For 30 advance/early chapters : p atreon.com/AutumnXd
Ryan left Patricia's office and went to Storm Bay, where the three triplet sets were still running connection training.
None of them had gone home for the new year. Frank Cabrera and the rest of his team had stayed too, keeping the pilots company.
"Happy New Year!"
The holiday atmosphere hadn't fully dissipated. When Ryan walked in, Cabrera's group of researchers and the pilots waiting their turn on the bench all greeted him with smiles.
One of them even produced a small holiday envelope and held it out, leaving Ryan caught between laughing and exasperation.
"This is basically a bribe."
He took it anyway. After exchanging holiday well-wishes with everyone, Ryan started getting Cabrera's team to brief him on the experiment progress over the break.
"Things are going well. Our best performer can now maintain a stable connection with the Storm arm for two hours," Cabrera said.
That settled Ryan's mind. He'd predicted the connection wouldn't be a difficult problem, but he'd still harbored some worry that the three triplet sets might fail to adapt.
He talked with the triplets next and learned that they hadn't only been adapting to the arm connection over the break. They'd kept up their three-person sync training too. After a long period of adaptation, their three-person sync duration had stabilized at over three hours.
Ryan ran the numbers. If they could maintain that duration inside an actual cockpit, they'd already meet the standard for live mech operation. The probability of that was low, though. Inside the cockpit, the baseline three-person sync load plus the additional load of an entire Jaeger would inevitably reduce their connection time.
"Keep it up. You've worked hard these past months." Ryan clapped them on the shoulders and left Storm Bay.
Like a lord surveying his territory, he made his way to the firefighting mech bay next.
This time he didn't see Kyle or Marsh in the bay, only a few technicians. After getting an update on the firefighting mech's progress, Ryan left and headed upstairs to the drift lab to check on recent developments there as well.
Marsh was currently leading a team to revise the firefighting mech's design, working to incorporate all the suggestions Ryan had made into the new specifications.
Reeves and his group were doing what they always did, collecting drift data and analyzing it.
That project had entered a plateau. The goals Ryan had originally set were now met, and the data analysis was more than thorough. Ryan had finished reading every analysis report Reeves submitted, and he'd fully mastered the liquid neural connection technology.
By rights, the project could be wound down, because Ryan had no immediate plans to pursue deeper research in that area.
Reeves and his team disagreed. They believed the project still had unexplored territory.
For instance: beyond the sync fluid media Ryan had specified, were there other viable media? What effects did the sync process have on the human brain and psychology over time?
Reeves's group had established several research directions and were now working toward them.
Ryan could only wish them luck.
The following days were ordinary base routine. He helped out at the firefighting mech bay when he had time, or checked on the experiments in Storm Bay. Otherwise he went outside to exercise, or confirmed the Crimson Typhoon preparation progress with Patricia.
Oh, and there was the system's third project. The progress had climbed to thirteen percent over this period.
The first batch of documentation had been released.
It was a project that made his scalp prickle just looking at it.
Ryan opened the system panel, looked at the dense blocks of code displayed there, and immediately felt himself developing reading comprehension issues and a phobia of densely packed objects.
The reason was simple.
The third project was, of all things:
MOSS.
The artificial intelligence from a science fiction film he'd seen in his previous life.
MOSS controlled global network resources. Operating through a space station, it could perform communication relay, data transfer and storage, real-time language translation, logical computation, and a long list of other functions.
Put simply: any device anywhere in the world, as long as it was connected to a network, MOSS could influence or manipulate it.
Backed by enormous server resources, MOSS had built a vast database of human civilization, paired with extreme computational speed, capable of simultaneously processing every request from both the ground and the space station.
Whether MOSS met the standard of a self-aware strong artificial intelligence, Ryan couldn't say. But its capabilities had been displayed in full in the film. At its most expansive, it could provide simultaneous translation for every person on the planet, eliminating communication barriers. At its most reduced, it could take a space station and set off alone on a path of escape.
However.
That was an awful lot of code.
Ryan looked at the code displayed on the system panel and felt his head swell.
The base architecture of MOSS alone ran to a line count measured in the billions.
Never mind the databases it might contain further down, the databases capable of reconstructing human civilization.
This one genuinely could not be copied.
The system didn't provide a real-world data-transfer service.
Even if Ryan tried to copy the code by hand, typing without stopping for a single moment, the base architecture alone would take him several years to transcribe.
Never mind the other functional modules that came after it. The communication and computation functions, each with code more voluminous than the last. Even typing at the speed of a Gatling gun, he wouldn't finish transcribing this code in his lifetime.
Since acquiring the system, this was the first project he couldn't simply copy.
Still, while the code couldn't be transcribed wholesale, he could study the concepts and algorithms inside it.
Those were the technology's true treasure vault.
If he understood the concepts and the algorithms, he might be able to create an artificial intelligence on the same level as MOSS. The databases and the functional modules could be accumulated gradually once the core AI existed.
So Ryan's research focus over this period became the system's decrypted base architecture, along with a particular artificial neural network algorithm.
A neural network algorithm was a model that simulated the structure of the human brain. It was composed of many layers of different units, each unit comparable to a neuron in the human brain.
These units were functionally and structurally simple, but like the neurons in a brain, they were interconnected. Each unit's computational result on a piece of data fed into the next layer of units. Computing layer by layer in this way, the network could ultimately perform extraordinarily complex calculations and arrive at the result people wanted.
Neural network algorithms had wide applications. Image recognition, speech recognition, and even AI face-swapping all had a neural network's shadow behind them.
It was also the basis for a computer's ability to "learn" on its own.
Take image recognition as an example. If you showed a computer a picture of a cat, the computer could use the algorithm to analyze and remember the cat's features. Then, in other pictures shown to it later, the computer could identify the cat in the image based on those features, and the identification would grow more accurate as the volume of identifications increased.
That was machine learning.