Project NSR // Incident Reports 001-002

Connection accepted from terminal:12@localhost
login T.Harvis
pass **********
... username ok
... password ok


** Previous entry (T.Harvis\entry.000) **


[-------]

I managed to get my hands on the incident reports that I mentioned earlier, most if not all of these have dates from before the start-date of the Project itself, but were apparently collected and indexed by the Project to be used as reference material for future issues of revisions... or maybe as an example of what can go wrong.

I'm verifying details on them as much as possible, but since most of these records are corporate or government domain and are coming to me through a rather dubious channel... I can only accomplish so much.

Incident report 001:

June 25, 2249, Bell-Mason Machines Research and Development center, Halyon 455 

Logged by Jason Carver, Safety Coordinator, Haylon 455 Center 

A Bell-Mason 45.73 Revision 403 Artificial Intelligence was tasked with power management across the entirety of a research facility. Given its prior successes in closed environments its engineers decided to open it up to the facility network. Within a month it had managed to dynamically allocate power across the network, reducing the overhead and power consumption by a measure of twenty percent.

When tasked with improving this by another five to ten percent the AI began stripping non-essential systems of power, including environmental and life support systems in a dozen orbital platforms, as well as non-active communication systems on them. 

By the time anyone had noticed the entire staff compliment on the orbitals had died, unable to wrest control from the Bell-Mason 45.73 on the ground, or call for help before they froze to death or suffocated, whichever came first. 

Engineers had to shut down the AI and restart the power distribution network before they could get power to the airlocks in the orbitals. The toll of this mistake was fifteen employees of Bell-Mason Machines. Settlements went out, and silence was bought.

Naturally, I couldn't dig up any record of this incident with the Bell-Mason company, though I did manage to come across a list of deceased Bell-Mason employees that died due to "accidents resulting from unsafe work practices" during June of 49.

[// Search results for BellMasonMachines(employees) status(deceased) date(june) year(2249) :

Susanna Deleon, Liza Newman, Margaret Wilson, Amalia McNeil, Conrad Compton, Nancy Reed, Max Pittman, Fredrick Jacobson, Herschel Finley, Irwin Avery, Jerome Yates, Judy Sanders, Dale Garner, Shirley Perez, Willis Moran

//]

Nautrally, the Bell-Mason company wasn't going to stop its development of the AI. The followup actions on the incident report state that a listing of vital/essential/exclusive systems were going to be drawn up and coded into the AI as untouchable for future power management tests.

Another incident report shows up two months later out of the same facility...

Incident report 002:

August 30, 2249, Bell-Mason Machines Research and Development center, Halyon 455

Logged by Jason Carver, Safety Coordinator, Haylon 455 Center  

The same Bell-Mason 45.73 (now revision 455) Artificial Intelligence suffered a crippling hardware failure resulting in a full-reboot by automatic systems. Hardware recovery completed as expected with no critical errors, and half a dozen service flags going up. System drives as well as tertiary solid-memory drives went through standard integrity checks to ensure that no data corruption had occurred because of the crash. Several thousand lost clusters were found, and repaired, the drive ran a second check to verify data integrity. The entire reboot process was two hours. Paul Lawson was the logged in system administrator at the time, and checked off on both status messages. 

The AI was back online in short order, and resumed its tasks of power management across the Halyon 455  R&D center without issue. At any given time the AI was only allowed to reduce power allocation to core systems by 20%, and most vital systems could only have a 10% reduction, and could not be shut off entirely. However, the Bell-Mason 45.73 was suffering from a hard logic fault resulting in it not recognizing the reductions already imposed on the systems, and applied them repeatedly every few minutes. 

The net effect was core and vital systems having their power allocation reduced by 100% over the course of  two hours. This resulted in safety systems not being properly powered, which resulted in various system faults across multiple research labs, which resulted in extensive data loss when a catastrophic system failure destroyed the integrity of most of the core network's drives.

Several employees were injured due to afflicted systems failing, the worst being third degree burns due to a magnetic compressor giving out and venting super heated gas in development lab. 

The 45.73 is now offline, and being shunted off to ██████████████ for further review and revision.

As far as I'm aware this is the last time Bell-Mason attempted artificial intelligence, though apparently they still have the original 45.73 sitting in an R&D center somewhere, untouched for decades. However, I'm fairly sure that it had a huge impact on the development of their smart-power machines that came out in later years.