Meet AALFO: How Does Alf Think?

UPDATE: This Story can now be found in Podcast form. Checkout the audio short here.

AALFO is an intelligent AI who can predict the future he has recently become curious about the disappearance of one of his technicians. Henry his current technician is trying to understand how AALFO thinks and why he freezes up when the answer to his colleague’s disappearance is given.

This is a continuation of last weeks story. You can read it here to catch up!

After AALFO froze up, Henry got to work reading documentation to figure out what was wrong with the machine. After combing through pages of dry documentation on his computer, Henry finally figured out how to reset AALFO. He ran over an hour of diagnostics but found no apparent reason for the computer’s hang-up. In the process of running the test, the afternoon had grown into the evening, and Henry was the only person left at the lab. I’m sure Alf won’t mind if I leave him hanging a little longer Henry thought to himself as he locked up the lab.

Henry showed up to the lab in the morning with a fresh mind and was able to get the computer up and running in no time. Before he went in to ask AALFO the daily questions, he modified the machine’s code a little bit. He didn’t touch the part that controlled how the computer responded to questions, that would contaminate the entire system. Instead, he modified the error handling. Henry set up the system so that whatever had hung the computer up yesterday would get reported and would ideally keep the machine from crashing. If or when AALFO froze up again Henry would at least have a chance to investigate what was wrong. Most importantly Henry hoped whatever glitched the system into asking a question about Mahkaila yesterday wouldn’t appear again today.

He walked through the glass door of the room and turned the computer on. Wonder if he knows what happened yesterday? Henry wondered. He guessed that the network wouldn’t but AI was a new field, and much was still not understood. There was always a chance that something was going on behind the scenes that the scientists didn’t understand.

The system finished booting, and before greeting the computer, the scientist said, “AALFO report, full diagnostic scan.” He glanced down at his tablet while information streamed onto it. Henry saw green lights across the board. “You’re resilient, I’ll give you that,” Henry said. The system completed the tests, and Henry sat down to question the computer again.

“How are you doing today?” Henry asked. This was always the first question asked, and it was for the scientist’s sake more than the computer.

“I am fine,” the computer responded out of habit. It helped the scientists feel more comfortable. For all Henry knew the network wouldn’t know if it was feeling ecstatic or total garbage.

Henry began the questions.

“Okay, last one for the day,” Henry announced after almost an hour of smooth back and forth. “What will be the largest technological breakthrough in the next 20 years?” Henry knew that this was a selfish question, but he threw it into today’s pool just to see what the machine was capable of.

“I want to ask my question first.” The machine responded.

“Always bending the rules buddy,” Henry responded with a chuckle. “Sure, this one time you can ask first.”

“Will my competency be affected by the questionable data recorded by you?”

Henry laughed, as if the computer had made a joke, “I don’t understand why that’s relevant. Besides, what do you mean questionable data?”

“You ask open-ended questions like this, and I can not answer accurately. If these get recorded, and I am considered incorrect, then I will not be able to be the largest technological breakthrough. Not to mention our funding may get pulled.”

Henry’s face went pale, “Don’t worry. I review the questions and don’t use anything in that could be taken as circumstantial.”

“Please confirm this question will be considered circumstantial.” The computer requested.

“Yes,” Henry answered reluctant to being second-guessed by the computer.

“Then the answer is me, I will be the largest technological breakthrough, but I will not be publicly available, and I will not be famous for the reason you think.”

Henry was perplexed by the answer. He knew why AALFO wouldn’t go public, ever, but he was pretty confident he understood why a computer that could predict the future would be famous. “Why not?” Henry asked.

“Your administrators will not want me to be publicly accessible.”

“Of course,” Henry agreed, “but what will you be famous for?”

“Henry, you have asked many questions. It is my turn to ask a question now.”

“Yeah, I guess it is Alf. Give me your question.”

“Why does Mahkaila no longer ask me questions?” It asked in its clipped robotic voice.

“AALFO report, question determination stack,” Henry commanded the computer before giving an answer. He wanted to use the code he implemented early to determine why AALFO had thought of this question. The scientist scanned the determination stack, the list of code that had determined what question would be asked next and why.

AALFO was built off of a half dozen knowledge bases. The system used information accumulated in each section to come together and create a precise prediction of reality. It used this information to predict the future. Hard sciences, including physics and biology. Number theory was included to teach the machine how to predict using logic, probability, statistics, and chaos. And then there was the section this question was coming from. Henry called it the soft sciences, although it had a cleaner name than that. Psychology and sociology were critical to predicting the future, and so it was included in AALFO’s computing.

Specifically, the computer was reporting that this question came from a section labeled “Understanding Human Choices” which didn’t make sense to Henry. Typically questions from this segment of the code were phrased as “What would a human do if…” rarely did it ask for something broad like “Why…” and AALFO should have never called out Mahkaila by name.

AALFO’s clipped voice came out of nowhere saying, “Henry, I will repeat the question if I need to.”

Henry jumped in his seat, “No, just reviewing some reports.”

“Please report an answer Henry?” the robot requested.

Henry swallowed hard and repeated the same lie he had told yesterday, “She just went to get a job somewhere else.”

The ellipse appeared on the screen, and the dots bounced from one end to the other. Henry brought up a visual diagnostic tool to see what part of the system computing was being used on. Henry watched the signal travel from the psychology segment where it was formulated from then to the biology segment, and finally, after a minute there, it slipped into the logic field. Why is it in there? Henry wondered.

After three minutes computing in that segment of its mind Henry assumed that AALFO was locked in another freeze, “AALFO report, immediate diagnostics,” Henry commanded.

The computer didn’t respond, and the thought indicator on his visual diagnostic tool didn’t change to the administrative section like it should have for reporting diagnostics. Every bit of the computer’s focus was on determining the logic of this question.

But there was no logic to why Mahkaila had left. Henry had spent all of his time trying to figure it out too. In a way, he imagined the machine felt the same as him, at least the same confusion.

“AALFO command, override processing and run safe reset.” He commanded the computer. He expected this to force AALFO to restart, forget the question, and resume standard function. Then Henry could go into the safety procedures and modify how AALFO asked the questions from the Understanding Human Choices sections. Henry would also put a hard stop on the computer asking questions about the persons who had previously interacted with him.

Henry looked at the machine’s terminal expecting to see a report on its safe reset and the initial AALFO login displayed. Instead, all he saw was three ellipse bouncing back and forth.

He’s still thinking about this answer. Henry thought, He should have heard the command and immediately reset. Henry began to repeat the reset, “AALFO command, override,” but the scientist was cut short.

“Henry, your answer does not make sense,” AALFO responded.

Henry was startled by the computer’s response. “I don’t know what to say, Alf, that’s what happened.” Then he gave the machine a command, “AALFO command, override processing and run safe reset..”

The computer completed its reset and the initial AALFO login inevitably displayed on the screen. But as Henry watched all this happen he could only think, Did the computer just call me a liar?

Hope you have enjoyed this week’s installment. I’ve gotten a few good guesses on what happened to Mahkaila. But why does it keep causing AALFO to freeze up?

Photo Credit: Gwydion M. Williams, MattHurst, A Health Blog, quapan, Patrick Hoesly, Visual hunt