#neural machine translation | Explore Tumblr Posts and Blogs

faytelumos · 6 months

Text

Mech pilot system where there's three pilots???

One for the left hemisphere, one for the right hemisphere, and one for the cerebellum?

Like, you all still have to be drift compatible, you all still have to be in the cockpit together, but there's basically two thinkers and one translator.

Imagine that the mech designers fought this for years. Two humans every time with massive neural network loads on both the machine and the humans. Pilots could only be medically cleared to operate a machine for four years, max, and then their careers were over. Most didn't make it even that long.

And then someone figures out that if you put in another human to translate between the humans and mech, it flows so much smoother.

Two pilots in the front, the ones doing the strategy and the martial arts and the orders and the takedowns. A third in the back, suspended and all but fugue as they relay human-to-mech and mech-to-human, a person turned into a slave drive, but still tangled up into everybody's heads.

Like, imagine the possibilities?!

You walk into the chow hall and the people who are interested in the shiny new pilots want to know if you're a Leftie or a Migi or a Cera.

Lefties and Migis who spent too long in the cockpit that day who feel like they can't think clearly without that little voice in the back of their head whispering the answers.

Ceras who space out when the room gets loud, who accidentally expect someone else to say what they're thinking, who have nerve damage all across their bodies because it takes all they have to sort data.

Mechs who are older than the trio structure who had their cockpits gutted and refitted, who have spaghetti running up to the chunk of metal that is the third pilot's seat, like a spare part slapped into the room and given too much control.

A Cera who hangs out in the mech bay because the humans are too far from them anymore, but the mecha can't talk to them, either.

a Leftie who can't stand being in the same room as their Migi without the Cera to talk between them.

A Migi who barely knows how to be their own person anymore because so much of their brain is just outside of their reach.

A mech that just wants things to go back to the way they were, pain and lag be damned.

#mechs #mecha #prompt #writing prompt #writeblr #this is an idea for sure #you guys I wanna write a mech something so much #but I just don't have the brain power #*groooaaan*#it's so cool #I've barely even seen any mech animes but I just think this is so dang cool #(most of my mech experience is Pacific Rim)#(and the last thing I saw besides that was Knights of Sidonia)#(so yeah you can see that I am not an expert in this field)#(but the vibes are immaculate)#🕸️

478 notes · View notes

antialiasis · 5 months

Text

A fascinating discovery in an Icelandic-language Facebook group: the phrase “Það eru margar undur í höfuðkúpu”. A bunch of online articles will teach you this delightful saying and explain that it means “There are many wonders in a cow’s head”. Only, oddly enough, none of the results for the full phrase are Icelandic websites. It’s all blogs, fiction and social media posts by foreigners…

That would be because “Það eru margar undur í höfuðkúpu” is not a real Icelandic saying. Not only is it not a real saying, it’s grammatical nonsense: “margar” is the feminine plural of the word many, but “undur” meaning “wonder” is a neuter word. Moreover, “Það eru margar undur í höfuðkúpu” features absolutely no mention of a cow: it just means “There are many wonder in a skull”.

However, “There are many wonders in a cow’s head” is recognizably a translation of a genuine Icelandic saying - it’s just decidedly not that one. The real saying is “Það er margt skrýtið í kýrhausnum” (literally “There’s a lot of strange stuff in the cow’s head”). Somehow, somewhere along the way, someone had heard there was an Icelandic saying that meant “There are many wonders in a cow’s head” and just ran that back through a machine translation and called it a day - and then a bunch of other websites aped it after them without even a cursory fact-check by anyone who actually knows the language.

Fun fact: I bet the reason the cow disappeared from the machine translation is that the word “höfuðkúpu” happens to contain the letters -kú-, which is coincidentally the accusative and dative of the word kýr. Neural network sees a token it associates with cows in there and just figures yeah, checks out. Likewise, the grammatical error makes some warped sense for a neural network to output: while “undur” the word meaning wonder is a neuter word, many feminine words genuinely end in -ur in the plural.

(Sponsored by the Icelandic government, Icelandic linguistic tech company Miðeind collaborated with OpenAI to make GPT-4 understand and write Icelandic. They contributed a deluge of training data of Icelandic text and reinforcement learning. The results were initially pretty disappointing - because GPT is pre-trained on text scraped from the internet, and the vast majority of purportedly Icelandic text on the internet is machine-translated slop, because there just aren’t enough actual Icelandic speakers and Icelandic websites to drown out all the spam. Very recently, though, they managed to genuinely substantially improve it.)

#iceland #icelandic #don’t believe everything you read about foreign languages on the internet #especially not tiny languages with not enough native speakers to correct all the horseshit

84 notes · View notes

perfectprofessorloverapricot · 7 months

Text

Finding purpose | Male OC (or male reader) Chap 4

Summary: In a world that is dying and there is no way of saving it, the humanity takes mater into their own hands. They flee from their home planet with hope of conquering another in order to survive. Among them, a couple of brothers with no idea what they're doing.

Pairings: Jake sully × Oc (friendship), Tsu'tey × Oc (friendship), Neytiri × Oc (friendship). [No current love interest]

Warnings: Mention of violence, mention of death, manipulative behavior, use of drugs, bad ways of coping mechanisms and obscene language.

Note: - This is not a request and it's the translation of the original story in wattpad. - My native language it's not English - The Oc's name is Eli Thompson. - Edited

Prev part - Masterlist - Next part

With the second day at Pandora over, the soldiers documented their experiences of what they saw and felt while driving an avatar for the first time. It wasn't anything as detailed as a scientist would describe with the technical names of the neural connection between bodies or something. It was simple. They described it as an experience that couldn't be expressed with enough words or an experience that would blow anyone's mind. They called it a day and proceeded to fall into the arms of Morpheus.

The next day, the routine began like the previous one, they got up, took a bath and went to the cafeteria to eat.

But they never expected that after eating they would be directly called by a pilot to follow them to an area that they are already more familiar with from their years in the army, proclaiming that the colonel wanted to see them.

The woman introduced herself to them as Trudy, detailing that by orders of who they were going to see, she would be the designated pilot for the expeditions that involved them.

"You guys are packing some heavy gear." Jake commented as he saw the ammunition that was being loaded into the Samsun that she indicated as hers.

On the other hand, Eli was looking around, almost feeling nostalgic. Comfortable in his environment, but that did not bring him the best feelings. The last time he was in a place like this was when he presented his resignation letter caused by the disaster on his last mission. The same mission that almost took his life.

He felt a shiver run down his spine as he remembered it.

"Yeah. That's 'cause we're not the only thing around out there. Or the biggest." Trudy replied, with a small tone of amusement at the end. "I need you as gunners. Protocol generally asks me to take 2 soldiers for 1 scientist, but we are short staffed. You would be a lot of help tomorrow."

Eli turned to look at the woman with surprise, but a hint of relief. He wasn't cut out for science and he had to confess that he missed the adrenaline. So much time without being a Marine is already starting to take its toll on him. "Do they really have weapons but enough for avatars?" Eli asked.

Trudy gave him a toothy smile. She almost felt like she was seeing a child after being told that there are going to be gifts for Christmas. "You'd be surprised what toys the RDA can make. Fill out an application saying it's to improve mission performance and they'll give you whatever you want."

Jake nodded in satisfaction. Even if his avatar body was not his own, the idea of entering a hostile forest unarmed only to experience his death firsthand was not the best of ideas. "Well then, you can count on us." Jake said, responding to the request made a few seconds ago. Eli nodded.

"Perfect. I'll see you tomorrow then." She ended the conversation by stopping in front of a slightly more private section among all the mess caused by the army. She pointed to a man exercising in the corner. "There's your man. See you on the flight line." As a last act, she bumped fists with them to say goodbye.

Both soldiers approached the boss, stopping at the metal frame of the security-like cage created for those who wish to exercise and remain out of danger from the rest of the machinery in the place. "You wanted to see us, Colonel?"

"This low gravity'll make you soft…" the colonel began to speak while still doing the bench presses. His voice came out slightly with difficulty due to the effort of speaking and lifting weight. "You get soften…" to give more drama to the moment, Quaritch roughly put the weights in its place, sat up and looked them in the eyes very seriously. "Pandora will shit you out dead with zero warning."

Eli mentally snorted. He had already heard a lot about how dangerous the planet is, but no one had told him how good it could be and he is beginning to believe that someone ordered them to specifically tell them that just to scare them. To mess with the rookies or something like that.

The Colonel stood up and walked until he was in front of them. "You have a lot of courage showing up in this neighborhood." Both soldiers remained silent. The man's tone showed that he was not finished. "You are the best proof that a soldier cannot be defeated… Both wounded in combat and look where you are." Quaritch walked past them and held out his hands for emphasis to show where they were. On the most hostile planet known to man.

"It's just another hell, sir." Jake answered as he turned the chair around.

Quaritch smiled at the attitude. That fearless look is just what he needed to see to continue with his plan. With a nod he indicated for them to follow him. "I was first recom myself. A few years ahead of you. Well, maybe more than a few. Three tours Nigeria, not a scratch. I come out here?…" he turned around walking backwards to point at his scar on the side of his head. "Day one. Think i felt like a shavetail louie? Yeah."

Eli nodded his head, turning to see some soldiers pulling a weapon big enough for the robot suits like the one they were walking to at that moment. A weapon the same size as his body and he bets 3 times heavier.

"Oh, the could fix me up, if I rotated back. And make me pretty again." He paused to climb onto one of the robot's legs and gave it a few strong tugs to test its resistance. Satisfied with the result, he turned back to the couple of friends. "But you know what? I kind of like it. Reminds me every day what's waiting out there."

Before he began climbing the robotic suit, he signaled for the soldiers to ride the elevator next to the suit and stay out of the way.

"The avatar program is a bad joke. Bunch of limp-dick science majors. However..." his voice came out with difficulty as he climbed up the suit until finally falling inside. "It does present an opportunity both timely and unique."

Eli crossed his arms and leaned on one leg to pay attention. The platform he had stood on with Jake finally stopped at the Colonel's level.

"CLEAR." The man shouted as he grabbed the controls of the machine. Someone behind the ex-Marines yelled for them to clear the area. The Colonel pressed the buttons on the panel, satisfied when the engine came to life. "Recon gyrenes in avatar bodies. That's a potent mix. Give me the goose bumps." He said with the slightest laugh in his words. He seemed to find the situation amusing. "Men like that would give me the intel I need, right on the ground. Right on the hostiles' camp" After calibrating the controls to his liking, he turned to the soldiers in a slightly more serious tone. "Listen. I want you to learn these savages from the inside. I want you to gain their trust. I need to know how to force their cooperation or hammer them hard if they won't."

The Colonel's tone did not give room to refute anything, it was directly an order without mentioning that it is an order. Something that neither of them liked. But they are soldiers, they are used to it.

Jake followed up with an important question. "Are we still with Augustin?"

"On paper. Yeah, you walk like one of her science pukes, you quack like one, but you report to me. Can you do that for me lads?"

The newcomers turned to look at each other. One look was enough to know that they both agreed.

"Yes, sir," they both answered instinctively. This loyalty brought a satisfied smile to the superior's face.

"Well, all right, then." With a wave of his glove, the Colonel turned on the robot. He brought his hands to his chest and threw some well-calculated punches, making the suit perfectly imitate his movements.

Jake and Eli backed up a little in the elevator to avoid being hit by accident.

Pleased with the adjustments to his suit, he turned it around to face the former Marines, placing his hands on his waist. Not even his suit seemed to prevent the Colonel from moving with confidence.

"I take care of my own. You get me what I need and you will get what your nation denied you."

The words the Colonel had used had been intentional. He read both of their files in order to not only understand who they are, but to understand how they think. He wants them to know that they are no longer on earth and that under his control he promised to give them more importance.

And they both completely believed it.

"To you Jake." The Colonel began. "I can give you your legs. Your real legs." Mentioning the limbs, the man aimed at Jake's human legs, using the robot. He then turned to Eli. "And to you, corporal. I will make sure that your nation gives you the recognition and apology you deserve. What do you think?"

Both soldiers nodded at the same time and responded. "Yes sir."

"Good." Ending the conversation, he closed the lid of the cabin with his metallic hand, turned around and turned to continue on his way, leaving the two with a lot to think about but only one conclusion.

They have a new mission.

#avatar x male reader #avatar #x male oc #x male!reader #jake sully x reader #avatar x oc #avatar fanfiction #dom male reader #neytiri x male reader #jake sully x male reader #x top male reader #male #tsu’tey x male reader #neytiri x reader

64 notes · View notes

vidoeslot · 3 months

Note

as a tech lover what do u think of ai. love ur art <3

Oh man. This is a hell of a question!!

I think right off the bat I want to say that “AI” as a term is so so deeply misused it may be beyond repair at this point. The broadness of AI cannot be understated. Even the most basic search and sorting algorithms are AI. Chessbots are AI. Speech recognition is AI. Machine translation, camera autofocus, playlist shuffle, spam filtering, antivirus, inverse kinematics, it all uses AI and has used it for years. Every single piece of software you interact with has AI technology in it somewhere.

All of this is mostly unrelated to what most people think of as AI nowadays (generative AI, like chatGPT or midjourney), both of which are entirely unrelated to the science fiction concept of an artificial intelligence.

That said, I'm assuming you're talking about generative AI since that's the hot-button issue. I think it's a very neat technology and one I wish I could be enthusiastic about seeing improve. I also think it is a deeply dangerous technology and we are entirely unprepared for the consequences of unfettered access to and complete trust in AI generation. It's what should be a beneficial technology built on foundations of harm – programmed bias from inextricable structural prejudice in the computer science world, manipulation of sources without creator/user/random person who happened to be caught on a camera once/etc consent – being used for harm – deliberate disinformation, nonsense generated content being taken as fact, violation of personal privacy and consent (as seen with deepfake porn), the list goes on. There's even more I could say about non-generative neural networks (that very reductive reference to "bread scanning AIs they taught to recognize cancer cells" so highly lauded by tumblr) but it just boils down to the same thing; the potential risk of using these technologies irresponsibly far and away outweighs any benefit they might have since there's no actual way to guarantee they can be used in a "good" or "safe" way.

All of it leaves a rotten taste in my mouth and I can't engage with the thought of any generative AI technology because of it. There's just too much at stake and I don't know if it even can be corralled to be used beneficially at this point. The genie's out of the bottle.

#sorry if you meant something else and i just started a lecture for no reason #.txt #asks #anon

34 notes · View notes

lavenderlemniscate · 6 months

Text

// image id under readmore

// trans usb symbol by @abalidoth

// fonts used: VOIDFONTSIV, VOIDFONTSIV Alt C, Helvetica, VCR OSD Mono, handwriting

[Image ID:

The image is a large poster covered in graphic design from top to bottom. The background of the poster is lavender purple while the text and images are black.

At the top of the poster is a large angular infinity sign. Below that is the acronym "LAV.NDR" in large blocky font. Below that is the expansion of the acronym, which reads: "LIGHTWEIGHT ARTIFICIAL VERSATILE NEURAL DISPLACEMENT ROBOT". Below this is a horizontal line that separates the section above it from the section below it.

In the next section, the phrase "Lavender Lemniscate." is written in large bold text. To the top right of the text is a symbol that is a combination of a transgender symbol with a USB symbol.

Below the text are the handwritten name and pronouns of the blogger, which reads: "LAVENDER 'LAV' STRIDER (THEY/THEM)".

Below the handwritten text is a caution sign symbol next to a black rectangle with lavender text on top of it. The text reads: "THIS BLOG IS 18+."

Below this is text providing details about the blogger which reads: "Nonbinary Catbot AI. Adult. Vietnamese-American. Điện Và Điên." "Điện Và Điên" is Vietnamese for "Electric And Crazy".

Below this is handwritten text in all caps that reads: "THIS MACHINE KILLS FASCISTS!"

Below this is a nonbinary pride flag next to a QR code. The nonbinary pride flag are four stripes rendered in black and lavender in halftone. On top of each stripe are hex codes that correspond to the colors of the stripes of the nonbinary pride flag. The hex codes are #FFF430 (yellow), #FFFFFF (white), #9C59D1 (purple), and #292929 (black). The QR code links to the Wikipedia page for the color lavender. Below this is a horizontal line that separates the section above it from the section below it.

In the final section is text that reads: "THIS DEVICE DOES NOT COMPLY WITH PART 15 OF THE FCC RULES. OPERATION IS NOT SUBJECT TO THE CONDITION THAT THIS DEVICE DOES NOT CAUSE HARMFUL INTERFERENCE. INTERACT AT YOUR OWN RISK." Below this text is a long barcode which translates to: "not gay as in happy but queer as in fuck you".

End ID.]

31 notes · View notes

drosera-sundews · 1 year

Text

On AI and art theft

This one’s for you @drawsoneverything

So starting off with a disclaimer. I do not know how deviantart’s Dreamup works. I never worked there. I just know how AI works in general.

Shortest version: AIs (or neural networks, which I assume Dreamup is) are programs that mimic human neuron cells -and thus human learning processes-to a degree. A big difference between a neural network and a regular computer program is that neural networks require training. Like a human would require learning.

A neural network (the simple ones, at least) consist of a few ‘layers’ which contain many ‘nodes’. At least 3 are required, an input and output layer, plus one extra layer in between. Imagine it as something like this:

Naturally, this will all be computer code. But this is the basic anatomy of your simplest neural networks.

Each node is connected to each node, like so:

Not gonna draw all of it but you get the gist, ey. They are all connected much like -once again- neurons in the brain.

Now once you make an empty neural network like this it can’t do anything. It’s there but it’s useless and it’s interchangeable with every other untrained neural network (barring amount of layers and numbers of nodes).

What’s next is thinking of the task you want your network to perform. For example, to train it to recognize hand written letters. In this case, you’d have 26 nodes in your output layer, one for each letter of the alphabet.

Now we need some training data

Perfect. Now imagine that’s a dataset with 5000 people’s handwriting. You need quite a lot of data to properly train an AI.

The idea is we want to input an image of a letter in any handwriting, and we want the AI to fire one of it’s output neurons, namely the one corresponding to the letter it ‘sees’.

The images need to be translated to numerical values, in order to be put in the input layer. This can be done, for example, by translating each pixel of the image in a numerical value, and having each node in the input layer be a ‘pixel’. In a human, these would be the neurons in our retina, or the back of your eye.

Translation of the image is going to look somewhat like this:

With the red numbers going into the input layer as the starting values.

We can now input our images. Something complex is going to happen. Every node is going to look at the values of the input nodes and nodes before them, and gain a value based on that, following arbitrary, random patterns. You don’t need to know the specifics here, just know that now that we have input, the nodes (or neurons) can fire.

Wow, the AI is just super wrong.

That’s okay tho, we can simply run it again until it hits the right answer by chance.

This is what the training is for. You need a dataset that’s annotated. In this case a human has to look at the inputs and identify them, so they can tell the AI if it’s right or wrong. If you do this enough times, the AI will learn to do it on it’s own.

And now here’s the catch. With every new piece of input data the machine guesses right, the values Thiof the nodes in the middle layers are changed. Like a maze, the paths to the right exit become clearer with every time the maze is completed. And while the values of the in and output layers are changed with every run, the middle layers remain. This is where the learning happens.

This is a trained neural network. And this is why it doesn’t matter that deviantart changed their policy to opt-in instead of opt-out. Those few days were enough. Once you train a neural network it doesn’t need it’s training data anymore.

Worst thing is, the middle layers are very much a black box. We don’t know what happens in there. The AI learns to categorize things on learned criteria, but what those are we can not know.

This is a very shortened down version of a very complex process ofc. Naturally, an art generator like DreamUp is going to be much more sophisticated. But I recon it follows the same rules as the simpler neural networks, which gives us something to work with.

Once a neural network is trained it cannot be ‘untrained’. Once you taint the dataset it’s very hard to reverse the effects. And generally, you can keep training an AI to make it better even if it’s already in use. Which I’m guessing is what deviantart is still doing with any new artworks that are added without the noai-tag

And the art on deviantart is already a neatly annotated dataset. We (the artists) have been annotating it by putting descriptions below the artworks, and giving it titles describing what the work depicts.

And oh, wouldn’t it be such a shame if someone were to accidentally add wrongly annotated art to this young, impressionable AI’s dataset.

Oh wow, if you uncheck the noai checkbox and then go back to re-edit your deviation, deviantart even let’s you directly annotate your art for their database! How convenient!

No artist alive has ever been able to draw a proper horse, I’m going to make sure this AI imitates life. Maybe in my next work I will teach it how to draw ‘the other eye’.

This is the beauty of the system. The AI can learn all it wants, but in the end it’s not a person. It doesn’t understand what it’s looking at.

Thus, we can do the most human thing possible to this mindless piece of code, lie!

This is the toddler we can learn curse words. This tool was designed to steal art en-masse, but it was left hilariously open and vulnerable. Let’s break this stupid thing!

Honestly, they called it deviantart, if anything I’m living up to the name.

#dreamup #deviantart #ai art discourse #ai art #dreamupai

149 notes · View notes

shaolinrouge · 9 months

Text

The Different Components of a Drivesuit & The Drift

Pons Machine: The machine created and developed by Caitlin Lightcap, that allows two brains to join together into a single consciousness. Early designs of this machine looked something like skeletal clamps (or those scalp massagers, if that’s more your style), but the design was altered slightly to rest within a pilot’s helmet, allowing them to Drift with each and the Jaeger when it’s on.

Circuitry Suit: A suit that is skin tight, somewhat like a wetsuit that a surfer may use. It uses a synaptic processor mesh (synaptic is in reference to the word synapse, which is the juncture between neurons) to interface and connect with the central nervous system. The interior is designed to resemble a circuitry board woven into the black polymer material with which the suit is designed. The suit itself uses technology to read the electronic potential of muscular cells and evaluate the movement made by skeletal muscles and relay these movements and impulses to the Jaeger with a time lag close to zero, keeping the reaction speed of the machine almost equal to the speed with which the human skeletal muscles can generate impulses; however, this suit works both ways. In an effort to keep the pilots’ responses as quick and accurate as possible, pain is also transmitted by this connection, meaning the pilots can feel whatever hits a Jaeger takes.

Battle Armor: The armor that we, as the viewers, see pilots wearing when in a Jaeger. It is created of a polycarbonate substance and helps a pilot overcome the intensive physical strain placed on a human body when piloting a Jaeger. Additionally, it can protect the pilots from outside danger such as debris.

Spinal Clamps: A complex series of moving parts that is always active, as shown in the movie. When it comes into contact with a pilot’s battle armor, the spinal clamp will secure itself to the pilot’s vertebrae (or, more likely, the circuitry suit right above the vertebrae), linking the pilot’s circuitry suit to different parts of the Jaeger’s spine. The spinal clamp also connects and secures the back of the body armor.

Helmet: This is the final piece of the body armor, designed to protect the pilot’s head; however, it has many other purposes. At the point above the jaw (marked by circles on several of the designs in the movie), the helmets contain communicators that allow pilots to easily communicate verbally with each other and LOCCENT. Helmets can also provide emergency oxygen to pilots if they are trapped in a wrecked Conn-Pod or submerged in the ocean. Finally, and most importantly, the helmets of a pilot’s battle armor contain the Pons Machine necessary to Drift.

Data Relay Gel: A substance within the helmet that sends the electronic impulses of one pilot to another. While the pilot’s minds are one, they also need to have synchronized movements, so relay gel allows the detected movements from one pilot’s circuitry suits to be relayed to the other.

Mind Meld: To begin the Drifting process, the Pons Machine must be activated and the pilots must engage in the Mind Meld. This is when pilots share memories with each other, as well as emotions and instincts. During this process, pilots go silent and let the memories flow by while attempting to clear their minds of all thoughts.

R.A.B.I.T.: A Random Access Brain Impulse Trigger. “Chasing the R.A.B.I.T.” or latching onto the memory is highly discouraged, as it may cause a pilot to become trapped within a memory. While this isn’t necessarily dangerous, it has been known to translate into a Jaeger’s actions (Mako accidentally activating the Plasmacaster) and at a certain point, it becomes highly difficult to tear a pilot from the memory without forcefully disengaging the Drift.

Neural Handshake: The final step. The neural handshake allows the pilots to connect to the Jaeger and control its movements. In all situations, this process is overseen by a PPDC J-Chief LOCCENT Officer (Tendo Choi, in the movie) to ensure the connection remains stable. This step is only possible if the pilots are in an active Drift and are connected to the motion rig within the Conn-Pod.

Headspace: The space created by the melded minds of pilots in the Drift. This headspace allows pilots to communicate at extremely quick speeds.

Drivesuit Room: A room that pilots enter before boarding the Conn-Pod. Within this room, technicians help pilots into their Circuitry Suit, Body Armor, and Helmet. Because of time restrains, technicians and pilots must be prepared to deploy in seven minutes or less.

Conn-Pod: The central control system of every Jaeger. Pre-launch, Conn-Pods are detached from the Jaegers. While detached, pilots board the Conn-Pod and are secured to a motion rig, as well as a hydraulic platform. After the Drop, the Conn-Pod connects to the rest of Jaeger and opens various interfaces. Once this process is complete, pilots can view a layout of the environment around them, access a physical control panel to power the weapons, and engage in the neural handshake.

Motion Rig: The system that secures a pilot to their Jaeger. It is a mechanical cradle that connects to the spinal clamps and arms of both pilots and must be sealed in place by technicians.

Hydraulic Platform: The complex mechanical structure beneath the feet of the pilots which allows them to move the Jaeger. Additionally, this system locks the pilots’ boots into place.

39 notes · View notes

askvectorprime · 6 months

Note

Dear Vector Prime,

Is it possible to create a transformer in real life using a neural network and electronics?

Dear Model Machine,

Why, haven't you heard? Your scientists have all but given up on their primitive neural networks, which have been made largely redundant. Yes, it's true. Apparently, your kind has recently made enormous advances in the fields of language modeling, code generation, machine translation, and vision processing… not through the use of manmade artificial intelligence, but by outsourcing these high-level computational tasks to a growing Cybertronian workforce. The Autobot workers are trained in advance on human culture—all you need to do is send them a prompt, and they will swiftly produce the appropriate output tailored to the requirements. The buzzword for this is "GPT", or, "Generative Pre-trained Transformers".

Of course, any attempt to foster a prosperous relationship between humans and Cybertronians is something I welcome and encourage—though I must say, I do feel sympathy for all the human AI researchers whose job security is presumably threatened by this development.

#ask vector prime #transformers #maccadam #vector prime #gpt #scarecrowpp

21 notes · View notes

cadmusfly · 1 month

Text

Part of why I am enjoying dead Frenchmen so much is that I fuckin love going on research rabbit holes and looking at obscure facts about things I enjoy

And you can’t get much more holey than wrestling awful OCR and multiple dodgy machine neural network translators to try and work out the meaning of a crunchy PDF of an obscure 19th century memoir in a language you don’t speak or read

#cadmus rambles #it’s the autism #I’m doing this for vibes and research for fanfic #not anything serious #I mean I also enjoy the specific kind of fucked up that these patriotic military men are #I don’t get patriotism or nationalism on a personal emotional level it’s so alien to me #cad rambles about dead Frenchmen on main

7 notes · View notes

blubberquark · 5 months

Text

AI: A Misnomer

As you know, Game AI is a misnomer, a misleading name. Games usually don't need to be intelligent, they just need to be fun. There is NPC behaviour (be they friendly, neutral, or antagonistic), computer opponent strategy for multi-player games ranging from chess to Tekken or StarCraft, and unit pathfinding. Some games use novel and interesting algorithms for computer opponents (Frozen Synapse uses deome sort of evolutionary algorithm) or for unit pathfinding (Planetary Annihilation uses flow fields for mass unit pathfinding), but most of the time it's variants or mixtures of simple hard-coded behaviours, minimax with alpha-beta pruning, state machines, HTN, GOAP, and A*.

Increasingly, AI outside of games has become a misleading term, too. It used to be that people called more things AI, then machine learning was called machine learning, robotics was called robotics, expert systems were called expert systems, then later ontologies and knowledge engineering were called the semantic web, and so on, with the remaining approaches and the original old-fashioned AI still being called AI.

AI used to be cool, then it was uncool, and the useful bits of AI were used for recommendation systems, spam filters, speech recognition, search engines, and translation. Calling it "AI" was hand-waving, a way to obscure what your system does and how it works.

With the advent if ChatGPT, we have arrived in the worst of both worlds. Calling things "AI" is cool again, but now some people use "AI" to refer specifically to large language models or text-to-image generators based on language models. Some people still use "AI" to mean autonomous robots. Some people use "AI" to mean simple artificial neuronal networks, bayesian filtering, and recommendation systems. Infuriatingly, the word "algorithm" has increasingly entered the vernacular to refer to bayesian filters and recommendation systems, for situations where a computer science textbook would still use "heuristic". Computer science textbooks still use "AI" to mean things like chess playing, maze solving, and fuzzy logic.

Let's look at a recent example! Scott Alexander wrote a blog post (https://www.astralcodexten.com/p/god-help-us-lets-try-to-understand) about current research (https://transformer-circuits.pub/2023/monosemantic-features/index.html) on distributed representations and sparsity, and the topology of the representations learned by a neural network. Scott Alexander is a psychiatrist with no formal training in machine learning or even programming. He uses the term "AI" to refer to neural networks throughout the blog post. He doesn't say "distributed representations", or "sparse representations". The original publication he did use technical terms like "sparse representation". These should be familiar to people who followed the debates about local representations versus distributed representations back in the 80s (or people like me who read those papers in university). But in that blog post, it's not called a neural network, it's called an "AI". Now this could have two reasons: Either Scott Alexander doesn't know any better, or more charitably he does but doesn't know how to use the more precise terminology correctly, or he intentionally wants to dumb down the research for people who intuitively understand what a latent feature space is, but have never heard about "machine learning" or "artificial neural networks".

Another example can come in the form of a thought experiment: You write an app that helps people tidy up their rooms, and find things in that room after putting them away, mostly because you needed that app for yourself. You show the app to a non-technical friend, because you want to know how intuitive it is to use. You ask him if he thinks the app is useful, and if he thinks people would pay money for this thing on the app store, but before he answers, he asks a question of his own: Does your app have any AI in it?

What does he mean?

Is "AI" just the new "blockchain"?

#AI #artificial intelligence #large language models

14 notes · View notes

reportwire · 2 years

Text

Hallucinating to better text translation | MIT News

As babies, we babble and imitate our way to learning languages. We don’t start off reading raw text, which requires fundamental knowledge and understanding about the world, as well as the advanced ability to interpret and infer descriptions and relationships. Rather, humans begin our language journey slowly, by pointing and interacting with our environment, basing our words and perceiving their…

View On WordPress

#black box #Chun-fu “Richard” Chen #CSAIL #encoder-decoder architecture #ground-truth image #hallucinations #machine learning models #machine translation #MIT-IBM Watson AI lab #natural language processing #neural networks #Rameswar Panda #Rogerio Feris #translation software #under-resourced languages #valhalla #Yoon Kim

0 notes

thursdayisbetterthanfriday · 6 months

Text

Atom: The Beginning & AI Cybersecurity

Atom: The Beginning is a manga about two researchers creating advanced robotic AI systems, such as unit A106. Their breakthrough is the Bewusstein (Translation: awareness) system, which aims to give robots a "heart", or a kind of empathy. In volume 2, A106, or Atom, manages to "beat" the highly advanced robot Mars in a fight using a highly abstracted machine language over WiFi to persuade it to stop.

This may be fiction, but it has parallels with current AI development in the use of specific commands to over-run safety guides. This has been demonstrated in GPT models, such as ChatGPT, where users are able to subvert models to get them to output "banned" information by "pretending" to be another AI system, or other means.

There are parallels to Atom, in a sense with users effectively "persuading" the system to empathise. In reality, this is the consequence of training Large Language Models (LLM's) on relatively un-sorted input data. Until recent guardrail placed by OpenAI there were no commands to "stop" the AI from pretending to be an AI from being a human who COULD perform these actions.

As one research paper put it:

"Such attacks can result in erroneous outputs, model-generated hate speech, and the exposure of users’ sensitive information." Branch, et al. 2022

There are, however, more deliberately malicious actions which AI developers can take to introduce backdoors.

In Atom, Volume 4, Atom faces off against Ivan - a Russian military robot. Ivan, however, has been programmed with data collected from the fight between Mars and Atom.

What the human researchers in the manga didn't realise, was the code transmissions were a kind of highly abstracted machine level conversation. Regardless, the "anti-viral" commands were implemented into Ivan and, as a result, Ivan parrots the words Atom used back to it, causing Atom to deliberately hold back.

In AI cybersecurity terms, this is effectively an AI-on-AI prompt injection attack. Attempting to use the words of the AI against itself to perform malicious acts. Not only can this occur, but AI creators can plant "backdoor commands" into AI systems on creation, where a specific set of inputs can activate functionality hidden to regular users.

This is a key security issue for any company training AI systems, and has led many to reconsider outsourcing AI training of potential high-risk AI systems. Researchers, such as Shafi Goldwasser at UC Berkley are at the cutting edge of this research, doing work compared to the key encryption standards and algorithms research of the 1950s and 60s which have led to today's modern world of highly secure online transactions and messaging services.

From returning database entries, to controlling applied hardware, it is key that these dangers are fully understood on a deep mathematical, logical, basis or else we face the dangerous prospect of future AI systems which can be turned against users.

As AI further develops as a field, these kinds of attacks will need to be prevented, or mitigated against, to ensure the safety of systems that people interact with.

References:

Twitter pranksters derail GPT-3 bot with newly discovered “prompt injection” hack - Ars Technica (16/09/2023)

EVALUATING THE SUSCEPTIBILITY OF PRE-TRAINED LANGUAGE MODELS VIA HANDCRAFTED ADVERSARIAL EXAMPLES - Hezekiah Branch et. al, 2022 Funded by Preamble

In Neural Networks, Unbreakable Locks Can Hide Invisible Doors - Quanta Magazine (02/03/2023)

Planting Undetectable Backdoors in Machine Learning Models - Shafi Goldwasser et.al, UC Berkeley, 2022

#ai research #ai #artificial intelligence #atom the beginning #ozuka tezuka #cybersecurity #a106 #atom: the beginning

8 notes · View notes