The h Bar

Novalium Company

6:00 PM

Yay I am embarrassing myself.

Loong

@NovaliumCompany What's your question?

Novalium Company

ahh... I forgot about that part.

Loong

And don't post text as a picture.

Novalium Company

At least it's a nice clickbait.

Just... I'll see what people think and I'll delete it I guses

ACuriousMind

@NovaliumCompany That's...not something to be proud of if it were true. But to me it looks less "clickbait" and more the equivalent of a physics.SE question titled "My idea for a fundamental theory of everything", which deters users here more than it attracts them.

Novalium Company

6:04 PM

Fine, I agree. I just want to see if my idea has any potential.

@ACuriousMind Btw, someone coming up with general AI is much more likely than a fundamental theory of everything

JMac

@NovaliumCompany Doesn't mean that posting your personal theory is a good SE question.

ACuriousMind

@NovaliumCompany Probably true for "someone" = "expert in the field", probably false for "somone" = "layman".

Novalium Company

@JMac agreed

@ACuriousMind plp don't know Iam layman.

ACuriousMind

Who's 'plp'?

Novalium Company

people.

peolpe.

i'm stupid :D

Abhas Kumar Sinha

6:17 PM

Meanwhile, inside the box, Schrödinger's cat plans its revenge.

3

Aaron Stevens

@AbhasKumarSinha But cats love being in boxes

Abhas Kumar Sinha

@AaronStevens She'll take revenge one day!!

@PM2Ring Really, it's the 7th non rainy season, raining here! :(

Aaron Stevens

7:13 PM

For the first time in a while (or maybe ever) I made a plan to do something in my research and it is actually going smoothly.

ACuriousMind

7:25 PM

that's a dangerous thing to say out loud :P

Aaron Stevens

7:43 PM

@ACuriousMind I understand and accept the consequences of my comment

ACuriousMind

You'll become the next Pauli

Physics Meta

0

Q: Sudden large drop in standing ("top percentage overall") without numeric change in "reputation"

My standing on the Physics Stack Exchange went from "top 16% overall" to "top 67% overall" during a period when I posted only one question or answer (a question posted Jan.5, 2020, MST), plus a few comments. The question has received 0 votes so far, and I hadn't noticed any up-votes on the comme...

discussion

Jake Rose

Any GR people around?

Can you do this?

Nvm

Loong

8:01 PM

Hm

twitter.com/StackOverflow/status/1214274211105198081?s=20

ACuriousMind

gnnnnnh...that's not a bot, is it?

Loong

Doesn't look like a bot.

The twitter bot for physics looks like this: twitter.com/stackphysics

ACuriousMind

Yes, I know that one (from the Community Ads)

Loong

yeah

ACuriousMind

Aha: twitter.com/benpopper/status/1203016391202394112 resp. stackoverflow.blog/2019/12/06/… introduces this. I cannot find any mention of this on mother meta or any of the child science metas

Loong

8:10 PM

ah

Novalium Company

I wrote to Ray Kurzweil to see what he thinks of my opinion that thinking, writing and talking about what happens after general AI is useless. (excluding entertainment purposes)

Semiclassical

About the only firm belief I have re: future tech is that, if a future human finds themselves on a starship, then it's far far far more likely that they're in a simulation than experiencing it in reality

JMac

@NovaliumCompany Isn't that a large focus in Kurzweil's writing?

Semiclassical

(or, to put it another way: the Holodeck is the most realistic part of the Enterprise :P)

Novalium Company

@JMac I don't know. I have read a few books and I don't think he predicts much but I'm just asking his opinion on that topic.

JMac

8:22 PM

@NovaliumCompany I had done quite a bit of research into his ideas like 7 years ago. From what I remember though, making predictions about humans with general AI is like one of the primary themes of his writing. It's not directly based on AI, but AI is built into the timeline and assumptions pretty strongly.

Novalium Company

@JMac Well, it's good that I'm asking him then.

Cuz if he can't prove me wrong, then what the hell...

Semiclassical

"It's tough to make predictions, especially about the future."

Novalium Company

That's not the answer to the question I'm asking him. Tough and useless are different words.

JMac

@NovaliumCompany What I'm pointing out is that it kinda sounds like you're asking him "Do you think that thinking, writing, and talking about the things that you frequently think, write, and talk about is useless?" ... It seems like it might seem like a bit of an insulting question.

Novalium Company

@JMac Ok? I'm pretty sure he doesn't get insulted from the opinions of random dudes on internet.

Semiclassical

8:28 PM

good luck with that, then

Novalium Company

Well, I just hope he answers :P

JMac

@NovaliumCompany If you got an email asking "Do you think everything you're doing is useless?", wouldn't you find it a bit rude? The fact that it's from some random person online would just make me really likely to ignore it, or respond in a pretty negative way.

Novalium Company

@JMac Well, if it's not useless, then let him prove me wrong.

Semiclassical

particularly when you've already acknowledged that you don't know his work that well

lol, what?

ACuriousMind

@NovaliumCompany You can't have it both ways: Either he considers questions from strangers worth reading and answering, in which case he also might be insulted by one insulting his work, or he doesn't. But expecting him to potentially answer but not potentially be insulted is inconsistent.

Semiclassical

8:31 PM

It's not his responsibility to prove you wrong.

JMac

@ACuriousMind Yes, that is a very good way to articulate the feeling I had. Thanks.

Novalium Company

I agree with everything you say. I might insult him, he might not answer, but in the best case, I get a response with his opinion.

ACuriousMind

"In the best case I haven't insulted anyone" is a pretty weak defense. Have you thought about at least trying to not insult people even in the worst case?

JMac

@NovaliumCompany Worst case you've made an enemy with a man who will upload himself into a supercomputer and spend immense computational resources making your life hell.

2

Novalium Company

If he has reasonable amount of emotional intelligence, he won't get insulted. You're making a big deal out of this, it's just an email.

JMac

8:41 PM

@NovaliumCompany Perhaps you're making too small a detail about personal interactions. Yes, most likely he will shrug it off, but that doesn't mean that it was a good way to go about talking to him. How do you know he even has a "reasonable" amount of personal intelligence? For all you know, this may actually upset him, or make him angry. I definitely wouldn't be happy about it. Maybe in a "Can you believe someone actually emailed me this?" sense.

Novalium Company

@JMac While I was writing the email, the thought of insulting him a bit crossed my mind, but it's still worth it. He might learn something from my opinion, he might rethink what he is doing. If he says I'm wrong, I need proof, that's it.

ACuriousMind

Do you realize how entitled that sounds?

Novalium Company

I realize and I feel a bit bad.

ACuriousMind

You sent an unsolicited email to a guy who has written extensively about AI risk, asking him whether that isn't a bit pointless, and expect him to rethink what he's doing? How unreflected do you think other people are if you don't think they haven't asked themselves whether what they're doing is worth it?

7

Novalium Company

Hello, I'm a 17 year old boy from Bulgaria and I've been looking for someone to explain to me why people talk and write about what happens after we create superintelligence (general AI...). I find it useless. I've tried asking in forums and so on but no one can prove me wrong. Who else to ask than Ray Kurzweil himself.
I see all of those people talking and writing about the things that will happen after we create general AI and how we can protect against it and so on... but my opinion is that we can't know what AI will do since we do no think like an AI. I'm not superintelligent, AI will be…

This is the email.

All I want is his opinion. I believe he has asked himself that question and he must have answered it. All I ask is the answer.

@ACuriousMind "How unreflected do you think other people are if you don't think they haven't asked themselves whether what they're doing is worth it?" All I ask is, why is it worth it?

JMac

8:53 PM

@NovaliumCompany "but no one can prove me wrong"; but there was quite a bit of reasoning given to you for why we would want to study the consequences of a general AI, even from users here. This whole attitude of "no one can prove me wrong" really rubs me the wrong way. The way I see it, you've actually gotten quite a bit of perspective on that, and dismissed it anyways.

Novalium Company

@JMac PLEASE, link me an argument, so we can end this.

And I agree that the "no one can prove me wrong" part is a bit assholish.

JMac

@NovaliumCompany Again, there was plenty of discussion on this the other day. If you want to say that writing about post-general AI advancement is useless, please provide your full logical argument first. I'm not convinced that it is useless. You're the one making the claim, so the burden of proof would be on you.

Novalium Company

@JMac "my opinion is that we can't know what AI will do since we do no think like an AI. I'm not superintelligent, AI will be, what's the point of me trying to guess what it will do when I know that I don't have the required cognitive abilities?"

JMac

@NovaliumCompany "my opinion is that we can't know what AI will do since we do no think like an AI" and do you have backup for this opinion? A lot of AI is actually modeled off how real neural networks work. If we model AI thinking after our own thinking, it seems like a tough claim to assert that we don't think like the AI. We obviously wouldn't think identical to it, but it seems like a tough assertion to prove that it won't think like us.

What's the point of trying to understand your parents as an infant? To improve your own cognitive abilities and to be able to better interact with them are two good reasons I can think of. Now replace parents with general AI, and infants with humans.

Even if you don't get immediate benefits from studying it, it can still have a lot of value.

Novalium Company

9:14 PM

The analogy with the parents and infant is irrelevant. Now thinking about what general AI will bring has nothing to do with us learning from it in the future (if that even happens). We have no idea what general AI will be. In fact, it's thinking and functioning might be alien to us. Since we don't know what superintelligence will be, thinking about what it will bring is even more absurd.

JMac

@NovaliumCompany We know it will be superintelligence. That is already something that allows us to propose issues that it may cause, potential complications, etc. Just because we don't know the specifics of it, it doesn't mean we should ignore any of it's general implications, or that considering the general implications has no value.

Novalium Company

@JMac We don't know the specifics nor the generals. A mouse cannot think like a human no matter how hard it tries

We don't know anything about what superintelligence will be like

Making it useless to even attempt to guess what it will do.

JMac

9:31 PM

@NovaliumCompany Depends on what you mean by "think like". I would argue that mice can think like humans, and that's one reason we can do experiments on mice and other animals to help understand cognitive functioning in general. We actually do know things about what superintelligence would be like. We can infer off our own intelligence and make predictions that inform how we proceed as safely as possible. Not knowing everything isn't the same as knowing nothing.

Novalium Company

Forget about the mice analogy, it's incorrect.

We do not know what superintelligence will be like.

It's really late, let's put a pause? I have fracking school tommoro :(

JMac

9:53 PM

@NovaliumCompany Again, not knowing everything is not the same as knowing nothing. We don't know everything about superintelligent AI; but that doesn't mean we don't know anything about it, or that it's pointless to consider the ramifications and potential issues.

Emilio Pisanty

11:59 PM

@Semiclassical CC @NovaliumCompany Indeed. If you haven't already, read this:

195

Q: I believe I have solved a famous open problem. How do I convince people in the field that I am not a crank?

I am interested in the situation where you have a very interesting result. For instance, you have solved a very important open problem. However, you are not known in the field and do not have any remarkable publications. Your supervisor thinks the work is good and you submit the work to a high pr...

publications paper-submission independent-researcher

written for a slightly different situation, but you should pay close heed to the advice as it pertains contacting people who are busy and have no real call or need to read your unsolicited emails.

Transcript for