The Obelisk

Part Two

Tim felt a punch on his shoulder, which shook him out of his reverie. "Dude," he heard Jansen's voice. "You look tired, man. How long has it been since you last slept?"

"Mmpthbth." Tim mumbled drearily.

"That's it." Ignoring Tim's weak protests, Jansen grabbed the iPad in his hands, slammed it shut, and tossed it onto the sofa next to him. "You're getting some sleep. Come on."

___

It was 2 months since the start of the Obelisk Project. Yet, in this two short months, both Tim and Jansen had witnessed some truly amazing wonders of science. What made it even more amazing was that those wonders were created by they themselves.

First, there were the basics – the fundamental programming of the ANI's processing unit and the enhancement of the visual and audio detection systems – the Obelisk can now detect movement and sound from almost half a mile away and make the appropriate internal preparations. Deciding that a feminine presence will be less likely to evoke suspicion than a male one, Tim made sure that the Obelisk was crafted with exquisite feminine features.

Then, came the intricate algorithms that made up the Obelisk. The main priority at hand – as said by Harold – was to reduce the signs that indicated the Obelisk's robotic features to the minimum. In other words, what they had to do was to make the Obelisk as humanoid as possible.

The most obvious solution was of course, to import Tim's Diversity Algorithm into the Obelisk. Equipped with the algorithm, the Obelisk will refrain from uttering the exact same words in the exact same tone, making it seem more convincing as a human.

The main objective of the Obelisk Project was simple: to create an ANI that is capable of welcoming speeches, simple pleasantries and basic human interaction. And of course, the ability to self-improve.

The self-improvement code Jansen came up with wasn't that complicated as one might've imagined. Prior to the installation of the code, the Obelisk has been uploaded with hundreds of short pre-recorded workaday conversations provided by the employees of Ricardo Tech – not just the branch Tim and Jansen were working at, but from various other branches across the world as well. Hence, whenever the Obelisk utters a phrase during its trial run, it'll record the phrase and compare it with the pre-recorded phrases. If the uttered phrase resembles a certain threshold of uploaded recordings, they are labelled as SATISFACTORY. Otherwise, they're rated as UNSATISFACTORY and is eliminated from the Obelisk's system.

___

Beep. Beep.

"Unhh." Tim groaned and stretched. His head felt like it had been through Vietnam without a helmet. He groped his way out of the dark Shuteye Room.

"Told you you'd feel better after some well-deserved rest." Jansen handed Tim a steaming cup of coffee when he saw him make the awkward zombie walk out of the room.

"How long was I out?" Tim slurped the coffee gratefully.

"About four hours?"

Tim choked on the coffee. "Four hours? What's the status–"

Jansen placed a placating hand on Tim's shoulder. "Relax, she's in good hands. And yes, she's doing extraordinarily well. In fact, while you were catching some forty winks, she developed a whole new algorithm that allows her to detect human response after she had uttered a phrase, rate it based on the pre-recorded responses that we'd uploaded into her system and alter her response accordingly so that it yields the best, most positive responses from the target."

Tim's jaw smashed into the marble floor and shattered into a million pieces. "H...how is that even possible?"

"I know, right?" Jansen blubbered excitedly. "It's like she's getting better at getting better. As incomprehensible as it might seem, her progress is completely mind-blowing."

Tim was quiet for a few seconds. "Do you think that Professor Klusmov was right?"

Jansen, sensing an abrupt change in atmosphere, frowned. "Professor who?"

"Klusmov." Tim repeated. "He warned us that this would happen, remember?"