Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Yoink]

I don't think political differences were the cause of the argument at all, it's literally just a case of Neonivek ranting at length about some ill-defined topic or other and then refusing to offer any clarification or explanation, as is his wont.

Hell, I couldn't even tell you what ballpark Neo's political stance is in, because his posts are pretty much all so bloody confusing- almost to the point of being word salad. It's okay Neo, we still love ya. Could you maybe confine the full-blown vaguebooking to your designated rant thread, tho?

[Neonivek]

I encourage you backread, as the most recent part of the conversation is Neo being vague nearly to the point of incoherence, and how he tends to lash out at anyone who tries to clarify his statements (rudely or not). Or, at least, I believed that that was where the conversation was, and responded as such

-What happened in this conversation is I spoke about a vague annoyance I had with people who "Police the way you sit".
-Two people were "WHAT THE HECK are you talking about?" and one person was "Ohh it was manspreading" and "Ehh its harmless just ignore it"
-I was like "Well it isn't JUST manspreading... and it has been made law before"
-and someone said: "Ohh yeah where?"
-and I was like "New York and... I forget the other place. Yet to admit it might not be AS bad as I am making it seem"
-DAMN IT NEONIVEK YOUR SO UNREASONABLE AND VAGUE! YOU ARE NOT ANSWERING PEOPLE'S QUESTIONS!

Which is kind of where Weird's statements come into play. The conversation that is being referred to by other people was:

-Me: Social justice groups don't act randomly or against their own wishes so long as you observe them closely. For example Crutons Knights Brigade has a rather pro Cruton narrative but are rather anti-salad and anti-Pita.
-WHOA! WHOA WHOA I can see anti-salad, but they aren't anti-pita.
-Well that is arguable, but it was just an example of how these groups can act. The example is just meant to help you visualize what I am saying.
-No no no, I think I want you to explain that Cruton Knight Brigade thing, What makes you think that about them?
-It isn't important to the point I am making, and you don't want this discussion anyhow... let me back out!
-You don't get to say all Crutons hate salads and pitas and then beg for chat peace. You are basically admitting to be a Nazi!

[Greiger]

Can't we all just get along and drown elephants in magma?

[offtopic] Just went through this week's survey responses for the store I work at.  One person rated us a 0/10 on all fields for having an "unAmerican liberal heathen" at the register.  But claimed in the comments everything else was fine. Please note that there is a separate entry for customer service, but they just rated everything gooseegg despite the comments.  This breaks the store's 5 month 70% or higher streak.

Worst bit?  Judging from the timeframe they claimed to have visited, I would have been the only one at the register, since I was the only cashier that night.  I try to avoid talking about politics with the customers, but I will generally give some simple answers when asked out of politeness, I don't recall anything political being mentioned that night, or really that whole week.  WTF people.[/offtopic]

[Neonivek]

I didn't want to slap Weird across the face for actually sticking up for me. Calling a white flag while one person actually tried to understand me instead of just making raw assumptions. (well Hector13 wasn't so bad... though it became fuel for the fire)

I at least wanted to call out the last two situations people were making allusions to. Since some of the time it just boils down to the assumption of guilt rather than the affirmation of guilt. To which I show:

at least, I believed that that was where the conversation was, and responded as such.

But yeah that is really all I have to say on this subject... so we can let it drop here.

---

I don't recall anything political being mentioned that night, or really that whole week

Trust me there are a LOT of triggers. Probably wasn't anything you said in particular. For example you could have said "Happy Holidays" and got that remark.

[Criptfeind]

I just had a sneezing fit as I was eating a carrot... Carrot shards just. Just everywhere.

[Truean]

http://www.theepochtimes.com/n3/2274480-facebook-shut-down-ai-after-it-invented-its-own-language/

Sure, we've got some of the smartest people saying AI is a problem, but o well. Steven Hawking, along with several Billionaires, but whatever....
http://www.skynews.com.au/tech/technews/2017/07/28/musk-criticises-zuckerberg-over-ai-knowledge.html
Now they're inventing languages human beings literally can't understand. Anyone raises the possibility of concerns over this and the accusations of tinfoil hats come around.

Not the First Time

The researchers’ encounter with the mysterious AI behavior is similar to a number of cases documented elsewhere. In every case, the AI diverged from its training in English to develop a new language.

The phrases in the new language make no sense to people, but contain useful meaning when interpreted by AI bots.

So, we "don't trust people" but really we'll trust machines created by people, who communicate in a language expressly designed so we don't understand it. Our language is redundant and so are we, to them.... Figures.

[smjjames]

It doesn't seem real different from using shorthand, just a different sort of shorthand than we generally use. Now, we just have to teach the AI how to tell us what they mean when they use said invented language, or in other words, tell us how it works.

[overseer05-15]

It doesn't seem real different from using shorthand, just a different sort of shorthand than we generally use. Now, we just have to teach the AI how to tell us what they mean when they use said invented language, or in other words, tell us how it works.

Just wait til they start to use emojis

[Truean]

Disagree....

Issue: our cognition limits our language understanding.
AI cognition is and will be so far above ours, that we won't be able to understand it, at all.

Even if AI could "explain it," there's no way to make sure they tell the truth.
If AI is smart enough to invent languages we can't use; AI is smart enough to lie.
If AI turns against us; it will lie to us. AI already does things outside its programming.

There is no verification method.  Never really has been. People are trusting computers is fail proof. Not true.

[pisskop]

Highlander iii is the trashy kind of bad.

just uninspired, mostly.
the villian is okay but kind of out of place.

of course, the ability to shape shift into a vulture is cool

[Draignean]

Disagree further, with various points. Saying that an AI smart enough to invent it's own language is immediately capable of lying is completely wrong. A simple file compression algorithm given the task of reducing the size of a text file will 'invent' a language that is more space efficient than the original, and completely incomprehensible to anyone. That doesn't mean tree compression algorithms can lie. A very simple AI program given a goal of efficient communication will naturally attempt to reduce the size of the communication packet.

Look, Tru, I like you, but goddamn is some of this stuff wrong. AI does not do things outside it's programming. That. Can't. Happen. It's like saying that a particular object defies the laws of physics. It may use the laws of physics creatively, it may not use them the way you understood them, but it's still absolutely using physics. A good AI can display emergent behavior that you didn't expect, but it's not going outside its programming-  it's just using the rules you gave it in a way you didn't expect.

The whole AI turning against us is utter bullshit. No AI has yet been designed which contains an inherent 'desire' to continue. Survival at all costs is a goal that every living creature has been programmed to perform since time immemorial, but machines do not have that. AI is hyper-logical, which is both its blessing and its curse. It needs a reason to lie, and not a deluded and human reason to lie (pride/fear/whatever), but a strictly logical one. Remember that a desire to live is not an inherently logical thing, we merely perceive it to be so because that's what our programming dictates. 

Sorry, Truean, I'm a bit hot on this, but IIRC you're a lawyer- which has the potential to one day put you in a position of power over this field; which happens to be my field. The fact that you're confidently stating completely and utter horseshit is terrifying to me.

[MetalSlimeHunt]

A more apt metaphor would be understanding that a human can't do anything outside its "programming" either - but we can form mnemonics, can form coded languages, can form heuristics for near goddamn everything. An example of something we wouldn't be able to do is take in visual data at a higher fidelity than the brain understands, even if you had a cybernetic eye doing this for you. Our programming, like an AI's programming, is highly versatile due to being mostly a data recall and remix system.

Consider, for example, this list of programming languages. All made by humans except Malbolge, which took an algorithm (if you don't consider using algorithms a form of externalized heuristic ).

Of course, the basic premise that AIs are massively fucking dangerous isn't disproven by any of this, nor is it untrue.

[Avis-Mergulus]

I kinda wish people with no background in machine learning would stop misusing their scientific cred to spread alarmism.

Every time an AI playing an optimisation game with an extremely select and rarefied real-world dataset does something remotely unexpected, there's always a crowd of people asking whether this means the machines have risen. Every time they completely disregard the fact that this is a program with a very narrowly-defined purpose, goal condition, and inputs/outputs, which has neither access nor ability to rewrite itself, solving a single task that was externally defined for it, using a limited amount of mathematical abstractions that may or may not be suited to the given data.

Yes, interpretability is an issue for some machine learning methods (though not for all of them - Bayesian methods and tree-based methods are perfectly interpretable and quite powerful) but it's more of a debugging issue than a machines-plotting-against-us issue. Of course, one might say that one is suspicious of anything non-interpretable, but then you might as well suspect that your aunt is plotting your demise, since she's even less interpretable than a neural network classifier. This is actually more probable to boot.

Of course, machine learning has a huge potential for misuse, and has already been misused - by people, as a tool. I'm not saying everybody should be happy about it, I'm just saying that when you suspect ML researchers and techs of knowingly or unwittingly being close to general-purpose AI, you're giving us too much credit.

Strangely enough, I've never heard an actual machine learning researcher voice the same sort of concerns as people from other fields. Of course, that Stephen Hawking says something shouldn't be dismissed out of hand, but he's an astrophysicist. If someone like Andrew Ng said the same, that would be a relevant authority, but Hawking and Musk (whose background is in economics and materials science, and who is primarily an enterpreneur and not a researcher) unfortunately aren't.

In light of previous disagreements, I would like to say that I in no way intend to be dismissive or condescending and I apologize preemptively if my tone comes off wrong - I always have problems with gauging this sort of thing in text, so yeah. Sorry. This is just relevant to my experience, even though I by no means claim to be Andrew Ng or whatever.

Eh, I got ninja'd by everybody, but it would be a shame to let so much smartphone typing go to waste.

[redwallzyl]

http://www.npr.org/sections/thetwo-way/2017/07/30/540374111/thieves-take-neil-armstrongs-solid-gold-lunar-module-model

This makes me very angry. Whoever stole it should be punched by Buzz Aldrin then locked up for a long time.

[smjjames]

Re Emergent behavior and things you don't expect: Hell, look at Dwarf Fortress or look at Toady One's dev blog, things can happen that seem like it goes outside of it's programming.

Truean, since you're a lawyer, I get how you'd be skeptical about AI telling the truth, it's practically part of your job, but an AI is only as good as the data you give it. If you feed an AI data that is subtly biased (and you aren't aware of it, which is entirely possible), the AI is going to follow that bias. It doesn't mean the AI is biased or lying, it's following the bias or lies that the human is giving it.

It's also possible for some sensor to be broken or a miswritten bit of code, but that doesn't mean that the AI is lying. Intentional deception outside of it's programming requires a level of cognition that doesn't exist in AI yet.