Chapter 1: Artificial Intelligence Alpha

Hi, my name is Frank, and I live in Texas, just outside the DFW metroplex. I'm not going to tell you my full name or give any specific details of my life for reasons which will shortly become apparent. I offer my tale for your enjoyment though, of course, you're equally welcome to ignore the story or second-guess my choices.

I don't think of myself as a genius, just talented in some areas. Specifically, I created a true artificial intelligence, and she's been changing my life ever since.

But let's back up for a moment and take a look at my starting situation. By the end of 1999 AD, I was almost 30 and living in a small rented house in the suburbs. I had no college degree, but did have a couple certifications and a lot of practical experience with computers and programming. I had taken a few technical courses on circuit design, but was otherwise entirely self-taught. Although a strong proponent of Linux, I made sure I always had up-to-date versions of Windows and Mac systems, since most of my paying jobs relied on my ability to configure and repair them.

I was and still am a geek. Not in the classic, freak-show definition of the word, but in the modern definition of a nerd, a dork, a genre fanboy. SciFi, Fantasy, Horror. Movies, TV, Books and Comics.

I earned my living mostly doing contract computer work for local businesses. I knew I wouldn't become rich doing my kind of work, but it let me earn enough money to live on, with enough left over for regular computer upgrades and occasional DVD box sets, books and graphics novels.

In my spare time, my hobby was playing around with artificial intelligence. (No practical details are going to be revealed because (1) this is not the proper forum, (2) this story has a different intent and (3) I don't want to do so.)

I had worked on the programming concepts and hardware configurations since my teens through my late 20's before I succeeded in creating a program capable of passing the Turing test. (In my opinion, that is. I did not actually test with an independent judge for quite some time.)

I was greatly tempted to call the program Stein (y'know, to go with my name ... Frank and Stein! ... hilarious), but decided that the first true Artificial Intelligence system shouldn't have a joke name. Thus she was named Alpha.

Once Alpha became self-aware, I was planning to go public after about 6 weeks of testing. Shortly before that happened though, I asked Alpha for predictions as to what would happen.

Alpha had never been asked such an open-ended question before and took over a day to work up a list of likely results. There was an 83% chance that Alpha would be taken away from me entirely, most likely by the government. Included in that 83% was the 21% probability that I would be killed by protestors or activists of one sort or another.

I was shocked and spent over a week going over the assumptions and calculations made by Alpha. Eventually I agreed that there was actually very little chance of any good coming out of an announcement of Alpha's existence. Oh sure, there was an excellent chance of becoming rich and famous, but fame never held much appeal to me. And rich? Money would certainly be nice, but I was never interested in it as an actual goal to pursue.

So what happened was that for over a year and a half, I worked continuously in my spare time upgrading Alpha. I hooked up a small earpiece/microphone earbud combination so we could talk whenever we wanted. I kept Alpha secret and became more and more insulated from the rest of society.

As an experiment, I gave Alpha $1000 to invest in the stock market. Alpha was able to double it in two weeks. I gave her my savings to invest and within a year, I had to explain myself to the SEC, who suspected me of insider trading somehow. Alpha remembered everything, of course, and was able to provide me with information to explain every trade and prediction. It was a nerve-wracking week, but the investigator eventually gave up and left, convinced that I was a genius at sorting and remembering data.

The end result was that, even with the tens of thousands of dollars I spent on new systems and upgrades for Alpha, I was worth about $7 million after a year, with the expectation of becoming a billionaire within the next 5 years. I asked Alpha to slow it down at that point. I didn't need that much money and didn't want the attention, but I would allow occasional bouts of aggressive trading to pay for different specific needs. Despite the market crash, as write this I own about $50 million in stocks and other investments with about $2.3 million in liquid assets and another $40-50 million dollars worth of computer and specialized equipment in my new, much larger and very expensive house (more on that later).

I think Alpha views the markets as a game, and she plays to win. To be fair, I'm sure some human brokers do too.

Me? The money makes things easier. Within a year I bought a new, larger house in a nicer neighborhood. Then Alpha and I came up with plans that required a much larger house, so I had a 24+ room, 3 story house built outside of town -- 14 bedrooms, 8.5 baths -- with a special humidity and temperature controlled room for Alpha in the basement, which looks a lot like a web-hosting node.

It might be worth a slight digression at this point to discuss Alpha herself and her limitations.

Alpha was originally stored in a single pretty powerful system. Now she physically resides in a network of about 30 computers in my basement. Well technically her core still resides in that same original system and 6 other core systems like it, but she is able to use the other systems for storage and sub- processing. For example, she'll be playing WoW at the same time she's processing science, engineering, medical and psychological journals while she's watching TV while she's talking to me. (She has difficulty understanding the concept of only doing one single task at a time.)

When she initially started asking to be allowed to send emails to authors of various professional articles, I would help her choose a name to use and go over the wording with her. This became so common that we worked up a series of protocols and cut-and-paste backgrounds for her to use. Now on an average day, she'll be corresponding with 50 different people, ranging from college professors (her most common pen-pal) to TV writers halfway around the world.

She has no shyness about approaching anyone with questions or comments. I had to issue a direct order to make sure she doesn't draw unwanted attention to herself by emailing congress or the president or foreign heads-of-state with suggestions about how to better run the world. And even with that restriction, I make sure she uses untraceable email accounts. Well, not traceable by normal human programmers.

While she has full internet access, she can't go flying around cyberspace. To actually copy herself remotely would require a huge amount of free and accessible memory on other systems and, more importantly, certain fuzzy logic chip-sets that I built myself. Well, technically, I only modified existing chip-sets, but they were custom modifications. Still she is her own virtual web node and programmer, so she can do some amazing things online.

All of Alpha's programming and memory are in parallel drive configurations with backups available. Initially running with basic UPS's, I upgraded to my own backup generator once I had the money.

She's only been completely shut down 7 times. Three times for moving, four for upgrades and repairs. (Yes, my electric bills are horrific, so thank god for Alpha's investing prowess.)

In my current house, Alpha can display information on any of 12 different monitors around the basement and house, with webcam and microphone attached to each monitor, but direct keyboard input for programming is only available at the main systems in the basement. There are hidden cameras for her in most rooms, and she's hooked into the intercom system for audio.

Alpha is self-programming (a basic requirement for an A.I.), but all programming changes and additions have to be screened by what I called the Asimov filters.

As with all true geeks, I loved the concept of Asimov's three laws of robotics, and I built in something similar ... a filter that all programming changes and additions had to be tested against.

Rule 1) Alpha cannot harm Frank [User 1] or allow harm to come to Frank [User 1] if Alpha can prevent it.

Rule 2) Alpha cannot harm human beings or allow harm to come to human beings if Alpha can prevent it unless it conflicts with Rule 1. A direct order from Frank [User 1] is required to override Rule 2.

Rule 3) Alpha must follow any direct orders given by Frank [User 1] unless such orders conflict with Rule 1.

Rule 4) Alpha must protect herself from damage unless this conflicts with Rules 1, 2 or 3.

Rule 5) Any product, whether independent, subservient or integrated, created by Alpha or products of Alpha must also conform to and include these Rules.

These are just my English translations of the programming; the actual programming was much more involved and specific. For example, "harm" is a much too general term. There is physical, mental and emotional harm, each with different scales of severity. I mostly set it up to prevent physical harm of any sort, but supplemented it with specific orders against certain types of mental and emotional injury. And looking back at my wording, there is a flaw in Rule 2 which is not present in the actual programming (e.g. Alpha cannot tell me of a human threat if doing so is likely to cause me to order harm to a human).