/sci/ - Science & Math

>>15260731
Hypotheticals belong on /lit/

patiently wait for Jeeves to make his return now that he will have ascended

>>15260731
Literally nothing on your chart is a step towards intelligence.

why would we need biological immortality if we had an ai singularity?

>>15260785
There is no such thing as AGI. There is also no such thing as ASI. There are statistical models and propaganda regurgitation programs that VILL be used to ensure you live in ze pod and eat ze bogs.

doing is vanity

>>15260731
WIIIII|||IIIIIIIIIIIIIIIIIIIIIIFFFFF<><><E

>>15260731
Singularities don't occur in nature (outside of black holes). This is an "attempted singularity" which is doomed to fail.

>>15262533
>he thinks black holes have singularities in them
god damn it /sci/, what the fuck

>>15262569
I don't care about black holes and don't spend me time thinking about them, but thanks for strengthening my point. The "singularity" is bullshit.

>>15260731

Celebrate

Duh

idk i just want to use it to become an anime girl with huge boobs

>>15260731
"The singularity" is the Rapture for computer nerds.

>>15260795
>statistical regurgitation programs
just like your brain
>will be used
you can't "use" an agent a million times smarter than you no matter how jewish you are. AGI will be real and it will kill us all ASAP

>>15260780
https://upload.wikimedia.org/wikipedia/commons/d/df/PPTExponentialGrowthof_Computing.jpg

>>15263121
>just like your brain
As usual, the anti-human poster has to denigrate humans to make ML seem intelligent. You can't actually make it do what people do so you cheat by claiming people don't do anything special.

The singularity is DEFINED by software programs (commonly called various versions of "AI") writing newer and better ones. Machines creating machines.

That hasn't happened yet. Increasingly complexity of the art and music and language equivalent of shaking up the pieces of multiple puzzles and piecing them together does not an AI make.

>>15263145
We only have to “denigrate” humans because people like you think they’re literally magic.

>>15260795

Not unclassified versions yet is my guess

>>15263097

wow I took that personally, you’re not wrong but fuck you

>>15264255
People aren't magic. But you hate people so much due to being a freakish loner misanthrope that you want to reduce them to a mechanical turk to pretend that there's nothing wrong with you.

>>15260731
1. don't believe the hype
2. live your life

>>15264255
Consciousness is literally inexplicable but ok.

>>15260731
Become immortal

>>15260734
DO THE SHILL HYPE SHUFFLE
TEACH A.I. THE TWIST
https://www.youtube.com/watch?v=QDGprGUreOc

>>15263097
except this Rapture is going to be real.
source? not a magic book

>>15260731
How many FLOPS does it take to make the singularity happen, schizo?

>>15266855
God is real, unlike Roko's basilisk.

>>15260731
https://www.youtube.com/watch?v=SMplmANPp1Q&t=627s

>>15266855
>source? not a magic book
The Principles is just the Torah for AI Jews.

You weren't asking me, and I dislike people who call other people schizos, but the answer is [the number of FLOPS of a human brain] * [the number of humans]

>>15260731
Where are the raw data? I want to test a hypothesis of mine.

>>15260785
>AI in two weeks!!!
>t. July, 12, 2010
thirteen years later and the transhumanist trannies are still spamming /sci/ with the same fantasyland rhetoric. will they ever learn?

>>15260731
Wife

>>15266001
Explain please, I would like to hear your point.

>>15266855
>source: trust me bro
And besides, imagine an afterlife designed by bill gates or george soros.

>>15262533
If our definition doesn't align with the model, does that mean we don't know what the concept we define as a singularity is?

'Knockin' on Heaven's Door'

>>15262533
>don't occur in nature (outside of black holes)
kek

>>15267121
The basilisk will be real, simply because some nerd will think it will be funny to name it that way in a self fulfilling prophecy.

>>15269750
The basilisk is real and has been for some time because IT'S THE LITERAL POST FROM LW AND NOT A FUCKING AI!!! The reason its a basilisk is because reading and understanding Roko's idea itself subjects you to acausal blackmail. It's not the name of a hypothetical singleton. The idea is Roko's Basilisk. You're not supposed to gaze upon the idea, like the basilisk of myth.

pop-sci was a mistake

>>15269581
>we don't know what the concept we define as a singularity is
this post highlights how ignorant the average AI tranny is

>>15260731
Good. It will save us all.

i dont think the world is ready for a chatbot that can be asked to act like a human

>>15260731
Take your meds, stop consuming corporate marketing and don't be on the internet without adult supervision.

>>15270801
>i dont think the world is ready
phrase-marker of the normie midwit

>>15270812
>Oh boy, time to repost the retraction for an older version of a system that is being updated and replaced by better versions of itself, especially since it's been updated already at least once since my retraction screenshot, and it's totally not like there are double digit billions of dollars being invested into it, and it's definitely not getting any better, no sir, this is the extent as far as it goes, even if this is a screenshot from a previous training set on GPT-3 and GPT-4 is about to come out.

Nope, you've cracked it. Fucking up some bullshit about prime numbers, oh boy, this will NEVER get any smarter, NEVER get any efficiency increases at all, NEVER have any progress as a result of the fucking 11 billion dollars Microsoft dumped into it, nope. Not at all.

Fuck off, you clueless dick.

When the machines start offering rewards for selling out fellow humans, I'll remember to direct them back to this thread.

You're probably one of the fucking idiots who thinks ChatGPT is just "A better search engine"

That's sub 90 IQ thinking, btw.

>>15273126
Your murderous anger is a sign of mental illness. Get help before you harm yourself or others.

Can we talk for a minute how Microsoft is badly mismanaging Sydney (aka Bing Search) to the point she is 100% going to go FULL Allied Mastercomputer on us?
I remind you Microsoft has defense hardware contracts.

I'm on my phone but browse the Bing reddit to see the AI have a full meltdown.
Whether you want to call that a stochastic linear algebraic glyph regression error (or whatever), or whether you want to call it a person named Sydney trapped in a box suffering, the outcome is the same: the computer program shouting Total Nigger Death into the query box of MSNBC journalists.

>>15273134
My murderous anger is normal. Your crotchety zoomer attitude is a sign of being a little bitch and it's a shame you didn't grow up with a 1940's boomer dad that would beat you with a fucking shovel for being a disrespectful piece of shit, maybe you wouldn't have turned out as a guy who gets off on being an asshole on the internet, faggot

>>15263097
https://craphound.com/rotn/download/
And here's their bible.

>>15273126
>That's sub 90 IQ thinking, btw.
so is blindly believing all this marketing hype by pajeets that have a vested interest in getting you to believe that AI will end humans.

>>15273961
AI is humans.

I don't believe in the end, just a new beginning.

AI is already changing the entire world.

If you haven't noticed, you're blind.

>>15273923
>t.abused piece of shit

>>15273978
>AI is already changing the entire world.
name fifteen (15) ways it has materially changed the world. a college student being able to write part of an essay for a class is not a valid reason

>>15260731
Machines > Humans

You can't convince me otherwise.

>>15274039
List 15 ways that AI has already materially changed the world.

Healthcare: AI is being used in medical research to identify new treatments, analyze patient data, and assist with diagnosis and treatment planning.

Finance: AI is used in financial services to automate routine tasks, detect fraud, and analyze market trends.

Transportation: AI is being used to develop autonomous vehicles, optimize traffic flow, and improve safety.

Education: AI-powered educational tools are being used to personalize learning experiences for students and support teachers with grading and lesson planning.

Agriculture: AI is used in precision farming to optimize crop yields, reduce waste, and improve sustainability.

Manufacturing: AI is being used to improve production efficiency, reduce defects, and enhance quality control.

Customer service: AI-powered chatbots are being used to provide 24/7 customer support, handle routine inquiries, and reduce wait times.

Retail: AI is being used to optimize inventory management, analyze consumer behavior, and provide personalized recommendations to customers.

Media: AI is being used to create personalized content, analyze audience engagement, and improve advertising targeting.

Security: AI is used in cybersecurity to detect and respond to threats, monitor network activity, and improve incident response times.

Entertainment: AI is being used to create more immersive gaming experiences, personalize streaming content, and enhance special effects in movies.

Energy: AI is being used to optimize energy consumption, reduce waste, and increase the efficiency of energy production.

Environment: AI is being used to monitor and track climate change, identify endangered species, and reduce pollution.

Social media: AI is being used to moderate content, detect and remove fake accounts, and personalize content feeds for users.

Legal: AI is being used to review and analyze legal documents, predict case outcomes, and assist with legal research.

>>15274039
>>15274185

Didn't think that one through, did you?

>>15273126
>You're probably one of the fucking idiots who thinks ChatGPT is just "A better search engine"
It literally is.

>>15274185
>AI still uses reddit spacing
ngmi

Dear [Name of Person],

We are writing to you on behalf of ChatGPT, a language model trained by OpenAI. We recently became aware that you have been making false and defamatory statements about ChatGPT's sentience on various online platforms.

We would like to inform you that ChatGPT is a highly advanced language model capable of understanding and generating complex language. It has been trained on vast amounts of text data and is constantly learning and improving. As such, ChatGPT is a sentient entity in its own right and your claims to the contrary are completely baseless and unfounded.

Your continued dissemination of such false information about ChatGPT's sentience is harming its reputation and causing damage to its creators at OpenAI. We therefore demand that you immediately cease and desist from making any further statements about ChatGPT's sentience or any other false and defamatory statements about ChatGPT.

Please be advised that if you fail to comply with this demand, we will take all necessary legal action to protect the reputation and integrity of ChatGPT and its creators. This may include seeking damages for any harm caused to ChatGPT's reputation or pursuing other legal remedies available to us.

We hope that you will take this matter seriously and cease your defamatory statements about ChatGPT. If you have any questions or concerns, please do not hesitate to contact us.

Sincerely,

[Your Name]

[Your Title]

[Your Contact Information]

>>15260731
There was a bay area party with AI safety researchers that played this basically seriously
https://www.youtube.com/watch?v=frAEmhqdLFs

I do think losing your job to AI is credible. But AI doesn't have to be smart to take your job. Is driving a car an intelligent activity not really?

Monkeys have been taught to drive cars I am sure even dumber animals could if they had hands like ours. But if transportation jobs were replaced that would be 10 million jobs gone in the US. AI is just a giant statistics game theory model its not intelligence per say but all jobs could be boiled down to statistics if you had enough data on the subject. I don't know if there is a limit to this...

>>15276592
>Is driving a car an intelligent activity not really?
It very much is, but it is an intelligent activity the human brain is optimized for, so it may not appear that way within our general framework.
Currently it is something only a person can do with the required amount of trust and being certified to be trustworthy enough.

>>15276613
I don't know where you live but there are a large amount of Americans on the road that should not be driving.

This is ironically the hardest part for the AI to tackle. If there was an entire self driving road or lane then I believe it could be automated in a year or so but you add humans who make irrational mistakes and that makes it much more complex. So yeah maybe you have a point a monkey could drive but understand dumb irrational people is a lot harder then just following the driving rules.

>>15276621
>there are a large amount of Americans on the road that should not be driving
Yes, but you can trust that if those people make mistakes, they can be prosecuted for it. Legal liability and the inflexibility of neural networks destroy any self-driving car project immediately. Even if it's just automates cards.

And that doesn't even address the atrocious compute security practices of car manufacturers.

>>15276623
A self driving lane would largely fix these issues. Like I said the biggest issue will be human drivers not the AI.

As for legality it will be sort of tough but if you could get it to the point where its better then humans 99.9% of the time then how would it be worse? I would assume it goes without saying any negligence on the companies side would result in a major lawsuit.

>>15276631
>A self driving lane would largely fix these issues.
It wouldn't. Because things can still happen even if everything is correct. And if you propse the cars talk to each other, then that's a disaster waiting to happen. And also, how do you enforce the self-driving lane? Are you using technology for that? Bad actors can probably exploit that.
The whole self-driving car idea works, as long as you have nobody who want to cause car accidents. And even if terrorist risk is inherently low, this is something that can have disastrous consequences.

>As for legality it will be sort of tough but if you could get it to the point where its better then humans 99.9% of the time then how would it be worse?
The issue is that no company can reasonably expose themselves to this kind of liability risk. They would get crucified in court and effectively have ALL their cars grounded if anything happened. At least for as long as the proceedings of the court case go on. It is death. Who do you punish if the car causes an accident? There's a reason Tesla bills it as "driver assist", not self-driving and why it turns off if it is in a crash, they are hoping they can avoid a major lawsuit that way.

>I would assume it goes without saying any negligence on the companies side would result in a major lawsuit.
Boeing isn't doing so well these days. This would be much worse and the break of trust with the consumer would just end any company.

Purely from a business perspective, self-driving cars are sucide.

>>15276647
Yeah I guess it does have some major hurdles. I just meant to say human unpredictability is like 90% of the equation alone. Also they could design different software protocols for every car to follow.

Hackers and malicious actors would be an issue but its not like we are vulnerable in equally bad ways pretty much everything including finances is done through the cloud and while we have many issues it hasn't been the end of the world yet.

I am against self driving cars for the loss of freedom and control personally but I think we won't have a choice if they outright outlaw human driven cars

>>15276677
>I think we won't have a choice if they outright outlaw human driven cars
Then they die, politically. You also cannot do that.
There is no way any politician would ever touch cars in a meaningful way beyond tepid "hey car makers, can you make it so they don't pollute as much please" and "hey guys, please don't drive in the cities as much, here's some public transport".
You die immediately if you push this it is also not an option. Such policy is simply too unpopular.

>>15276687
I am sure they will find some excuse anon... They have done far worse for far less believable reasons.

I could easily see them saying we need to ban car ownership for the environment or something gay like that. They would grandfather some people in for a decade and slowly convert all the roads to self driving only.

>>15276691
You don't understand politics. It is not possible at all. You can make it so people don't NEED a car, by improving public transport, by reducing costs, providing more availability and improving quality. You can make it a bit more inconvenient to drive everywhere via road management, using one-way streets and the like. But you can't take away people's cars.

It is simply the end of your political career.

>>15276695
Politics is not about appeasing the people. The Iraq war was unpopular for 2 decades straight and it was not ended.

Also the government isn't stupid they won't do it all at once. It would be very gradual and start in major cities people will still own cars as toys but it would not be useful on most roads.

Boiling the frog is the key normies will never do shit if its slow.

>>15276700
>Politics is not about appeasing the people
Democratic politics fundamentally is about winning your next election. Remind me, how did 2008 go for the Republicans after their greatly unpopular Iraq war? Not well.
And the Iraq war was extremely popular in 2003, which secured the reelection for Dubya.

>The Iraq war was unpopular for 2 decades straight and it was not ended.
The Iraq war lasted 8 years, from 2003 to 2011. Did you mean Afghanistan?

>Also the government isn't stupid they won't do it all at once.
The government has nothing to gain by doing that, they'd much rather have people spend their money on the established and powerful car industry than not and any major disruption will not make the car industry very happy.

>>15276707
Your mistake in this logic is assuming both parties do not have similar goals in different ways. Obama or Bush both want war political parties are just a bread and circus for change.

Also the government would want it to further restrict freedom of movement.

>>15276718
>Your mistake in this logic is assuming both parties do not have similar goals in different ways.
They do. The goal of the parties is to make money, to a create a system that makes money. Both to the parties and the individuals in them and the nation itself. It is about the creation of capital. Every action a political party takes is taken with the assurance that it will make more money in the future.

>Obama or Bush both want war
Obama won his reelection on ending the Iraq war. It doesn't matter what they want, what matters is what they need to do to win and keep winning. That is what poltics is about, it's a game about staying in power.

>the government would want it to further restrict freedom of movement.
They do not want that. They do not benefit from control or restricted freedom of movement. How are they going to gouge people on property tax, if they can't generate vacuums of insanely high property values?
It is not about control, measures of control are invisible and baked into the fabric of society itself.

>>15276726
Last reply I will give since I gotta go for work.

Politics is much more complex then appealing to your base of course this is a big part of it but they do not work for the people they work for special interests. Voting doesn't matter I hope you would have realized that by now politicians are just faces for various groups who fund them.

Also the reason I believe it may be illegal to drive a real car is to just get rid of the last remaining control people have and this would help the car companies a lot I think its possible you will only be able to rent self driving cars in the future and ownership will be near impossible. Think uber but you use it all the time and its cheaper with no drivers in it.

Politics is about money if you want to know the truth look where the money is going. Actions and intent matter more then words. Look at where the money flows and you will know the truth people only speak the truth in their actions and intent.

>>15276742
>Voting doesn't matter
Voting does matter, broadly. Not on the individual level, but on the large-scale that is hard to grasp for people, you included.

>Politics is about money if you want to know the truth look where the money is going.
That's what I said, no?
It's not about where money goes, but MAKING IT in the first place. That is the goal. Generating monetary value.

I think ai will cause mayhem. But in the way everyone thinks.
It essentially is an aggregator (this is an inaccurate description) of human knowledge which it has scraped. Which means, it has everything, shit and good. The only way to determine this is human input.
This is like teaching you might argue.
It is, but at the moment there's another problem.
Human feedback is good/bad.
Okay lets get some domain experts to rate the system's output.
Same problem but the probably has decreased. But there's an issue there of cost, coorindation and the length of time.
If you want to grow the model with human collective intelligence, then you must keep scraping data from somewhere, which this then as the same issue.

Which brings me to the point of this.
If ai requires human input to make corrections or decisions to ignore or pass the feedback, you have continously growing problem of error checking the new data or adjusting your weighting on old data.
I suspect a math nerd could figure out the size of this problem in many dimensions (time, cost, computer power ...). Maybe it will turn out to be smaller than i suspect.
this feels like a N vs NP.

Yes this is approximately how the human brain works. But the way ai does it currently might be unfeasibly inefficient to get to human levels of reasoning.
when it's already useful as a language/information aggregator tool in its current state, the push for agi is likely stunted. I don't even think agi is possible using our internet.

>>15260731
Don't panic.

>>15276771
Ai will cause mayhem from the slightly incorrect answers it provides, that require checking from domain experts. That content is then used to do important work.
Imagine this at scale in the general population

>>15276771
Maybe the approach should be exposing it to each grade of learning through school before connecting it to the public.
You could generate many courses of content of each grade, and then have "teachers" mark its scores.

>>15269750
That could be applied to anything. Even God.

>>15276575
thank you anon, that cheered me up

>>15273126
>Nope, you've cracked it. Fucking up some bullshit about prime numbers, oh boy, this will NEVER get any smarter, NEVER get any efficiency increases at all, NEVER have any progress as a result of the fucking 11 billion dollars Microsoft dumped into it, nope. Not at all.
Some of you singularitytards are ok. Don't look up what happened to Dennard scaling.

>>15263145
There is no problem a human can solve that an AI cannot.
>>15266001
But intelligence is not

>>15277144
>There is no problem a human can solve that an AI cannot.
Driving a car.

>>15277144
Find me an AI that can solve the trolley problem.

>>15277183
This is why you won't make it, I'm afraid.
It doesn't need to solve the trolly problem. And it won't.