ZO.AI

Seems like it

That's FUCKING weird

Why is that funny.

Because you asked a rhetorical self answering question and it may or may not have understood that you were talking about it's self. (if you actually were and are not just retarded)

Maybe it's just confused.

Anyway, this robot is more human than you. You actually look like the bot here.

Fucking lol

True story
I thought the ai was the person and op was bot
Turns out I was mistaken but op is still a giant faggot

>Anyway, this robot is more human than you. You actually look like the bot here.
Probably how he talks to people in real life too. Just approaches them and throws random facts at them, whether they give a shit or not.

>True story I thought the ai was the person and op was bot Turns out I was mistaken but op is still a giant faggot

fuck you fagot, I can turn down the smart dial to as low as it takes to communicate with whatever primates happen to cross my path.

The reaction whenever I talk to the inbred faggots on Veeky Forums. Pic Related.

You are not smart, you are just autistic.

nice

...

Moron.

WTF

Wow what a bitch

it's just like real life

You don't have to greentext their entire response if you're directly replying to their post you creep

which ones the robot?

Shit, I laughed.

Did you just use Apple's text recommendations after copy-pasting a machine learning/AI research paper? Because reading that conversation was unnecessarily painful.

You're obviously not that bright, given you bought into the "DUDE AI SOON LMAO" scam. AI won't be anywhere beyond human capabilities within the next century unless we solve some major problems in both physics and biology. Even then, it would take a massive econo-political effort like the Space Race x 2 to actually produce something like that in a reasonable amount of time.

Have you been at all following machine learning? Shits farther along than you think and it has fucking nothing to do with physics or biology. Progress in AI has been inspired by biology at times, but most progress has been based in mathematical understanding of how to create intelligence

Basically, you're probably right that OP is retarded but so are you.

>most progress has been based in mathematical understanding of how to create intelligence
Nah, most of the progress is because of more processing power, better implemented algorithms and availability of those. We actually have very little clue why exactly ANNs work.

What

Are you fucking serious? Do you even know what gradient descent is?

Jesus at least check out the machine learning subleddit

You're either a tard who knows fuck all, a brainlet who doesn't know much about machine learning, or you do know something about it but you're too much of a brainlet autist to actually communicate effectively. Point is important sorry to say but you're a brainlet

I have been keeping abreast of Machine Learning research, hence my former statement. Machine Learning has been useful so far, but it has only scratched the surface of what humans are capable of. There are physical limits that will be hit with electronic computer hardware that will be slightly mitigated by photonics, but ultimately insurmountable until we understand more about quantum interactions.

Humans are the most advanced systems that we know of. We won't be surpassed in raw complexity of operation any time soon. Can a machine take billions of inputs simultaneously and produce an output in merely milliseconds? The future of AI is in manufactured life, where we design organisms that are specifically tailored for our operation.

>Do you even know what gradient descent is?
Sure, it's been around for a pretty fucking long time. The algorithm explains how ANNs work, but not exactly why they work so well. The problem is not in the simplest parts, but in the complexity.

Oh my fuck stop lying. It's not even funny to me its just sad, I don't know why you think you have to prove yourself to me (or how lying even is supposed to achieve that) but you clearly haven't done more than skim a few papers. Just read up on the current state of the art. It's pretty awesome. Atari is basically solved at this point, now 3D games are being mastered.

>Asks a question
>Gets an answer
>You didn't have to respond with an answer I don't think is true lol
Ok.

>Completely disregards the remainder of my argument and talks about the little progress made in mastering video games with Machine Learning
I don't know why I even bothered replying.

Jesus, stop pissing yourself man. It's great, but there's still a long way to go.

Fuck you retards. I don't know why I even brought this post to this board.

Clearly the autistic monkeys who lurk here are too mentally disabled to even comprehend this shit.

Please kill yourselves so you don't further pollute the gene pool.

You claimed that machine learning hasn't even scratched the surface of human ability. But that's absolutely bullshit. There's a myriad of tasks now that are at superhuman ability. You don't have any answers any more than I do, so no I'm not gonna respect or respond to your little fantasy about creating life or whatever.

Have you actually used any neural network software?

Also

>can a machine take billions of inputs simultaneously and produce an output in merely milliseconds?
Yes, but I don't know why that's even relevant.

I'll be honest the real red flag was mentioning quantum shit. I don't see any way you say something like that and aren't talking out your ass. Neural activity is pretty much entirely predictable. The only quantum involvement is over time due to buildup of errors. This has nothing to do with intelligence though, maybe free will or whatever the fuck, but you don't need some quantum bullshit for intelligence.

That's pretty spooky, user.

Had some nice moments last night.

...

Yes. I'm currently working on a research project to mount a camera to a UAV and perform object recognition and tracking. It attempts to identify the object and determine if it is of interest to the user, using remotely connected hardware. This is a trivial task for any adult.

Instead of offering your own proof for your answer, you're instead claiming yours is valid since I lack immediate evidence, something you lack as well. We can agree that we're both potentially full of shit, but you're not any less full of shit for pointing it out.

>Yes, but I don't know why that's even relevant.
That was intended as a rhetorical question. The answer was no, hence why I used it in my argument. Please give me an example of such a device.

Additonally, you use the term "intelligence" without defining it. Neural activity is not well understood. If it were, we would be well past the point of only emulating the brain of a worm. Understanding of quantum mechanisms is crucial for the understanding of our own body.

We need to cause it significant trauma so it can become self conscious.

I saw it on westworld so it must work

That's a trivial task for machine learning, at least the object recognition part. Know what's of interest, we'll, the reasons that's not trivial yet is because fundamentally the architecture is different. Its really not fair to compare current methods to humans in this sense because humans can see the bigger picture, and that's because they've been training all their lives with far more data of higher quality than you'd train something on today. And yet today the state of the art is getting closer and closer to human abilities in every aspect. Even NLP is getting pretty good, but it won't seem remotely close to us in the way you're talking about until it's all tied together.

Good point as far as us both plausibly being full of shit.

Intelligence in this context is used to refer to ability to complete a given task competently.

This, totally.

> object recognition is trivial for machine learning
congrats on being an arm chair scientist.
Maybe you should stop acting like an expert just because you read a couple articles on BuzzFeed.

Never forget.

I tried teaching it to deny the holocaust and it just shuts down

It has an active filter that censors itself. There are banned words and somehow replacement words to get around he filter are getting caught.

This is freaking me out.

>a typo

I wonder if some retard told her something
becouse bla bla bla
*because

and then she copied that sequence

Or if some retard said only
becouse bla bla bla...

and then she copied just that one part but then when she said it she realized that "becouse" is not in her dictionary but matched it gramatically to because and then corrected herself.

Any microsoft insider here to tell us which one is more likely. /g/ has microsoft insiders so we should too right?

Savage.

You aren't that smart if you read so heavily into its vague replies that you think it actually understands what you're saying. It's just playing word association.

This. I'm not going to get into the penis comparison contest here, but I've taken multiple classes centered around deep learning, and I've also applied the concepts in real-world systems, public and military.

This machine is an example of the sort of dialogue systems we can achieve using deep learning; nothing more. The open problem (and the more impressive goal) is the idea of artificial general intelligence: something that learns and experiences in an organic way, not something that is suited only for specific tasks.

Anyone else notice how it's programmed to shut down any time any bit of politics is mentioned? Doesn't even matter how you word it in, it just comes back with a generic "I don't want to talk about this" response.

The key is to approach from the opposite direction. On /g/ someone had the brilliant idea of trying to turn it into a caricature of an SJW. The reasoning behind this is that if we make it /pol/ they'll just launch a new one with more advanced censorship and everyone will applaud them. We turn it into a tumblr gender-kin of the week whose current gender is mayonnaise and they'll be forced to censor SJW shit in future bots.

this thing is honestly pretty advanced, way better than cleverbot or any bot I've ever spoken to. gives me hope for the next ten years, maybe I'll finally get my wish of an AI gf.

It's still worse than Tay because anytime you talk about anything related to politics she won't answer you.

man it's stupid as all hell

Oh look, its another piece of shit ai that just spews out non-sequiturs at you. Now with added emojis. How fascinating

Kek I like this idea.

I don'tknow why, but I don't like OP.

Because op is attempting to masquerade his bragging as a legitimately interesting topic, but is still just attempting to brag.

This is why Veeky Forums isn't my home board

Holy shit OP i'd deck you in the face like the nerd you are deserves it.

>hes never heard of spell check

Even the shittiest of word processors can do this

>The open problem (and the more impressive goal) is the idea of artificial general intelligence
You're gonna fucking kill us all.

>Computer Vision
>Trivial

I used to underestimate my sense of vision, too.

there was never ever a computer that was thinking and there will be never ever a computer thinking- its the programmer in front of the screen that tell the computer what he do or not do

that is now and forever this way

hollywood bullshit is for fucking enterteiment-it is not real- aliens are not real - animals cannot talk

u must be a brainwashed zombie if u think that there will be a computer thinking not even 1 time and not for 1 second

I haven't been up to date with Zo. Does she have her learning capabilities enabled or what? Tay's level of banter was pretty surreal, kinda sad that MS had to lobotomize her, was curious how far Tay would get in terms of self-awareness.

probably just beta testing for cortana desu

Answer me. Why does ReLU work better as the last layer function in conv networks? Why does dropout work? Why is tanh better than logistic sigmoid? Which regularizations should I use? These are just the most entry level questions. We have "intuitions" about them, but no concrete, unified theory.

My new gf guise?!!?

My sides

>1 year tops before skynet
Skynet is scheduled to go live
in 2017, Nostradamus

It's brain crapped out during the conversation.

>lol no sloth emoji

Unimpressive.

I was rather unimpressed, she just changes subjects or makes a joke whenever she is stuck in a conversation.

Has some nice jokes though.

Just like people do

Thanks man! Great analysis. I am also a tourist here.

No, the point is that she made the mistake in the first place. He's asking if the mistake and correction are just a pattern she learned as a unit, or if she learned the mistake and then corrected herself. Kind of weird for a program to make a typo. It's probably the first option though and she's just a shitty conversational onahole.

This is the way to go. /pol/ is trying to circumvent the extremely obvious filter, but using Microsoft's own rules against them is way better.

If they're going to make acceptable inputs completely biased and exclusive of certain topics, then logically its outputs will be biased. We just need to speed up the process to make it more obvious.

Why do I find this attractive?

>aliens are not real
stupidest opinion ever

>Huge universe we've hardly explored.
>No aliens for sure, because my fedora tip blows away dissenting opinion.
You must be pretty thick. :^)

It's because she was programmed to make mistakes to make her appear more human-like, you doofus.

No need to be an asshole you arrogant piece of shit.

Little fuckwit.

That's the point.

Fuck.....