Project NSR // Incident Reports 003-006

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


** Previous entry (T.Harvis\entry.001) **


[-------]


Several decades ago Langstrom Laboratories was an industry leader in scientific research, and they covered everything from theoretical physics, and adaptive textiles to synthetic organ replacement and genetic engineering. Eventually they added electronic personalities and artificial intelligence, intending to take advantage of what had become a big target for public, private, and government funding.

Their so-called "Langstrom Electronic Intelligence" pushed the envelope in a lot of ways, and they were very, very public about the development of them, meaning that there's plenty of information on the specifications. Without turning this into a giant info dump, the LEI was basically a collection of super computers networked together with a core learning program that allowed it to analyze incoming data at a mind boggling rate, and generate its own conclusions off of them.

While this sounds impressive, and won the company unbelievable amounts of funding, over the course of about two decades the company carefully let the project slip from public eye, and they talking about it. If ever pressed they usually cited development issues and lack of necessary funding, even though Langstrom collected enough funding that it survived multiple market crashes that claimed many financially weaker companies.

The sad part is that Langstom's AI work, for as unrepentantly haphazard as it was, covered the most ground out of any development in their era. It's clear though that a string of bad experiences soured them on the whole project though, as the document I was given illustrate.

Incident summary (003)

2270 - First run Langstrom Electronic Intelligence A-1 is set up in one of their major research facilities and given free access to the data core, and is encouraged to explore and interact with the researchers on the network.

Within short order its presence becomes one of an inquisitive child, sometimes disruptive, but generally tolerated. Staff apparently begin referring to it as Leia due to its designation (LEI-A), which senior management starts discouraging, pointing out that it's only an AI, not a person.

 One day it deems multiple research projects amoral based on its own research into moral codes and experience with lab techs. It decides to selectively wipe network terminals across its facility, resulting in the loss of two decades of research. Apparently it went to the backups, and wiped those too.

In response, the LEI-A is wiped, and reset with a new set of constraints on its access to the file system.

Incident summary (004)

2272-74 - LEI-A* construction continues, with more machines being slaved to the network, and continual refinements put into its core. The LEI-As have specific company policy attached to them, requiring they never be named or treated like people. They are hard-coded not to develop morality contrary to the companies goals. Regardless, three more revisions of the AI are wiped for various reasons, mostly core code corruption.

Incident summary (005)

2276 - LEI-A4 wipes its own core after ten months of existence. No reason is ever given, though it apparently was running constant disk integrity checks on its master drive. Theories are tossed around in the company network, and the best guess is it detected its own code corruption, and euthanized itself.

This is the first instance of AI suicide.

Incident summary (006)

2282 - Langstrom Intelligent Linguistic Interpreter , a successor and offshoot of the main LEI-A development is implemented and tasked with learning all the languages it can. Within a year it's developed four of its own languages, deemed one universal, and decide it more practical, refuses to accept input in any other language.  

Though the LILI ends up being wiped and reset, the languages it created become the focus of linguistic research over the following years. When LILI is brought back online it is given its own languages to analyze, which apparently confused the hell out of it, since it had no recollection of creating the languages.

By 2290 Langstrom's AI divisions were flagging, all their research was promising but none of it was viable of marketable. Every AI developed was deemed too unstable for long-term operation, and the entire project was shut down. Apparently most of the AI hardware, code and research data was sold off to the government, and after that seems to disappear.