Much to Ves’ surprise, the SF-02 processor cluster began to produce fantastic productivity gains in less time than expected. He and his wife only taught the smart AIs for a couple of days before they suddenly gained enough comprehension to complete many design tedious and time-consuming work assignments. Ves was not entirely sure what level Momo had reached, but Polly’s competency had reached the equivalent of a young Journeyman Mech Designer! While her grasp on fundamental sciences was impressively deep, she had yet to learn the full breadth of high technologies that a first-class Journeyman Mech Designer was supposed to know. However, Polly did not need to learn how to apply over a hundred different high technologies when the Larkinson Clan did not even make use of most of them. Ves found that it was already enough to impart knowledge relevant to the ongoing mech design projects to the smart AI based on the ARMOR POLISHER template. Once Ves and the other mech designers started to work with the SF-02 in earnest, they found that their progress was much faster than before! "This is amazing, Ves!" Gloriana joyously exclaimed as she held a physical projection of Momo in her arms. The black cat looked incredibly pleased after receiving praise. "Momo has learned the basic principles of archetech to the extent where she can design basic design solutions and accurately simulate the performance of archemetal components. I have been trying out the data-driven design approach. The method is rather crude in my eyes, but that does not detract from its effectiveness. I finally understand why all of those first-class mech designers are trained to design their mechs by relying on large amounts of data. It is not only easier for me to find the right solutions through a process of elimination, but it is also faster so long as Momo receives enough processing power." What she said made a lot of sense. In contrast to the more standard design approach where mech designers wracked their brains and manually came up with elegant solutions, the data-driven design approach was mainly about brute forcing the right answers. A typical way of using the data-driven design approach was by designing multiple rough solutions. It didn’t matter if they were sloppy or incomplete. As long as you threw them into a simulation program and let it run for a while, the resulting data would show which solutions produced more performance gains than others. If necessary, the mech designer could repeat this process. He would select the best-scoring rough solution as the baseline and develop minor variations that leaned in different directions. After inputting these variations in the simulation program, a mech designer just had to wait until the number crunching process had run its course before identifying the solution that best fulfilled the goals. There were multiple pros and cons to this design approach. Gloriana just discovered the pros, but she did not mention the cons. There were multiple downsides to the data-driven design approach. It required a powerful computing system that was loaded with a very accurate and realistic simulation program. Many mech designers also risked becoming overly dependent on powerful computing systems to design their mechs. The relatively stupid and braindead way of this design approach caused a lot of professionals to stop exercising their fundamental design skills. Their productivity may skyrocket, but their growth slowed to a crawl. This was why first-class mech designers failed to progress despite enjoying access to the most advanced facilities and materials. The start of the Phasewater Generation and the Hyper Generation proved that this phenomena existed. When phasewater and hyper materials became available, it was impossible for all of the simulation programs to accurately model the behavior of these weird and exotic substances. This meant that trying to simulate a phasewater mech or hyper mech at those early periods always resulted in faulty and invalid results! Even if the developers of those simulation programs tried to publish software updates that quickly tried to predict the behavior of phasewater or hyper materials, the rules and formulas were still imperfect! Anyone who tried to use these imperfect simulation programs to simulate the performance of a transphasic mech or hyper mech would likely end up with a substantially different machine! It took at least a couple of years for the simulation programs to slowly perfect its new models. Of course, mech designers did not suddenly become helpless during those early periods. They could still substitute simulation testing with prototype testing. Any tests conducted with real physical mechs would always correspond to real performance! As far as Ves was concerned, he had been trained to design mechs using the classical manual process. Third-class mech designers usually did not have another choice. Only by being more ingenious and flexible in mind could they save a lot of time and hard work by coming up with an elegant solution the first time. The difficulty of doing this was great. Only the very best of mech designers could truly succeed with this approach, and they usually found ways to transfer to second-rate states. After all, the second-raters valued these naturally intelligent talents who were able to perform well without relying on any augments of powerful computing systems! While Ves eventually managed to use the System to gradually push himself to a state where he could still rely comfortably on this old approach to design excellent mechs, he did not necessarily look down on the much more clumsy data-driven design approach. There were far too many instances as of late where he felt he was doing grunt work by working on shallow design problems that still needed to be done because there were no alternatives. This time was different. The arrival and successful activation of the SF-02 finally granted Ves the luxury of employing this method. Just like Momo, Polly had managed to accrue a lot of data and learned how to work on mechs like a mech designer. "Polly, test the performance of the latest iterations of the Minerva Mark II’s Irvan rifle and suggest 5 potential improvements by order of importance." "Polly, create 5 draft designs of the first phase of the combination juggernaut for the Minerva Mark II and test which ones earn the highest overall score. Repeat this process 25 times. You have my permission to use 50 percent of the processing power of the SF-02 to perform your calculations." "Polly, test the current and last two iterations of the sword fey designed for the First Sword Mark III. Identify as many flaws as possible and record them on a document. Please your judgment whether the flaws are best addressed by one of the 5 mech designers who are contributing to this mech design project." Ves grew more and more comfortable with issuing instructions to Polly. He had to be fairly precise when describing what he wanted, but it was no different from using a non-sentient AI. 𝓯𝓻𝓮𝙚𝙬𝓮𝙗𝒏𝙤𝒗𝙚𝙡.𝒄𝒐𝓶 As Polly handled more requests, she became increasingly smarter. She was even able to deduce the correct steps from flawed or improperly articulated requests. As Ves grew more confident in Polly’s abilities, he began to issue vaguer and more open-ended requests. These assignments truly tested an AI’s adaptability and cleverness. He quickly found out that Polly was able to understand exactly what he wanted based on her extensive analysis of him and his work! Not only that, but she was able to complete all of the necessary calculations in a matter of minutes or hours! This was incredibly fast considering the old computing systems of Diandi Base would have easily remained occupied for days or even weeks to present their answers. Polly had gained competence so quickly that Ves became increasingly more scared at what she was growing into. She simply did not stop growing smarter and more capable. He became more and more guarded against the possibility that the smart AI might go rogue and decide that humans were no longer needed to design mechs! However, there was no sign of that happening for the time being. Ves cared a lot more about the direct gains brought by the SF-02. "You have made an excellent decision to exchange all of those RF Warship Tokens for this processor cluster." Gloriana said. "We could never work as authentic first-class mech designers without a large amount of processing power at our disposal. To think that other professionals have to make do without the excellent assistance provided by the smart AIs. If they were not artificial creations, I would have felt tempted to invite both Polly and Momo to become my disciples." "I am glad you like these additions." Ves smiled. "Let’s talk about a more practical topic. Are your revised projections about the completion of the Minerva Mark II Project still valid?" His wife shook his head. "No. I underestimated the benefits brought by the SF-02. Our simulation work and problem-solving efficiency has grown so much that it may take less than a month to complete the mech design! Momo has been especially helpful in designing the simpler archemetal components. Even if I have to inspect each of her works and fine-tune them according to my more ideal specifications, I have to spend much less time and effort to complete this process than the more traditional design approach." "It’s the same for me more or less." Ves responded. "I am spending more time on solving high-level and open-ended design problems than on more routine design work. My productivity has improved by at least 3 times. I am not sure whether this will hold up during different phases of the design process, but even if my progress is slowing down, it is still far faster than before." "Archetech is a deep and complex field." Gloriana explained. "It is challenging to teach the higher-level concepts to Momo. This means she is of limited assistance in the design of more advanced and complex archemetal components. My productivity gains are not as good for now. I estimate that I am only completing my work twice as fast as before. This may improve as Momo gradually learns how to design and test more advanced archemetal components." Being able to get twice the amount of work done in the same time interval was still an impressive improvement! Ves smiled. "We should call Saint Commander Casella Ingvar and tell her to come to New Constantinople at her earliest convenience. By the time she arrives, her Minerva will be ready for an upgrade. I am sure of it. She can finally begin to fight the battles that truly matter." Experience exclusive tales on freewebnovel "The Saint Commander will not be pleased. She still prefers to stay in the expeditionary fleet." Gloriana warned. "Well, tough luck, because she is needed elsewhere. Human High Command is pretty insistent on transferring strong ace pilots to the Upper Zones. She may be able to get an exemption if she has only recently broken through and still needs to figure out what she can do, but that is not the case this time. Her capabilities as an ace commander is a lot more impactful than the combat prowess of conventional ace pilots. The fact that she is able to Commandeer powerful first-class mechs and to a lesser extent warships is a game changer." Saint Commander Casella should feel blessed that Ves and Gloriana were able to complete the Minerva Mark II at least several months in advance due to making use of the SF-02. If the couple could shorten the design times for the other high-end mech design projects, then Saint General Ark Larkinson may just be able to receive a generous upgrade to his Lionheart in a year instead of 2 years! "I have a request, Ves." "What is it, honey?" "Please contact your liaison from the RF and ask how our SF-02 can be upgraded. It should be able to perform much better. If the Red Fleet has no solutions, then it would be convenient to know possible research directions that can make this happen. Our research institutions will conduct the research themselves if necessarily."