The intelligence test consisted of two segments: Calculations and a Maze.
For the calculations, the golems solved problems simultaneously, wrote their answers on boards, and held them overhead as soon as they were done.
For the maze, a small-scale structure had been erected on the parade ground, and the golems competed based on the time it took to navigate it.
"And now, we will begin Part 1 of the intelligence test: the calculation questions."
The eight golems lined up in a single row, each holding a board. These were whiteboards treated with Ground Spider resin (GSP).
"It appears everyone is ready. Here is Question 1: 123 + 456 + 789 =?"
Almost in unison, the boards shot into the air. The answer was 1368. Every single golem answered correctly.
"Question 2: 12 × 34 × 56 =?"
Bibs #3, #4, and #5 raised their boards almost simultaneously. About one second later came Bibs #7, #1, and #2. Another second after that followed Bibs #6 and #8.
The answer was 22848.
"Question 3: 12.3 to the 4th power =?"
Bibs #3, #4, #5, and #7 raised their boards almost at once. About two seconds later, Bibs #1, #2, #6, and #8 followed.
The answer was 22888.6641.
(Hmm. Even with arithmetic, a noticeable gap appears.)
"Question 4: Please calculate the sum of all prime numbers between 1 and 100."
Bibs #3, #4, and #5 held up their boards almost simultaneously. About three seconds later came Bibs #7, #1, and #2. Another two seconds behind them were Bibs #6 and #8.
The answer was 1060.
"Question 5: 96498 ÷ 64 =?"
Bibs #3 and #4 raised their boards almost at the exact same moment. About two seconds later came Bibs #5 and #7. Two seconds after that followed Bibs #1 and #2. Another second later came Bibs #6 and #8.
The answer was 1507.78125.
"This concludes the calculation questions."
(This really highlights the performance of the control cores and the strengths and weaknesses of their control systems.)
Jin pondered what other kinds of tasks might draw out those functional differences even more clearly.
(Puzzles might work well. Or head-to-head card games. Even chess or shogi could be interesting.)
Then again, Jin critiqued his own thought—games like that would take far too long to produce a winner.
(Would a memory test serve any purpose, I wonder...?)
He turned that idea over in his head as well.
(At the very least, it would be pointless for Hourai Island's side.)
When it came to short-term memory, the golems simply were not capable of forgetting. In terms of long-term memory, any unnecessary data was simply purged.
"...Ngh, how humiliating."
A few of the participants ground their teeth, frustrated after having it driven home that their golems' calculating capabilities fell far short of their expectations.
"Freddie truly has been blessed with a wonderful teacher and great companions, hasn't he?"
Meanwhile, others looked on warmly, genuinely happy to see Freddie's growth.
"Now that the maze is ready, the contestants will tackle it one by one, starting with Bib #1. We will broadcast a bird's-eye view captured from the airship above. Please note that this view is completely hidden from the participating golems."
A feed of the maze was projected onto the gigantic magi-screen.
Jin's initial thought was simple:
"Hmm. At this scale, it's just like a maze built inside a classroom for a school festival."
Followed immediately by:
"...I was wondering what this reminded me of, and it hit me. It's like watching mice run a maze in an experiment."
Because the competing golems had no aerial view, whenever they hit a dead end, they had no choice but to backtrack to the last fork and try another route, repeating the process until they successfully navigated the maze.
Jin figured that movement speed, coupled with the ability to map and analyze routes in real time, would be the deciding factors.
"Bib #1, Avalon general-purpose golem: 1 minute 56 seconds."
Since it served as the baseline benchmark, no one in the audience knew whether this was fast or slow, prompting only polite, scattered applause.
"Next up is Bib #2. Ready... Start!"
(Hmm. Because of its heavy build, its movements inside the maze seem just a hair sluggish.)
True to Jin's assessment:
"Bib #2, Avalon combat golem: 2 minutes 4 seconds."
It came in slightly slower than the general-purpose golem.
(...Given how close their times are, their situational judgment algorithms must be identical.)
Jin saw right through the design.
Meanwhile, the trials pressed on.
"Bib #3, 'Prototype Type 8': 1 minute 32 seconds."
Groma's team had cleared the maze in an impressive time.
"Next up is Bib #4. Ready... Start!"
(Oh... Hmm. Their algorithm still has a long way to go. There's plenty of room for improvement.)
The control core—specifically the control system—of the golem built by Freddie and the others was still quite rough around the edges.
"Bib #4: 1 minute 12 seconds."
A low roar of "Ooooh!" swept through the crowd, followed by a resounding wave of applause.
Freddie, Greena, and Rubina, however, wore expressions of clear dissatisfaction.
(Looks like the three of them realized it too.)
From Jin's perspective, given their technical caliber, failing to break the one-minute mark was a shortfall.
(They haven't studied enough when it comes to constructing control systems... or rather, is it just a lack of practical experience?)
In that respect alone, Jin felt this exhibition held immense value for them.
"...Bib #5, 'Lotte 53': 1 minute 20 seconds."
Applause rippled out in appreciation of its solid run.
(On the other hand, that one must be running a thoroughly mature system.)
Jin surmised that such an impressive time could only come from layer upon layer of meticulous refinement.
"...Bib #6, 'Jupitera': 2 minutes 38 seconds."
(Ah, so they didn't put much effort into their software... Is that just their design philosophy?)
Every so often, you'd run into engineers who poured everything into hardware while neglecting software.
Hardware was visible, after all, making any improvements immediately obvious.
Software, on the other hand, offered no visible proof of where it had been upgraded. In the worst cases, a superior might even assume you were "doing nothing"... and while such bosses were exceedingly rare, they certainly existed.
(Whoops.)
Lost in thought, Jin noticed the next trial was already wrapping up.
"Bib #7, 'Aria': 1 minute 41 seconds."
(A very respectable time. Honestly, rather than evaluating only the physical frame, holistic tests like this really ought to become more common.)
Jin firmly believed there was plenty of merit in engineers who prioritized software.
"Bib #8, 'Great Rock': 2 minutes 1 second."
(Even if the results are rough right now, that just means there's plenty of room to grow.)
"This concludes the trials for all participants. Thank you all for your hard work!"
A warm swell of applause enveloped the venue.
As noted earlier, this was a demonstration rather than a tournament, so no official rankings were assigned. Even so, anyone could see that the performance of Bib #4, "Storm," had far outclassed the rest of the field.
"And now, we would like to invite Jin-sensei to deliver the general remarks."
(Wait, nobody warned me about this.)
Caught off guard, Jin decided he would simply share his honest impressions of the trials, and made his way toward the temporarily erected podium.