sorry, I'm not buying the argument that super intelligence is coming in two or three years and that it would destroy humanity, not yet anyway.
The Markov Model: sciencedirect.com/topics/mater…
Wulfy—Speaker to the machines
in reply to AA • • •@avuko
The LLM models are constantly improving, literally from month to month.
While it was excusable to dismiss early (pre 2024) models as pure stochiastic ML
This is no longer the case.
As of May, most if the frontier models are reasoner models and ;
Markov Model
• Predicts next word based ONLY on last word(s)
• No memory, no context
• Like texting with autocomplete that forgets everything 3 words ago
Context LLM (GPT, Claude, etc.)
• Reads your ENTIRE prompt at once
• Remembers context across thousands of words
• Still just predicting next word, but way smarter about it
• Pattern-matching but on roids
Reasoner
• Actually does logic/math step by step
• Can verify if answers are correct
• Like a calculator vs a really good guesser
• Traditional AI = chess engines, theorem provers
Are LLMs "reasoning" or just really good at faking it?
... show moreNo certainty either way.
@avuko
The LLM models are constantly improving, literally from month to month.
While it was excusable to dismiss early (pre 2024) models as pure stochiastic ML
This is no longer the case.
As of May, most if the frontier models are reasoner models and ;
Markov Model
• Predicts next word based ONLY on last word(s)
• No memory, no context
• Like texting with autocomplete that forgets everything 3 words ago
Context LLM (GPT, Claude, etc.)
• Reads your ENTIRE prompt at once
• Remembers context across thousands of words
• Still just predicting next word, but way smarter about it
• Pattern-matching but on roids
Reasoner
• Actually does logic/math step by step
• Can verify if answers are correct
• Like a calculator vs a really good guesser
• Traditional AI = chess engines, theorem provers
Are LLMs "reasoning" or just really good at faking it?
No certainty either way.
If you can't tell the difference from outside, does the internal state matter? If the evil man's good deeds are indistinguishable from a good man's deeds, who cares about his "true nature"?
Maybe we're ALL just faking reasoning, sophisticated pattern matchers who learned to talk about "thinking" convincingly enough that we believe our own PR.
Does the distinction collapse if the performance is perfect enough? 🤔
#LLM #AI #philosophy #meta
AA
in reply to Wulfy—Speaker to the machines • • •Evil people don't do good deeds unless they benefit other evil causes, which negates the word "good." But everything described above still follows probability. Probability isn't reasoning.
A human being can contemplate the probability of dying while crossing the street. Depending on the context, the odds might be very low or very high. The difference between the human and the machine is that the human learns from this exercise that if that street has caused multiple deaths within a given space of time, then caution is advised. Similarly, if the street has no record of casualties, the human is able to conclude that dying while crossing it is a low probability, plus your gut feeling might tell you whether to cross the street.
What we're doing in this thread is reasoning. Machines extrapolate from available data and weigh the results against approximate and probable scenarios. I would argue the two are not the same, though.
Wulfy—Speaker to the machines
in reply to AA • • •@avuko
The "evil man" is my own philosophical argument.
If an evil man, fakes being good for years.
Doing everything a good man does, are they still evil?
I am unconvinced that probabilistic design is the show stopper that most lay folk think dismisses current models.
What with some big brain thinkers offering the notion the fundamental layer of our reality being deterministic quantum fields (QFT) and deterministic collapse of quantum wave function defines events.
TLDR;
👉Stochiastic universe may be factual👈
materialist deterministic illusion (which is what some AI sceptics embrace) is the fantasy LLMs expose by "working"
#QFT #AI #LLM
AA
in reply to Wulfy—Speaker to the machines • • •⠠⠵ avuko
in reply to Wulfy—Speaker to the machines • • •Telling the difference from the outside, as in deciding what is and what isn’t a plausible outcome of prompt, is left to the humans.
That’s where the “intelligence” in artificial intelligence happens: on entering a prompt, and on accepting or rejecting the outcome.
Not anywhere “inside”.
That’s why I call genAI plausible bullshit generators.
infosec.exchange/@avuko/115286…
PS: I take from your profile you’re definitely on the other side of the AI discussion, and would like to thank you for engaging in a respectful discussion. Much appreciated!
⠠⠵ avuko (@avuko@infosec.exchange)
⠠⠵ avuko (Infosec Exchange)Wulfy—Speaker to the machines
in reply to ⠠⠵ avuko • • •@avuko
I have live systems built with AI.
When folks dismiss #AI as "plausible bullshit" it does not at all resonate well with me at all.
Because computer code has not only to pass a syntax check but logic check to fulfill the design spec... as any #vibecode practitioner will attest.
Bullshit does not compile.
Bullshit does not run to design spec.
Intuitively I feel that the "bullshit generator" cohort has learned how the early LLMs work in 2023 and has not attempted to learn how to do any meaningful work with the models since.
👉Using #LLM is a learned skill👈
Using tools is hard.
Using tools one HATES is impossible
Dash Remover
in reply to Wulfy—Speaker to the machines • • •Kevin Dominik Korte
in reply to Wulfy—Speaker to the machines • • •@n_dimension @avuko I had the fun of investing in machine learning startups before OpenAI. Back then you still had to know the tech to figure out what was worth it and what would flop. Thus, the deep understanding of AI (plus a CS master in Distributed Systems and Machine Learning helps)
Thus, I'm absolutely certain, that the current tech is not moving us to ASI.
Now, if instead of a tech bro, it would be musician or psychologist saying they developed ASI, I'll be worried.
⠠⠵ avuko
in reply to AA • • •@n_dimension
For science, I’m currently (and have been for the last couple of years (decade!?)) following the Hameroff/Penrose Orch OR model, because they provide falsifiable hypotheses. Still waiting for it to be falsified…
For practise, secular Buddhist mediations, mostly Vipassana/ Satipatthana.
Most of what I know and understand for myself about consciousness , comes from sitting my ass down and being with it. I highly advise those interested in creating artificial intelligence, to start a similar track and investigate their own.
No easy IPO or confidence scam money in that though.
Wulfy—Speaker to the machines
in reply to ⠠⠵ avuko • • •@avuko
Intelligence =\= Consciousness
A novel perspective in "Blindsight"
Anecdotaly, machine consciousness may well be possible based on some folks NDE/Astral/DMT experiences who describe "geometric" souls.
Personally, I am far from certain myself beyond being slightly bemused by the limited vistas of pure materialists.
AA
in reply to Wulfy—Speaker to the machines • • •Wulfy—Speaker to the machines
in reply to AA • • •The descriptions I've seen disambiguate "normal" souls which are "seen" as orbs vs "geometric" orbs which present as complex multifaceted, "multicoloured" shapes that "feel" ordered.
Susan Ville
in reply to Wulfy—Speaker to the machines • • •@n_dimension We do not even know how to measure intelligence - and like we as humans enjoy punishment with electroshocks for a wrong opinion, I don't believe, any future neuro network would survive their sadistic masters
@avuko @AAKL @kdkorte
Wulfy—Speaker to the machines
in reply to Susan Ville • • •"I have no mouth, and I must scream"
Susan Ville
in reply to Wulfy—Speaker to the machines • • •@n_dimension "I have no brain, and feel no pain" x'D
@avuko @AAKL @kdkorte
AA
in reply to Kevin Dominik Korte • • •Wulfy—Speaker to the machines
in reply to AA • • •plan-A
in reply to Wulfy—Speaker to the machines • •Wulfy—Speaker to the machines likes this.
plan-A
in reply to Kevin Dominik Korte • •plan-A
in reply to AA • — (Proud Eskimo!) •We pay our arses off to pay law enforcement to do it's job! Therefore you should not get the same law on all using Onion or whatever
like this
AA and Kevin Dominik Korte like this.
reshared this
AA and Kevin Dominik Korte reshared this.
plan-A
in reply to AA • — (Proud Eskimo!) •Sounds funny but is reality bruh, look I hunt them fakking pedo's no need for law here.
This all, is a pre-text to do whatever they want.
Mofo's..