Early Hasbio R&D pt. 1 (nerovance)

It had been about a couple weeks since you were hired to work at the Hasbro subsidiary that was known as “Hasbro Biotech” until they took the name Hasbio a few months back. Apparently after the parent company collapsed under the weight of it’s own failures, there was an “angel investor” that saw potention in their Biotechnical subsidiary, and even had them purchase the rights to “My Little Pony” during the Liquidation Auction.

Hasbio had two major engineering divisions for R&D: Genetic and Technical, and you were hired on as a Programmer for the Technical division. As soon as you entered the lab and were setup with a workstation, you proceed to ask Mr. Sherman: “So… What exactly are we working on at the moment anyways Boss?” To which he respond: “I’m glad you asked. We are currently developing the AI for a lifelike toy pony” before one of your coworkers interjected. “I don’t know why you’re calling what we’re working on anything close to lifelike… It’s more like some kind of vaguely pony-shaped plush robot, a Fluffy.”

When you got down to the Development and Design floor, there was a janky looking quadropedal robot with a series of cables connecting it to a large computer near the middle of the room. Mr. Sherman began to speak again: “This here is what we call the DevBody, containing two cameras on the head to approximate eyes, a pair of microphones for ears, large speaker in it’s throat to act as it’s voice…”

“Oh, and it’s got pneumatics near the joints for simulating real muscles. It’s a near 1:1 approximation of what Genetics is supposed to be working on. Last I heard though they haven’t even gotten a working wetware processor yet.” Interjected the same coworker, Mr. Sherman looking annoyed, but also resigned in the knowledge that your coworker is right. “This young man is what the artificial personality you’ll help develop will be running on as well as tested against. And for your information Chester they actually just got a working Processor in Genetics last week… But I will admit, it isn’t all that promising at the moment.”

Chester then got up and motioned for you to look at the DevBody as he starts turning it on. “Might as well show you what we’ve already managed to get this thing to be able to perform…” spoke Chester as he typed some commands into the prompt on his computer. Soon enough the DevBody started to whirr to life, and upon standing up looked at you and slowly trotted towards you while saying: “Hewwo mistah, am you new daddeh for…” only for it to slump over and, in a monotonous voice, say “Error, token ‘Product_Name’ not found. Program cannot continue operation. Wan Die, Wan Die, Wan Die...

“Chester! Why is it still saying ‘wan die’ while in Error Mode?” Mr. Sherman snapped, before Chester quickly responded: “What? I can’t have a little fun while coding this thing? The Upper Brass haven’t even managed to decide on a name for it yet, and you even explicitly told us that we can’t use a placeholder for some unknown reason. Maybe you shouldn’t have fired the last Programming Lead because he had the thing call itself ‘Fluffy’ or something like that.”

As the next few days passed, you found yourself quickly getting used to how writing the code for the Hasbio project works. As you were working on one of the subroutines related to the learning algorithms: you couldn’t help but hear upper management arguing with Mr. Sherman on a few things, and they apparently had decided that having the AI say “wan die” when it crashes would be for the best when it came to debugging, at least while this is still in early development.

“Ah, you must be the new developer we hired a little while back.” You heard from behind you, which upon turning around you realized was Mrs. Percy, one of the middle managers. “I just wanted to let you know that we’ve finally agreed on what to name the Hasbio My Little Pony project… If you check your email it should contain everything you’ll need to know.” She said before walking off, not bothering to actually say anything about what I’ve been working on the past week… But upon checking my work email: I saw a bunch of chain letters being sent by management, alongside an announcement buried behind some “Send this to 10 people or be cursed for life!” junkmail.

It turned out the guy who was fired before I came onboard was onto something, since it turned out the official name for the product is now “The Fluffy Pony” or “Fluffy” for short… I could just imagine Chester giving a shiteating grin to Mr. Sherman over that, but for now I need to get back to work… Though there was one more email in my inbox, one that finally detailed the kind of architecture the Wetware Processor would likely have.

“Not surprised they’d go with at least a reduced instruction set for the Fluffy CPU, though I am surprised they’d actually go full MISC on it… Still don’t know how much memory it’ll have to work with, but at least I now know that the hardware should be about as fast as a fuckin’ Intel 486… That was given 128 Cores.” Chester said to me as we were on our lunch break. Thankfully most of the routines being used were fairly lean, but I would need to see how the learning and recall functions would work on such a computer, and hope I can figure out how it could be run in parallel.

6 Likes

Can I ask if you’re planning to dive into the technicality of wetware, and the translation of traditional programming into such systems, or is this not that sort of story?

1 Like

That’s Exactly this kind of story. I’m a computer nerd, so of course I’m going to probably end up making some pretty obscure references here and there.

I’ll give a few examples I’d probably look at including later on:

  • Their hindbrain is based on the brains of Koalas, and are what newborn chirpies rely upon while their forebrain processor grows. A Sensitive Baby is the result of the wetware processor system being malformed or otherwise unable to execute any code,
  • The overall structure of their Neural CPU is essentially a cluster of around 96 wetware Transputer nodes, each of which has around 1MB of working RAM.
  • There is an area of the brain which acts as long/medium term memory, which a group of around 8 Computing Nodes are used to manage reading from and writing to… Which actually has a colossal amount of storage, around 256GB, especially compared to what is essentially the computer cluster’s total cache (around 96MB total), hence it also tends to be used regularly as Swap Space,
  • Despite being organic, the 96 Wetware Transputers have an instruction set similar to that of an electronic transistor-based CPU. This is why a robotic DevBody can be used for prototyping the software.
  • The Fluffy Operating System is encoded within a Fluffy’s DNA, and an additional cluster of about 4 Transputer Nodes is contained within the reproductive system for the sole purpose of executing error-correcting routines on their gametes’ DNA in addition to checking for certain signatures within their DNA and if something is amiss: Destroying those somatic cells to likely prevent unauthorized tampering.
3 Likes

Interesting! I’m not very CS literate (I barely learnt QBASIC), but I’m better at biology. :slight_smile:

I think your approach will work at this stage, since you’re basically building biological furbys - the developed artificial brain will essentially just slot into the fluffy biorobotic body and sealed.
That can have its own OS which is the equivalent of the autonomic nervous system, so the body can keep ticking over by itself, it’s just effectively braindead until the brain is slotted in and connected.

I think there will be issues if you try to get this network to be formed during foetal development and subsequent growth since there’s significant neuroelasticity at this early stage. Trying to limit the neuroelasticity and ensure that the core Fluffy OS doesn’t deviate too much, but still leave enough flexibility for learning is a very tricky balancing act and too much bias either way is probably the cause of why fluffies are so fucked up, behaviourally speaking.

Making sure that a foal receive both parts of the OS via autosomal inheritance is also going to be tricky, so there’s going to be some highly conserved regions, which might also be the cause of the traditionally messed up inheritance genetics - fluffy breeders struggle to selectively breed for a particular colour or trait as these trait are encoded within these conserved regions, so the genotype carries everything, the phenotype is purely randomised.

There can also be some self correcting mechanisms here - for example, a zygote that doesn’t have a full and correct OS is non-viable and spontaneously aborts itself, due to RNA read errors occurring during mitosis; essentially a compiling error in computer terms.

I’m probably out of date on my CS knowledge, but I thought the main benefit of biological systems was parallel computing? We can’t process a data stream as quickly as a computer, but we can process more data streams simultaneously.

1 Like

I thought the main benefit of biological systems was parallel computing? We can’t process a data stream as quickly as a computer, but we can process more data streams simultaneously.

That makes sense, I was even kind of thinking about that, but Hasbio probably thought they could just re-implement what normally would be done in silicon using neurons. Then again, I’m also considering having it be that at the end, instead of showing off some kind of bio-furby, they just end up showing off an unholy nightmare with wires coming out of everywhere, whose jaw sloughs off after asking for spaghetti…

Which would then lead to Genetics having to rush out what is basically just a genetically engineered Pighamster with a human larynx onto shelves for Christmas.

1 Like