For 40 advance/early chapters : p atreon.com/AutumnXd
Regarding the optimization requirements he'd raised at the meeting that day, Hargrove and the team under him had been working hard the whole time.
The difficulty of the work wasn't actually very great.
Ryan had already drawn out all the engineering blueprints for them, and next they only needed to follow the gourd to draw the ladle.
As a top-tier domestic team, their own strength couldn't be overlooked, and for work like this, they naturally handled it with ease.
And to achieve faster work efficiency, they expanded their manpower again, recruiting another batch of programmers to come pound out code, achieving a double exploitation of tools by tools.
Just like that, the optimization Ryan had requested was finished in a little over half a month.
Ryan came to the research center tower again to inspect the results of their work.
The AI translation supervisor Owen Castellano, who came out to receive him, walked in front, reporting on the work of this stretch as he went. "We've now completed the model optimization required last time. After the optimization, Argus's progress is astonishing, and the problems of translation selection and reordering have all but vanished!"
His expression was extremely excited, because those optimization theories Ryan had described last time were completely correct. Once implemented, the precision of Argus's translation had climbed yet again, and had now reached over 98 percent.
At precision like this, forget daily life, even in some deliberately arranged difficult scenarios, Argus's translations wouldn't go wrong.
This wasn't only Argus's progress. It was also a major advance for him personally and for the team.
And those theories Ryan had described had opened a new door in AI translation for him besides.
How could he not be happy?
Ryan walked into the AI center and immediately drew the attention of the programmers.
"Boss, still short a thigh ornament? I think I could do it."
"The sky's cleared, the rain's stopped, and you think you're up to it again! Boss, don't listen to him. I'm the ornament best suited for it, warm and considerate and able to crossdress too!"
"Get lost, you. A nearly two-meter big fellow wants to cosplay a muscle Barbie? I think I'm the more fitting choice for the ornament!"
The people inside still wore an exhaustion they couldn't hide on their faces, but like Castellano, they cast admiring gazes toward Ryan, spouting all kinds of nonsense with their mouths.
They were too excited. The results after the model optimization were far too astonishing, exceeding everyone's expectations by a wide margin.
And the theories Ryan had raised last time seemed to have room to play in the other problems facing AI translation too.
They'd already begun researching the other problems on the foundation of the theories Ryan had described earlier, and reckoned new results would appear before long.
By then, perhaps AI translation's precision would advance another step.
Ryan smilingly received all of everyone's enthusiasm, and to the programmer who wanted to crossdress, he offered the suggestion of crossdressing first before anything else.
Any crossdressing declaration with no actual action behind it was just being a hooligan.
Hargrove wasn't at the research center today, so Castellano was in charge on-site.
He brought Ryan in front of a computer. Aside from Argus Image, the most eye-catching thing on the desktop was another piece of software whose icon resembled it.
"This software is the Argus Translate we designed over this stretch." Castellano gave a simple introduction, then opened the software.
The software's startup page was its icon, an amiable peacock with a great many bubbles beside its beak, each bubble holding a character or a letter, so that a first impression was enough to tell this was software related to language.
Castellano continued from the side. "We've currently developed both a computer version and an Android version of this software, with identical functions."
"We integrated the image recognition function, enabling it to recognize and translate the text in a picture, and at the same time our latest voice recognition and voice synthesis technology was applied to it too, so it also supports voice translation."
"Beyond that, it can also translate through text input. All in all, the basic functions of this software of ours are about the same as comparable software on the market now."
"But our single greatest advantage is the precision of the translation."
Castellano had delivered this speech to the computer's microphone, and as the words left his mouth, they were simultaneously translated into another language on the computer screen.
"This translation quality is already sufficient." Ryan looked at the translation displayed on the computer.
For Castellano's speech, Argus had translated it just right, without the slightest excess or omission.
"On top of that, even facing spoken dialogue with a large amount of redundant words, Argus Translate's translation quality won't show an obvious decline," Castellano said.
When anyone converses day to day, it's impossible to be as neat and clear as a written draft. A large number of irregular redundant words with only transitional meaning are produced, common ones being things like "um," "uh," "this," and "that."
And different people, according to different language habits, have different filler words too.
These irregular redundant words are a difficult point for AI translation.
Because ordinary AI can't find an obvious pattern connecting these words to the context, sometimes running into the earlier problem of not knowing how to translate a word like the "book" that could be read or reserved.
But Argus's grasp of patterns and features far exceeded that of ordinary AI. It could automatically determine the transitional words in a conversation, so there was no need to worry about these problems at all.
After Castellano demonstrated all of Argus Translate's functions, Ryan expressed his satisfaction with Argus Translate's current level.
"This software can be released for testing," Ryan said.
To this decision, Castellano was of course supportive.
He said with a smile, "The moment Argus Translate hits the market, those competitors are probably going to lose sleep."
Both at home and abroad, everyone was now paying extraordinary attention to Prism AI.
The earlier Argus Image had already given some people quite a fright, to say nothing of this Argus Translate.
"Do our own work well and get the software testing started as soon as possible. We don't need to pay too much mind to the competitors' thoughts," Ryan said.
Competitors?
What sort of thing was that?
After receiving Ryan's instruction, in just two days' time, Castellano and his team had already made preparations for testing, ready to push Argus Translate onto the stage.
Before this, Prism AI's official account had also released a test teaser for Argus Translate.
For all of Prism's news, media at home and abroad were extremely attentive.
The instant the news was released, they noticed it and turned it into news to push to everyone.
In under a single day, the matter of Prism AI's Argus soon setting foot in the translation field was already causing an uproar online.
Everyone knew Prism would definitely enter the AI translation field, but no one had imagined the day would come this early, this fast.
---