Chapter 21 - Period Of Isolation (Part 2)

Since the workforce of my academy was a bit lacking, hiring some of the Earth-0 refugees as the teachers while simultaneously learning as much present knowledge from them as possible was tough.

Not only did I have to arrange their schedules to be as tight as possible, but I also had to reward them with something worthwile for their efforts.

The reason why it was so tough was because Heimer didn't necessarily have enough resources to properly satisfy their needs seeing as they were from a very technologically high-achieving background, so job promotions and higher salaries were the only two things that I could reward them with.

Nevertheless, they were willing to do these jobs. And I also could have used force on them, but that would just be detrimental in the long-term.

However, even if all of the Earth-0 refugees became teachers for my academy, they would still be unable to teach the 100,000 students in my academy.

Thus, for the past 6 months, I have been working with a few Earth-0 refugees with the background of machine learning engineers to design and build robots to work as teachers.

And as of right now, it was successfully accomplished.

These robots used a pre-trained transformer-based model that was commonly used as the base machine learning model of Earth-0 robots called Acclimated Robust Transformer, also known as ART.

I called my version of the model TART-X, which stood for the 10th generation Teaching-Acclimated Robust Transformer. It had around 200 quintillion parameters as opposed to the original ART's 30 quadrillion parameters.

The reason why it was the 10th generation was because the earlier versions were rough.

To train the model, I had to build a supercomputer capable of running weeks of training with my measly 21st century knowledge using the Synthesizer. Not to mention, I lacked a large enough dataset to train the model.

Thus, most of the engineers had spent their time creating the training dataset using all the knowledge I've learned so far. But of course, I didn't give them sensitive information like weapon-manufacturing knowledge or dangerous information used for harm.

I only gave them knowledge relating to STEM with the exception of dangerous ones and the countless harmless literature I've read. Cherrypicking the training dataset was that important after all.

Nevertheless, the Earth-0 refugees also contributed to this. After all, they were refugees from a world that was full of information.

Their personal belongings included PCs and items alike, and those items on average had petabytes worth of data inside of them.

After excluding the personal and sensitive data, we implemented their data into our dataset.

After we were done with gathering the training dataset, we moved on to building the supercomputer.

Luckily for me, my genius mind and nanobots helped a lot. And since the Synthesizer was also there to help with the manufacturing process, the supercomputer was built fairly quickly.

For the software part, the existing computers brought by the refugees were enough to get things covered. This was also the reason why building the supercomputer was a breeze, since I was able to reverse-engineer these powerful computers' architecture.

The last hurdle was the iterative training done to the model, which roughly took 2 months as a whole.

The reason why the model had to go through 10 generations was because of problems where the AI would go rogue and such.

The 10th generation of TART was the best out of all of its predecessors. Not only was it the safest in terms of human safety, it was also the best in teaching personality.

The only thing left to do was to put it into a robot's body, and that was a piece of cake.

Some of the Earth-0 human refugees were living as androids as they have discarded their biological bodies due to personal reasons, while some others were living with artificial limbs for similar reasons.

Elizabeth was currently studying mechanical engineering due to her obsession with the artificial limbs some of the refugees possessed. So she had volunteered in designing the teacher robots' bodyparts when she heard of me talking about it.

That was around 5 months before, and it came as a surprise to me when she said that it was already completed and ready to be mass-produced.

I never had thought that she was this talented in the field. She probably saw it more as a hobby than a job.

As of right now, using the Synthesizer and nanobots, I have managed to mass-produce 253 teacher robots, which I referred to as TART-X robots to stick true to the AI.

They didn't look human, but they were humanoids.

I asked Elizabeth to design them to be as harmless-looking as possible, while also looking as sleek and minimalistic as possible.

The final product was a robot that looked like it was made out of white cushion.

It looked humanoid, with a screen display on its face that showed 2 simple eyes that could show certain emotions. It had a tall and plain white body that was as soft as a pillow with the firmness of a hard sofa cushion.

And since it was a humanoid robot, it had hands and legs like that of a human.

The limbs weren't proportionate to a human's though, since they looked cartoonishly out of proportions to look as harmless as possible.

The brain of these robots had to contain exabytes worth of data storage. Luckily, the hardware of modern computers brought by the Earth-0 refugees was capable of handling it. So I reverse-engineered the hardware into one that was suitable for the TART-X robots.

These robots were first introduced into the academy as the teachers' helpers. But then I thought to myself that it was far more efficient to just replace all the class teachers with these robots.

The humans and aliens that were assigned to be the current class teachers would be moved to a higher position so that they can focus on overseeing the classes that the robots were teaching.

Of course this didn't mean that every single class would be taught by these robots, since many other classes required actual people to teach, like in the military classes.

Not to mention, homeworks, assignments, and student projects were still going to be handled by actual people, since the robots were just tasked to teach the class. This meant the same for certain learning activities.

As of right now, 100 classes were already being taught by TART-X robots.

In each class, 2 TART-X robots were tasked to teach 500 students. This meant that I had to mass-produce 400 of these robots to cover all 100,000 students' classes, excluding the military classes of course.

Essentially, the robots worked together to deliver the lessons as best as possible. And whenever students would ask questions, the robots will work together to bring the best answers to those questions.

The TART-X robots were versatile in all STEM subjects. So every class was taught by the same TART-X robots for every subject. The only thing that varied was the Earth-0 teachers supervising the subject taught.

And since I had already mass-produced 253 TART-X robots, it wouldn't take that long to reach 400 of them.