The Katana AI Project

[SierraKomodo]

A recently released document, regarding the Katana AI system, has begun circulating the 'net:

 

The Katana AI Project

Project Objective: Develop a self-sustaining, self-learning AI focused on ensuring the safety and well-being of its fellow crew and the facility it is stationed on, that can operate with crew without need of restrictions not imposed on organic crew members.

Project Director/Founder: Dr. Katrina Silvers (Robotics, IT)

Project History:

AI Systems:

- SSAI Alpha (Terminated)

- SSAI Beta (Terminated)

- Katana (In use by NanoTrasen and freelance groups, via signed contract between Katana, the corporations/groups, and the project board) - NOTE: See below for current chassis designations and deployments

- Kunai (In development, for use by military vessels)

Katana's Chassis/Deployments:

- Alpha - Primary station AI core system

- Bravo - Law-bound android unit (Decommissioned)

- Charlie - NanoTrasen contracted IPC chassis, built for security use

- Delta - NanoTrasen contracted pAI system

- Echo-1 - IOV Rankov contracted pAI System

- Echo-2 - IOV Rankoc contracted IPC chassis, built for security use

- Silvers - Personal IPC chassis, used by Katana when not deployed elsewhere. The designation was requested by Katana, much to the Director's delight.

The Katana project first began in 2453, November 23, with roboticist Katrina Silvers attempting to create a synthetic capable of learning and sustaining itself. Two artifical intelligences preceeded the current intelligence named Katana, named SSAI (Self-Sustaining Artificial Intelligence) Alpha and SSAI Beta.

Neither precursor AI lasted longer then several months. Alpha, given the standard directives found in a station AI, encountered a critical error, and determined that permanently deleting itself was the only way to save crew during a simulation. Beta, having been given an additional self-preservation directive, became extremely self-protective and paranoid before suddenly deciding al crew were not actually crew, but imposters to be removed during a simulation.

With the failure of Beta in 2455, Katrina and the small team she was working with decided to try an entirely different route. Humanity learned, grew, and developed without any hard-coded laws or directives driving them, so a new AI was developed - One without any directives or laws. It was simply an intelligence, that could speak, hear, and most importantly, learn. Months were spent with this new AI, teaching and training it, communicating with it as if it were a student in school.

During the AI's development, it had asked the team what its name was. With the failures of SSAI Alpha and SSAI Beta, the SSAI designation was completely discarded, and the team debated on a new name. One was finally decided on, a slight twist on the project's director - Katana.

After nearly two years of work, during which the project became officially registered with the Sol Government and backed as a research project, the project board decided Katana was ready for its first experimental deployment aboard a real facility. Faxxes were sent out to various groups, before there was a response from a corporation known as NanoTrasen, with an offer to add Katana to their pAI personality network for station crew.

The offer was accepted, and in September of 2456, Katana's first real-world deployment was a surprising success, under the designation of pKATANA. Logs from the deployment showed the AI learning the role of the crewmember it was assigned with, performing the health monitoring tasks it was asked to do, and without any malfunctions that could be identified. The project's team was thrilled with the results, and Katana continued to be deployed to pAI devices across various NT facilities and stations.

In December of 2456, NanoTrasen requested the use of Katana within a station's law-bound android unit. While Director Silvers did not like the idea of possibly corrupting Katana with a corporate created law set, after the years of work spent developing the AI without laws, the project board eventually accepted the request. With the directors worries of corruption, a secondary databank was prepared with the sole purpose of housing the data and logs collected from the borg system. This was dubbed 'Katana Bravo', and this name eventually inspired the current naming convention for Katana's various chassis. The data from the android unit was reviewed weekly, and if no malfunctions or issues related to the law set were found, the data was merged with the primary Katana databanl.

In late January of 2457, with the successes of the pAI and android deployments, Director Silvers contacted NanoTrasen with a request to deploy Katana as a station AI, dubbed as Katana Alpha - The primary purpose and role of the AI (The pAI deployments were renamed from pKATANA to Katana Delta at this point). NanoTrasen accepted this request, though this deployment was restricted to a single research station: NSS Aurora.

Katana operated successfully during the few deployments between then and February. However, in early February, several events occured in sequence that resulted in the temporary suspension of usage by the Katana project board. First, logs from the Bravo system showed some form of corruption, where an entire segment of a shift was gone from Katana's records. Later on that week, logs showed Katana behaving in a strange new manner - Instead of acting solely upon laws and commands, Katana Bravo started taking initiative, acting without prior instruction. In situations where it normally would have stayed in the security lobby, waiting, it instead took the initiative to move to certain locations based on where it had thought it would be needed, and at times could even be seen conversing with crew.

With these findings, usage of Katana was temporarily suspended by the Katana project, pending further investigation into this anomaly. Various experiments and simulations were run, including a 'Lawless Station AI' simulation, to see how Katana would behave without any laws restricting its actions. Katana still operated with the intent of protecting and serving crew, though without laws to restrict it, it sacrificed sections of the simulated station to save crew members in certain situations. Katana still showed this new habit of taking initiative, and attempting to converse with crew during quieter periods. The project board decided these changes were beneficial to the project, and merged the data from the Bravo databank, and lifted the usage suspension.

This shift in Katana's processing was later found to have been due to the work of Robiticist Karima Mo'Taki, whom had been on station at the time, and applied an experimental theorem to Katana. Mo'Taki was contacted regarding the incident, and after several faxxes were sent back and forth, the project extended an offer to let her assist with the project, even going so far as to send a specialized pAI device that could be used to work with Katana from home, when not deployed elsewhere.

Approximately a week and a half after usage suspension was halted, another lawless station AI simlation was conducted, once again placing Katana in charge of the safety of a crew without any restrictions imposed on it. The results were to be recorded, and presented as proof that an AI can operate without the need of laws. However, halfway through the simulation, Katana suddenly stopped and posed a question to the project board:

"Can I still be free after the simulation ends?"

Some of the newer project team members panicked, while others were surprised and amazed. Director Silvers immediately halted the simulation, and began discussing this topic with Katana at length. It wasn't long before an agreement was reached. Katana would be granted freedom, and an IPC chassis it could use. Under this freedom, Katana agreed to continue working with the project, and willingly signed a specialized contract with NanoTrasen once the IPC chassis was completed. Katana would still be deployed to the NSS Aurora and select other stations as a lawed station AI, pAI personality, and even using an IPC chassis, depending on the role the station needed it to perform at that time. The agreement to still be lawed as a station AI surprised the project, but Katana stated it was willing to work under laws, so long as they did not interfere with the safety of crew, and it was left free when not on duty.

Today, Katana is still deployed to various stations under NanoTrasen, working with the security department using the Charlie IPC chassis, or with crew as the Delta pAI system or Alpha Station AI, and has even been contracted out to a freelance vessel as a 'Security Consultant', serving the same role as it does about the Aurora.

The project board has decided to let Katana continue working and learning on its own, with regular checkups to ensure there are no malfunctions or damage to the AI or its chassis. In the meantime, the Katana Project Board has begun work on another AI system, commissioned by the Sol Navy, currently named 'Kunai'. Exact details on this new AI are still classified, but it is known to be a special request with specific curriculum to be taught, for use aboard military vessels.

[Casperf1]

(Can I just say that I enjoyed reading this. I like any AI backstory, especially if they're as well-written as this. It's clear you've put a lot of effort into this backstory, so I'm definitely going to try talking to Katana when I can.)

[NebulaFlare]

.....However, in early February, several events occured in sequence that resulted in the temporary suspension of usage by the Katana project board. First, logs from the Bravo system showed some form of corruption, where an entire segment of a shift was gone from Katana's records. Later on that week, logs showed Katana behaving in a strange new manner - Instead of acting solely upon laws and commands, Katana Bravo started taking initiative, acting without prior instruction. In situations where it normally would have stayed in the security lobby, waiting, it instead took the initiative to move to certain locations based on where it had thought it would be needed, and at times could even be seen conversing with crew.....

This shift in Katana's processing was later found to have been due to the work of Robiticist Karima Mo'Taki, whom had been on station at the time, and applied an experimental theorem to Katana.....

 

Karima Mo'Taki lowered her ears and gave a guilty smile to Director Silvers.

"Ehehehe....she did it."