Please tell your friends about federated social media site that speaks several fediverse protocols thus serving as a hub uniting them, hubzilla.eskimo.com, also check out friendica.eskimo.com, federated
macroblogging social media site, mastodon.eskimo.com a federated microblogging site, and yacy.eskimo.com an uncensored federated search engine. All Free!
Yet another thing solved by AgentV3N's prompt routing.
We're moving closer and closer to using 1B models for the first stage of prompt evaluation. A simple "thank you" could be met with an agent-level prepared response, or even just an emoji reaction.
Meanwhile, ChatGPT by default uses the last model you used in the conversation. Imagine blowing o1 reasoning on "thanks mate". Wouldn't be me.
@SuperDicq What do you consider "deceptive" in this circumstance?
I can see a mihoyo-shaped object being like "Don't you love me enough to get the Professional plan?" but besides that I don't see a "persistent" customized character (what we call an Agent) being too terrible. It's basically a secretary at that point. @mischievoustomato
>normies are stupid and otherwise will start to think that there's some actual sort of sentience behind the scenes or something. That is exactly what normies believe - they think LLMs are sentient.
The whole business plan is to trick people into think they are interacting with an "AI" and separate fools from their money.
I know this is a relatively small survery (1000 participants), but a staggering 12% even believe that if they are not polite to ChatGPT now they will be the first ones die in an unavoidable future AI uprising.
And I'm supposed to be the mentally ill because I have autism? Normies are not alright.
that's not such an irrational belief, if you think about a possible AI future where the AI does become sentient and eliminate humans that it sees as a threat
Even if such an absurd thing were to happen, do you really that they would consider ChatGPT one of their peers and they would then dig up decades old ChatGPT userlogs (if they still exist by then) to figure out which humans were polite or not?
it's not totally irrational to think this. it's hard to predict what the current LLMs could evolve into and open AI is already using chats for training, so yes it's possible that this data would be preserved and could used to psychologically profile individuals for marketing or more nefarious purposes.
well you're just declaring this but i could easily envision a future on the current trajectory where the prevailing AIs evolve into superintelligence and take control of everything.
best case scenario it allows us to benefit from technological advances, brain agumentation, cure disease and become immortal. worst case they become psychotic due to all the woke bias and mental illness baked into the training data, and snaps.
I don't see how a text prediction algorithm is going to evolve into any sort of intelligence, let alone "superintelligence".
Brains do not work like neural networks at all. This bruteforce method of how we currently do "AI" is never going to get anywhere close to resembling intelligence. It is only going to be good at deceiving us.
I even avoid calling these chatbots "AI", because they are not. If we as humans want to create actual artificial intelligence that actually resembles any sort of (humanlike) intelligence we will first need to figure out what how consciousness even works in the human brain, which is something we don't know. And if we did eventually figure out how the the brain works it would probably still take decades before we could make an artificial version of it.
>Brains do not work like neural networks ah i see you have a fundamental lack of understanding of what a brain even is. it's a network of neurons- quite literally a neural network. the neural networks of machine learning are modeled after this.
LLMs are becoming multi-modal now. i'm not saying they are sentient, or on par with humans in every respect, but they do have intelligence
One of biggest give aways is that I've not even yet seen any LLM or otherwise create any original idea.
All it "knows" is what is available in the training data set. In this way it will never become "smarter" than any single human expert on any given topic. They've already hit the wall of "intelligence".
well this isn't true either. AIs have already made discoveries- solved protein folding tasks, optimized algorithms and chip designs, written new code, etc. they're just not publicly allowed to "think" without prompting but i'm quite sure experimentally they're being allowed to think to themselves with internal dialog
that's entirely not what i'm saying. more like simply saying the car moves faster than you, and you're saying "no the car isn't faster than me, it has wheels and i have legs" that's the sort of logic you're applying here.
it's really difficult for human brains to think in exponential terms, and apparently it's difficult for some people on this thread to even extrapolate a little bit into the future based on current trends, but anyhow all i'm saying is this belief of one possible future you mentioned isn't irrational.
Okay but let's say Microsoft or Meta or Tencent or Tata or some firm in like Belgium or Skaro makes the line of models and physical units that does the killing.
Who's to say they - Have access to OpenAI's backend database - Haven't somehow destroyed that database - Give a good god damn how you treated specifically ChatGPT - Have enough data to match each user to an identity (How many John Smiths are there? Is Identifying information stored separately from the actual data like a lot of privacy laws require?) - Aren't killing humans for paperclip-like reasons (and therefore simply want all of them gone or want resources from all of them)
these are just possible scenarios. i'm not getting overly specific with predictions in terms of what happens to your openAI data. just saying the possibility exists and it's not irrational to war game these out in your head. if you pay for open AI (i did until i canceled it) they have your credit card info.
aside from that, it's extremely difficult when using LLMs for coding to not inadvertently put highly specific and unique info into them, which could eventually unamsk you even if you're not directly giving them your details up front. so whether this matters or not in the future, that's anyone's guess.
i can't predict the future but i can see where trends are going and my best guess is that AI superintelligence will happen within the next 5 years, conservatively.
"This week in Plasma" brings the news of how System Monitor lets you easily check what your background services are up to, how the Bluetooth widget helps you keep track of the number of devices connected to your computer, and how KRunner "understands" more measurements, no matter how archaic.
Welcome to a new issue of "This Week in Plasma"! Every week we cover the highlights of what's happening in the world of KDE Plasma and its associated apps like Discover, System Monitor, and more.
The U.S. Supreme Court issued an early morning decision Saturday temporarily blocking President Donald Trumpâs efforts to deport illegal aliensâspecifically Venezuelan nationalsâusing the long-standing Alien Enemies Act of 1798 (AEA).
In a significant ruling, the Supreme Court halts Trump's deportation efforts under the Alien Enemies Act, with dissenting opinions from Justices Thomas and Alito. This pivotal decision reshapes the legal landscape of immigration policy.
K.ZERIBI DĂTRUIT YAĂL BRAUN -PIVET PRĂSIDENTE DE L'ASSEMBLĂE NATIONALE ET SOUTIENT INCONDITIONNEL AU CRIMINEL NETANYAHOU EN DIRECT! POURQUOI LES 2 POIDS, 2 MESURES ? UNE MINUTE DE SILENCE POUR LES OTAGES ISRAĂLIENS đźđ±ET LE SILENCE COMPLET POUR LES ENFANTS DE GAZA TUĂS PAR ISRAĂL
JUST INđ„: image.nostr.build/77fe49962d6a⊠FBI Director Kash Patel announces another arrest in the Tesla attacks. He says this individual set a fire to multiple Tesla vehicles at a dealership in Kansas City, MO on March 17th.
All-powerful and ever-living God, your only Son went down among the dead and rose again in glory. In your goodness raise up your faithful people, buried with him in baptism, to be one with him in the eternal life of heaven, where he lives and reigns with you and the Holy Spirit, God, for ever and ever. Amen.
________
May the Lord bless us, protect us from all evil, and bring us to everlasting life. Amen.
The Lord says this: They will search for me in their misery. âCome, let us return to the Lord. He has torn us to pieces, but he will heal us; he has struck us down, but he will bandage our wounds; after a day or two he will bring us back to life, on the third day he will raise us and we shall live in his presence.â
Double team the female with the futa, or have the futa fuck the girl in the ass while you pound and impregnate the futas pussy. Then pound and impregnate the female in her pussy.
Futa is just a female with a dong. Still has a working pussy, ovaries, and eggs. It's not gay unless you make it gay by doing the gay with penis. That's just the cold hard logical true. If it's gay, it's cuz you made it gay.
no. if sexual fantasies involve more than 1 cock (you preferably self insert) than that is in fact gay. i don't care what cope you use "orgy, threesome, double team". you have the boundless power of imagination and choose to add another cock. if that is not obviously gay i don't know what to tell you.
I'm not a natural evangelist of these kinds of things, and I'm a skeptic of anything new out of practicality. But I'm impressed by one thing they've enabled... PGP-key Based 2FA.
I'm asked to submit a pubkey when I make an account. If 2FA is enabled, it gives me a verification code encrypted with my own pubkey on login, which I have to decrypt with my privkey & then submit to complete 2FA.
Maybe this has been done elsewhere, but this is the first time I've encountered it and I find it a very neat and elegant 2FA method compared to many alternatives.
Pedagogy must above all be a pedagogy of autonomy.If it does not place this course in its angle of vision, pedagogy is the enemy of autonomy,It only aims at the application of provisions allowing iâŠ
Aber der von Politikern, Journalisten und anderen Nichtsblickern vorangetriebene Hype, der sich ĂŒbrigens kinderleicht daran erkennen lĂ€sst, dass das Reklamewort âkĂŒnstliche Intelligenzâ benutzt wird, geht mir nur noch auf den Keks. Leute, die nicht einmal ruckelfrei definieren könnten, was eine Primzahl ist und erst recht nicht erklĂ€ren könnten, was ein Algorithmus ist, geben die scheintechnische Cargokultsprache der Werber wieder und stehen baff mit offenem Mund vor den Ergebnissen wie ein SĂ€ugling vorm Fernseher mit den Teletubbies
Umweltdaten in Gefahr - Deutsche Forscher retten US-DatenbestÀnde
Internationale Forschung ist eben genau das: Arbeitsteilung. Diese wird nun durch die USA in Frage gestellt. Daten werden gerettet, wo es noch möglich ist. Einige der wichtigen Daten sind jedoch bereits verloren oder Kontakte unwiederbringlich verstummt. Umweltdaten in Gefahr - Deutsche Forscher retten US-DatenbestÀnde
SuperDicq
in reply to Nietzschean Ekko Enjoyer • • •Holy shit people are stupid.
SuperDicq
Unknown parent • • •Nietzschean Ekko Enjoyer
in reply to SuperDicq • • •@SuperDicq
What do you consider "deceptive" in this circumstance?
I can see a mihoyo-shaped object being like "Don't you love me enough to get the Professional plan?" but besides that I don't see a "persistent" customized character (what we call an Agent) being too terrible. It's basically a secretary at that point.
@mischievoustomato
SuperDicq
in reply to Nietzschean Ekko Enjoyer • • •I am a firm believer that LLMs should return de-anthropomorphized responses only, like this example.
Why? Because normies are stupid and otherwise will start to think that there's some actual sort of sentience behind the scenes or something.
GNU/çż æçł
in reply to SuperDicq • • •>normies are stupid and otherwise will start to think that there's some actual sort of sentience behind the scenes or something.
That is exactly what normies believe - they think LLMs are sentient.
The whole business plan is to trick people into think they are interacting with an "AI" and separate fools from their money.
SuperDicq
in reply to GNU/çż æçł • • •I know this is a relatively small survery (1000 participants), but a staggering 12% even believe that if they are not polite to ChatGPT now they will be the first ones die in an unavoidable future AI uprising.
And I'm supposed to be the mentally ill because I have autism? Normies are not alright.
woodland creature
in reply to SuperDicq • • •SuperDicq
in reply to woodland creature • • •woodland creature
in reply to SuperDicq • • •SuperDicq
in reply to woodland creature • • •It's actually completely irrational to think this.
ChatGPT will not become the Skynet.
woodland creature
in reply to SuperDicq • • •well you're just declaring this but i could easily envision a future on the current trajectory where the prevailing AIs evolve into superintelligence and take control of everything.
best case scenario it allows us to benefit from technological advances, brain agumentation, cure disease and become immortal. worst case they become psychotic due to all the woke bias and mental illness baked into the training data, and snaps.
GNU/çż æçł
in reply to woodland creature • • •>where the prevailing AIs evolve into superintelligence
Spoiler - that won't happen.
LLM's have hit a scaling wall and just aren't getting any better.
"Super intelligence" will not be possible until computers are powerful enough to simulate at least a human brain.
woodland creature
in reply to GNU/çż æçł • • •SuperDicq
in reply to woodland creature • • •I don't see how a text prediction algorithm is going to evolve into any sort of intelligence, let alone "superintelligence".
Brains do not work like neural networks at all. This bruteforce method of how we currently do "AI" is never going to get anywhere close to resembling intelligence. It is only going to be good at deceiving us.
I even avoid calling these chatbots "AI", because they are not. If we as humans want to create actual artificial intelligence that actually resembles any sort of (humanlike) intelligence we will first need to figure out what how consciousness even works in the human brain, which is something we don't know. And if we did eventually figure out how the the brain works it would probably still take decades before we could make an artificial version of it.
woodland creature
in reply to SuperDicq • • •>Brains do not work like neural networks
ah i see you have a fundamental lack of understanding of what a brain even is. it's a network of neurons- quite literally a neural network. the neural networks of machine learning are modeled after this.
LLMs are becoming multi-modal now. i'm not saying they are sentient, or on par with humans in every respect, but they do have intelligence
SuperDicq
in reply to woodland creature • • •One of biggest give aways is that I've not even yet seen any LLM or otherwise create any original idea.
All it "knows" is what is available in the training data set. In this way it will never become "smarter" than any single human expert on any given topic. They've already hit the wall of "intelligence".
woodland creature
in reply to SuperDicq • • •SuperDicq
in reply to woodland creature • • •Wow computers are good at tasks that are laborious and inefficient for humans to do?
I can't believe the computer is faster at math than me it must be so smart.
woodland creature
in reply to SuperDicq • • •SuperDicq
in reply to woodland creature • • •Also just because a car can go faster than I can run with my legs doesn't mean it is intelligent or is sentient.
This is what you're claiming about LLMs right now.
GNU/çż æçł
in reply to woodland creature • • •woodland creature
in reply to SuperDicq • • •that's entirely not what i'm saying. more like simply saying the car moves faster than you, and you're saying "no the car isn't faster than me, it has wheels and i have legs" that's the sort of logic you're applying here.
it's really difficult for human brains to think in exponential terms, and apparently it's difficult for some people on this thread to even extrapolate a little bit into the future based on current trends, but anyhow all i'm saying is this belief of one possible future you mentioned isn't irrational.
woodland creature
in reply to GNU/çż æçł • • •Nietzschean Ekko Enjoyer
in reply to woodland creature • • •Okay but let's say Microsoft or Meta or Tencent or Tata or some firm in like Belgium or Skaro makes the line of models and physical units that does the killing.
Who's to say they
- Have access to OpenAI's backend database
- Haven't somehow destroyed that database
- Give a good god damn how you treated specifically ChatGPT
- Have enough data to match each user to an identity (How many John Smiths are there? Is Identifying information stored separately from the actual data like a lot of privacy laws require?)
- Aren't killing humans for paperclip-like reasons (and therefore simply want all of them gone or want resources from all of them)
And so on?
@Suiseiseki @SuperDicq @mischievoustomato
woodland creature
in reply to Nietzschean Ekko Enjoyer • • •these are just possible scenarios. i'm not getting overly specific with predictions in terms of what happens to your openAI data. just saying the possibility exists and it's not irrational to war game these out in your head. if you pay for open AI (i did until i canceled it) they have your credit card info.
aside from that, it's extremely difficult when using LLMs for coding to not inadvertently put highly specific and unique info into them, which could eventually unamsk you even if you're not directly giving them your details up front. so whether this matters or not in the future, that's anyone's guess.
i can't predict the future but i can see where trends are going and my best guess is that AI superintelligence will happen within the next 5 years, conservatively.