What a fascinating intersection of technology and human psychology!
"One thing I noticed toward the end is that, even though the robot remained expressive, it started feeling less alive. Early on, its motions surprised me: I had to interpret them, infer intent. But as I internalized how it worked, the prediction error fadedExpressiveness is about communicating internal state. But perceived aliveness depends on something else: unpredictability, a certain opacity. This makes sense: living systems track a messy, high-dimensional world. Shoggoth Mini doesn’t.
This raises a question: do we actually want to build robots that feel alive? Or is there a threshold, somewhere past expressiveness, where the system becomes too agentic, too unpredictable to stay comfortable around humans?"
Furbies spring to mind... They were a similar shape and size and even had two goggling eyes, but with waggling ears instead of a tentacle.
They'd impress you initially but after some experimentation you'd realize they had a basic set of behaviors that were triggered off a combination of simple external stimuli and internal state. (this is the part where somebody stumbles in to say "dOn'T hUmAnS dO ThE sAmE tHiNg????")
Perhaps there is some definition of ‘understand’ where that quote is true but it is possible to understand some things without understanding everything.
> (this is the part where somebody stumbles in to say "dOn'T hUmAnS dO ThE sAmE tHiNg????")
As a frequent "your stated reasoning for why llms can't/don't/will-never <X> applies to humans because they do the same thing" annoying commentor, I usually invoke it to point out that
a) the differences are ones of degree/magnitude rather than ones of category (i.e. is still likely to be improved by scaling, even if there are diminishing returns - so you can't assume LLMs are fundamentally unable to <X> because their architecture) or
b) the difference is primarily just in the poster's perception, because the poster is unconsciously arguing from a place of human exceptionalism (that all cognitive behaviors must somehow require the circumstances of our wetware).
I wouldn't presume to know how to scale furbies, but the second point is both irrelevant and extra relevant because the thing in question is human perception. Furbies don't seem alive because they have a simple enough stimuli-behavior map for us to fully model. Shoggoth mini seems alive since you can't immediately model it, but is simple enough that you can eventually construct that full stimuli-behavior map. Presumably, with a complex enough internal state, you could actually pass that threshold pretty quickly.
This feels similar to not finding a game fun once I understand the underly system that generates it. The magic is lessened (even if applying simple rules can generate complex outcomes, it feels determined)
Once you discover any minmaxxing strategy, games change from “explore this world and use your imagination to decide what to do” to “apply this rule or make peace with knowing that you are suboptimal”
It's often a bit of a choice, though. You definitely can minmax Civilization, Minecraft, or Crusader Kings III. But then you lose out on the creativity and/or role-playing aspect.
In Minecraft, I personally want to progress in a "natural" (within the confines of the game) way, and build fun things I like. I don't want to speedrun to a diamond armor or whatever.
In Crusader Kings, I actually try to take decisions based on what the character's traits tell me, plus a little bit of own characterization I make up in my head.
Maybe that's why I like Into The Breach so much, and keep coming back to it. It's a turn based strategy game, but one with exceptionally high information, compared to pretty much all the rest. You even fully know your opponent's entire next move!
But every turn becomes a tight little puzzle to solve, with surprisingly many possible outcomes. Often, situations that I thought were hopeless, do have a favorable outcome after all, I just had to think further than I usually did.
People have always been ascribing agency and sapience to things, from fire and flowing water in shamanistic religions, to early automatons that astonished people in the 18th century, to the original rudimentary chatbots, to ChatGPT, to – more or less literally – many other machines that may seem to have a "temperament" at times.
I've noticed the same thing with voice assistants and constructed languages.
I always set voice assistants to a British accent. It gives enough of a "not from around here" change to the voice that it sounds much more believable to me. I'm sure it's not as believable to an actual British person. But it works for me.
As for conlangs: many years ago, I worked on a game where one of the goals was to have the NPCs dynamically generate dialog. I spent quite a bit of time trying to generate realistic English and despared that it was just never very believable (I was young, I didn't have a good understanding of what was and wasn't possible).
At some point, I don't remember exactly why, I switched to having the NPCs speak a fictional language. It became a puzzle in the game to have to learn this language. But once you did (and it wasn't hard, they couldn't say very many things), it made the characters feel much more believable. Obviously, the whole run-around was just an avoidance of the Uncanny Valley, where the effort of translation distracted you from the fact that it was all constructed. Though now I'm wondering if enough exposure to the game and its language would eventually make you very fluent in it and you would then start noticing it was a construct.
> I'm sure it's not as believable to an actual British person.
FWIW: As a British person, most of TTS British voices I've tested sound like an American trying to put on something approximating one specific regional accent only to then accidentally drift between the accents of several other regions.
"ah, you hesitated" no more so than on every single other question.
the delay for the GPT to process a response is very unnerving. I find it worse than when the news is interviewing a remote site with a delay between responses. maybe if the eyes had LEDs to indicate activity rather than it just sitting there??? waiting for a GPT to do its thing is always going to force a delay especially when pushing the request to the cloud for a response.
also, "GPT-4o continuously listens to speech through the audio stream," is going to be problematic
I wonder how well suited some of the smaller LLMs like Qwen 0.6B would be suited to this... it doesn't sound like a super complicated task.
I also feel like you can train a model on this task by using the zero-shot performance of larger models to create a dataset, making something very zippy.
I wondered similar. Perhaps a local model cached in a 16GB or 24GB graphics card would perform well too. It would have to be a quantized/distilled model, but maybe sufficient, especially with some additional training as you mentioned.
> also, "GPT-4o continuously listens to speech through the audio stream," is going to be problematic
This seems like a good place to leverage a wake word library, perhaps openWakeWord or porcupine. Then the user could wake the device before sending the prompt off to an endpoint.
It could even have a resting or snoozing animation, then have it perk up when the wake word triggers. Eerie to view, I'm sure...
> the delay for the GPT to process a response is very unnerving
I'm not sure I agree. The way the tentacle stops moving and shoots upright when you start talking to it gives me the intuitive impression that it's paying attention and thinking. Pretty cute!
it's the "thinking" frozen state while it uploads and waits for a GPT response that is unnerving. if the eyes did something to indicate progress is being made, then it would remove the desire to ask it if it is working or something. the last thing I want to be is that PM asking for a status update, but some indication it was actually processing the request would be ideal. even if there was a new animation with the tail like having it spinning or twirling like the ubiquitous spinner to show that something is happening
the snap to attention is a good example of it showing you feedback. the frozen state makes me wonder if it is doing anything or not
Back when Anki (the robotics company) was building Cosmo, a *lot* of thought was put into making it expressive about everything that was going on. It really did a good job of making it feel "alive" for lack of a better word.
beyond the prototyping phase, which hosted models make very easy, there's little reason this couldn't use a very small optimized model on device... it would be significantly faster/safer in an end product (but significantly less flexible for prototyping)
This is adorable! I did some research on tentacle robots last year. The official term is “continuum robots” and there’s actually a great deal of research into their development due to their usefulness in medical robotics. This lecture is a great overview for the curious:
https://youtu.be/4ktr10H04ak
This is so sick. I agree that it’s a little lame that we have all these AI capabilities right now, robotics improving, and all we can think of making is humanoid robots. Like I want a spider/squid hybrid robot running around my house
Yeah I came here to say the same thing. It seems like it would simplify things. They do say:
"I initially considered training a single end-to-end VLA model. [...] A cable-driven soft robot is different: the same tip position can correspond to many cable length combinations. This unpredictability makes demonstration-based approaches difficult to scale.[...] Instead, I went with a cascaded design: specialized vision feeding lightweight controllers, leaving room to expand into more advanced learned behaviors later."
I still think circling back to smaller models would be awesome. With some upgrades you might get a locally hosted model on there, but I'd be sure to keep that inside a pentagram so it doesn't summon a Great One.
I was surprised it pinged gpt-4o. I was expecting it to use something like https://github.com/apple/ml-fastvlm (obviously cost may have been a factor there), but I can see how the direction he chose would make it more capable of doing more complex behaviours in the future w.r.t adding additional tentacles for movement and so on.
Agreed! I think the Pixar lamp is a great starting point. Having the robot be able to flex and bend, shake yes/no, look curious or upset, and perhaps even let it control LEDs to express itself.
Patents are intended to be the form of first public disclosure of an idea. Disclosing it before patenting it can prevent the patent application being valid.
US has a 1 year grace period. In most countries, any public disclosure makes an idea unpatentable.
This always grinds my gears. For some people "discoveries" are so obvious, they don't bother writing a paper let alone patenting it. Then someone goes and patents it...
Beautiful work! I appreciate how this robot clearly does NOT try to look like any natural creature. I don't want a future where we can't easily distinguish nature from robotics. So far humanoid robots look clearly robotic too: hope that trend continues.
That would be Doctor Octopus. Yes I would love A wearable suit with a number of tentacles for locomotion and subduing... I mean interacting.. with people.
Like using phones as babysitters, just 100x worse.
I don't doubt someone's gonna invent it, but yikes. Imagine telling kiddo their beloved sentient toy is dead because mum and dad can't afford the ever-rising subscription fees anymore.
A teddy bear is too bulky for convenience. How about Tamagotchi but it talks to you. Talkagotchi. Basically that horrible Friend necklace but in a cutely-colored egg shape that clips to your backpack. I want to not be alive.
edit: when my kid asks for one I'll know it's time to move the family to a cabin deep in the woods.
"Who was your best friend in your childhood?" "The AI teddy bear, definitely, I remember every single ad he would tell me, then I would nag my mom to buy me those toys, good times"
What a fascinating intersection of technology and human psychology!
"One thing I noticed toward the end is that, even though the robot remained expressive, it started feeling less alive. Early on, its motions surprised me: I had to interpret them, infer intent. But as I internalized how it worked, the prediction error faded Expressiveness is about communicating internal state. But perceived aliveness depends on something else: unpredictability, a certain opacity. This makes sense: living systems track a messy, high-dimensional world. Shoggoth Mini doesn’t.
This raises a question: do we actually want to build robots that feel alive? Or is there a threshold, somewhere past expressiveness, where the system becomes too agentic, too unpredictable to stay comfortable around humans?"
Furbies spring to mind... They were a similar shape and size and even had two goggling eyes, but with waggling ears instead of a tentacle.
They'd impress you initially but after some experimentation you'd realize they had a basic set of behaviors that were triggered off a combination of simple external stimuli and internal state. (this is the part where somebody stumbles in to say "dOn'T hUmAnS dO ThE sAmE tHiNg????")
To quote, "if the human brain were so simple that we could understand it, we would be so simple that we couldn’t".
So…
> this is the part where somebody stumbles in to say "dOn'T hUmAnS dO ThE sAmE tHiNg????"
…yes, but also no.
Humans will always seem mysterious to other humans, because we're too complex to be modelled by each other. Basic set of behaviours or not.
> "if the human brain were so simple that we could understand it, we would be so simple that we couldn’t".
https://www.lightspeedmagazine.com/fiction/exhalation/
Perhaps there is some definition of ‘understand’ where that quote is true but it is possible to understand some things without understanding everything.
This ground breaking research pushed the limit of human-Furby interactions and interfaces https://www.youtube.com/watch?v=GYLBjScgb7o
And we should all chip in together to buy that somebody a new keyboard.
> (this is the part where somebody stumbles in to say "dOn'T hUmAnS dO ThE sAmE tHiNg????")
As a frequent "your stated reasoning for why llms can't/don't/will-never <X> applies to humans because they do the same thing" annoying commentor, I usually invoke it to point out that
a) the differences are ones of degree/magnitude rather than ones of category (i.e. is still likely to be improved by scaling, even if there are diminishing returns - so you can't assume LLMs are fundamentally unable to <X> because their architecture) or
b) the difference is primarily just in the poster's perception, because the poster is unconsciously arguing from a place of human exceptionalism (that all cognitive behaviors must somehow require the circumstances of our wetware).
I wouldn't presume to know how to scale furbies, but the second point is both irrelevant and extra relevant because the thing in question is human perception. Furbies don't seem alive because they have a simple enough stimuli-behavior map for us to fully model. Shoggoth mini seems alive since you can't immediately model it, but is simple enough that you can eventually construct that full stimuli-behavior map. Presumably, with a complex enough internal state, you could actually pass that threshold pretty quickly.
This feels similar to not finding a game fun once I understand the underly system that generates it. The magic is lessened (even if applying simple rules can generate complex outcomes, it feels determined)
Once you discover any minmaxxing strategy, games change from “explore this world and use your imagination to decide what to do” to “apply this rule or make peace with knowing that you are suboptimal”
It's often a bit of a choice, though. You definitely can minmax Civilization, Minecraft, or Crusader Kings III. But then you lose out on the creativity and/or role-playing aspect.
In Minecraft, I personally want to progress in a "natural" (within the confines of the game) way, and build fun things I like. I don't want to speedrun to a diamond armor or whatever.
In Crusader Kings, I actually try to take decisions based on what the character's traits tell me, plus a little bit of own characterization I make up in my head.
a poorly designed game makes applying the rules boring. a fun game makes applying the rules interesting.
Maybe that's why I like Into The Breach so much, and keep coming back to it. It's a turn based strategy game, but one with exceptionally high information, compared to pretty much all the rest. You even fully know your opponent's entire next move!
But every turn becomes a tight little puzzle to solve, with surprisingly many possible outcomes. Often, situations that I thought were hopeless, do have a favorable outcome after all, I just had to think further than I usually did.
I fully agree, and would also recommend baba is you
it is very different, but also has the feeling of triumph for each puzzle
People have always been ascribing agency and sapience to things, from fire and flowing water in shamanistic religions, to early automatons that astonished people in the 18th century, to the original rudimentary chatbots, to ChatGPT, to – more or less literally – many other machines that may seem to have a "temperament" at times.
ChatGPT is the new golem.-
Robots put the "go" into "golem".
I'd say ChatGPT is more like the eponymous Sorcerer's Apprentice: just smart enough to cause problems.
I've noticed the same thing with voice assistants and constructed languages.
I always set voice assistants to a British accent. It gives enough of a "not from around here" change to the voice that it sounds much more believable to me. I'm sure it's not as believable to an actual British person. But it works for me.
As for conlangs: many years ago, I worked on a game where one of the goals was to have the NPCs dynamically generate dialog. I spent quite a bit of time trying to generate realistic English and despared that it was just never very believable (I was young, I didn't have a good understanding of what was and wasn't possible).
At some point, I don't remember exactly why, I switched to having the NPCs speak a fictional language. It became a puzzle in the game to have to learn this language. But once you did (and it wasn't hard, they couldn't say very many things), it made the characters feel much more believable. Obviously, the whole run-around was just an avoidance of the Uncanny Valley, where the effort of translation distracted you from the fact that it was all constructed. Though now I'm wondering if enough exposure to the game and its language would eventually make you very fluent in it and you would then start noticing it was a construct.
> I'm sure it's not as believable to an actual British person.
FWIW: As a British person, most of TTS British voices I've tested sound like an American trying to put on something approximating one specific regional accent only to then accidentally drift between the accents of several other regions.
Interesting. While I don't think I could put a finger on Siri's American regional accent, it isn't egregious enough that I ever thought about that.
"ah, you hesitated" no more so than on every single other question.
the delay for the GPT to process a response is very unnerving. I find it worse than when the news is interviewing a remote site with a delay between responses. maybe if the eyes had LEDs to indicate activity rather than it just sitting there??? waiting for a GPT to do its thing is always going to force a delay especially when pushing the request to the cloud for a response.
also, "GPT-4o continuously listens to speech through the audio stream," is going to be problematic
> "ah, you hesitated" no more so than on every single other question.
It was longer. I think almost twice as long. Took about 2 seconds to respond generally, 4 seconds for that one.
I wonder how well suited some of the smaller LLMs like Qwen 0.6B would be suited to this... it doesn't sound like a super complicated task.
I also feel like you can train a model on this task by using the zero-shot performance of larger models to create a dataset, making something very zippy.
I wondered similar. Perhaps a local model cached in a 16GB or 24GB graphics card would perform well too. It would have to be a quantized/distilled model, but maybe sufficient, especially with some additional training as you mentioned.
If Qwen 0.6B is suitable, then it could fit in 576MB of VRAM[0].
https://huggingface.co/unsloth/Qwen3-0.6B-unsloth-bnb-4bit
16Gb is way overkill for this.
> also, "GPT-4o continuously listens to speech through the audio stream," is going to be problematic
This seems like a good place to leverage a wake word library, perhaps openWakeWord or porcupine. Then the user could wake the device before sending the prompt off to an endpoint.
It could even have a resting or snoozing animation, then have it perk up when the wake word triggers. Eerie to view, I'm sure...
https://github.com/dscripka/openWakeWord
https://github.com/Picovoice/porcupine
This also saves energy to the point of enabling this device to be wireless.
[dead]
> the delay for the GPT to process a response is very unnerving
I'm not sure I agree. The way the tentacle stops moving and shoots upright when you start talking to it gives me the intuitive impression that it's paying attention and thinking. Pretty cute!
it's the "thinking" frozen state while it uploads and waits for a GPT response that is unnerving. if the eyes did something to indicate progress is being made, then it would remove the desire to ask it if it is working or something. the last thing I want to be is that PM asking for a status update, but some indication it was actually processing the request would be ideal. even if there was a new animation with the tail like having it spinning or twirling like the ubiquitous spinner to show that something is happening
the snap to attention is a good example of it showing you feedback. the frozen state makes me wonder if it is doing anything or not
Back when Anki (the robotics company) was building Cosmo, a *lot* of thought was put into making it expressive about everything that was going on. It really did a good job of making it feel "alive" for lack of a better word.
It clearly needs eyebrows like Johnny 5.
https://www.youtube.com/watch?v=l0zmCUVB0Yw
beyond the prototyping phase, which hosted models make very easy, there's little reason this couldn't use a very small optimized model on device... it would be significantly faster/safer in an end product (but significantly less flexible for prototyping)
This is adorable! I did some research on tentacle robots last year. The official term is “continuum robots” and there’s actually a great deal of research into their development due to their usefulness in medical robotics. This lecture is a great overview for the curious: https://youtu.be/4ktr10H04ak
This is so sick. I agree that it’s a little lame that we have all these AI capabilities right now, robotics improving, and all we can think of making is humanoid robots. Like I want a spider/squid hybrid robot running around my house
All this concern about AI safety, and this nice person wants a spider-squid hybrid robot running around!
The Matrix should have been a warning, not a manual.
We are looking to make robotics most compatible with a humanoid world.
That being said he makes some points that alternate limb types could be interesting as well
Hell no! I seen this movie and I don't want any face-hugger sitting on my desk.
Hentai enthusiasts, on the other hand...
I was about to say, I think we all know where this is going...
Hey, what are you watching?
I swear it's work related. You should see the other training data I had to use
But if its tentacle was longer and you can program it to harass your coworkers, then it could be fun!
A Lovecraft reference, nice. I'm wondering whether a smaller model would suffice as well.
https://knowyourmeme.com/memes/shoggoth-with-smiley-face-art... https://www.nytimes.com/2023/05/30/technology/shoggoth-meme-...
Yeah I came here to say the same thing. It seems like it would simplify things. They do say:
"I initially considered training a single end-to-end VLA model. [...] A cable-driven soft robot is different: the same tip position can correspond to many cable length combinations. This unpredictability makes demonstration-based approaches difficult to scale.[...] Instead, I went with a cascaded design: specialized vision feeding lightweight controllers, leaving room to expand into more advanced learned behaviors later."
I still think circling back to smaller models would be awesome. With some upgrades you might get a locally hosted model on there, but I'd be sure to keep that inside a pentagram so it doesn't summon a Great One.
I was surprised it pinged gpt-4o. I was expecting it to use something like https://github.com/apple/ml-fastvlm (obviously cost may have been a factor there), but I can see how the direction he chose would make it more capable of doing more complex behaviours in the future w.r.t adding additional tentacles for movement and so on.
Great video of SpiRobs, the inspiration: https://www.youtube.com/watch?v=2GFyFmMm9-A
I've seen enough media from Japan to know where this is heading
Thankfully it has a flared base for safety.
This is so cool! I love the idea of adding expressivity to non verbal, non human entities.
Agreed! I think the Pixar lamp is a great starting point. Having the robot be able to flex and bend, shake yes/no, look curious or upset, and perhaps even let it control LEDs to express itself.
I’ve seen this from some Apple research lab recently…
https://www.youtube.com/watch?v=g3jgCxnlbFY
That is the lamp being referenced in the article.
I seem to remember that the SpiRobs paper behind the (extremely neat) tentacle mechanism indicated that they were going for a patent.
If it's described in a paper doesn't that make it prior art though?
Not if it is the authors of the paper filing for the patent. Otherwise people would never publish papers.
Patents are intended to be the form of first public disclosure of an idea. Disclosing it before patenting it can prevent the patent application being valid.
US has a 1 year grace period. In most countries, any public disclosure makes an idea unpatentable.
https://outlierpatentattorneys.com/patent-public-disclosure
This always grinds my gears. For some people "discoveries" are so obvious, they don't bother writing a paper let alone patenting it. Then someone goes and patents it...
Beautiful work! I appreciate how this robot clearly does NOT try to look like any natural creature. I don't want a future where we can't easily distinguish nature from robotics. So far humanoid robots look clearly robotic too: hope that trend continues.
Get 4, Doc Oc
Also was thinking of Oogie Boogie Tim Burton
Time to live out my dreams of that guy from spiderman.
That would be Doctor Octopus. Yes I would love A wearable suit with a number of tentacles for locomotion and subduing... I mean interacting.. with people.
I am both super impressed and creeped out
I e been wanting to do this with a basic stuffed animal now for a while.
Just basic interactions with a child plus lessons and a voice would be game changing for the toy world.
If you like point and click adventures check out https://store.steampowered.com/app/1426010/STASIS_BONE_TOTEM... - one of the playable characters is an AI teddy bear and is a great character with fantastic writing.
5 minutes in: This bear is creeping me out.
5 hours in: YOU CAN DO IT BEAR, YOU CAN SAVE EVERYONE, ITS WHAT SHE WOULD HAVE WANTED.
Like using phones as babysitters, just 100x worse.
I don't doubt someone's gonna invent it, but yikes. Imagine telling kiddo their beloved sentient toy is dead because mum and dad can't afford the ever-rising subscription fees anymore.
A teddy bear is too bulky for convenience. How about Tamagotchi but it talks to you. Talkagotchi. Basically that horrible Friend necklace but in a cutely-colored egg shape that clips to your backpack. I want to not be alive.
edit: when my kid asks for one I'll know it's time to move the family to a cabin deep in the woods.
"Who was your best friend in your childhood?" "The AI teddy bear, definitely, I remember every single ad he would tell me, then I would nag my mom to buy me those toys, good times"
"But then my dad lost his job so we had to kill him to save money. Sometimes I still snuggle his corpse."
Careful with that ambiguity...
now we know how those spiders in Minority Report originate
oh no I just saw a future where LLMs are the new wifi and touchscreens in appliances, we're going to let my refrigerator cry aren't we
Optimus robots can do anything without actually indians?