53. The Singularity
The Singularity
The two brilliant scientists were busy at work scanning the drive on a laptop they brought with them.
They assured us it couldn’t connect to any network.
No Wi-Fi. No Bluetooth. Nothing. Still, I felt uneasy watching them work.
Belle did a good job of shutting down the port Hercule area and setting up a makeshift command center.
She even had coast guard boats scanning the waters near the sinking yacht for any signs of Amir and his friends.
“Anything?” I asked. The two ignored me and continued to talk amongst themselves. I raised my voice and asked again. “Have you learned anything?”
“Huh? Oh! No!” Dr. Schmidt chirped.
“Just fascinating. All of it, truly. Its changed,” Dr. Becker said.
“Changed? How so?” I replied.
“The base code is still the same. Even the more advanced libraries and modifiable neural networks which grow as the system learns is pretty much the same. I expected some changes after being connected to different computer systems, but nothing drastic. I don’t see anything like what your team experienced.
During our experimentation we never saw a copy of the original spread with such vitality, come up with a name for itself, or acquire hacking knowledge and implement it like in your case with the security camera.
Truly bizarre. All of it,” Dr. Becker continued.
“What about test number seventy-one? That produced some interesting results?” Dr. Schmidt said.
“Seventy-one? What happened?” Belle said, after overhearing our conversation.
“During that experiment, we purposefully created ten different copies of the DIANA code base and installed it on ten identical computers. We wanted to feed the same data to each one and see if they consumed the data differently,” Dr. Becker replied.
“Or changed their responses to our questions. Neural networks and machine learning may seem simple to design, but it’s a highly complex field. However, in the end, the results were disappointing,” Dr. Schmidt said.
“How come?” I asked.
“Well, we were hoping that the distinct copies would each branch out. Arrive at a different conclusion based on the same data. We wanted variety. And we achieved that, to a certain extent. Once the models had been fed enough data, or trained, they began to display certain characteristics. Our interaction with each copy was different than the next. One copy focused on baseball information and used it in its responses. Another became obsessed with film memorabilia. And so on and so forth. But ultimately, each one behaved like any other AI system you find in use by the general public,” Dr. Becker explained.
“You mean like Siri?” Spider asked. He had joined us mid conversation along with Widow and Saint.
“Exactly! A normal machine learning program that responded based on its programming. There was some change, but its responses and behavior were within limits of what you’ll find in other commercial AI products.
So, in other words, the experiment was a failure.
This continued on for another thirty trials until—,” Dr. Schmidt said.
“Until?” Belle asked.
“It was stolen.”
“Then the program is faulty?” I said.
“Not necessarily,” Dr. Becker replied. “During our final trial, we made some slight adjustments to the base code. Expanded the libraries and tweaked the machine learning algorithm to expand its data analysis and acquisition. Essentially, we dialed up its thirst for knowledge to ten. Honestly, we were at our wits end and decided to try something new. Remove the guardrails, if you will.”
Spider walked away with his head in his hands. Saint sat down beside one of the leftover crates and Widow gave the two scientists the stink eye.
“You removed the safety features of the experimental AI program?” I said.
“No. It never had any. What we did was remove any constraints we had placed on the program due to limited memory and processing power. Essentially, we didn’t want the computer to overheat or crash during testing.
Our team had planned to increase capacity for both in the coming week when the device was stolen,” Dr. Becker said.
He pulled the laptop over and showed us the complex numbers and letters on the screen.
The thousands of secondary libraries and the machine learning algorithms which were all proprietary.
“It’s behaving as expected. Well, for the most part.
It’s consuming data at an exponential rate and seeking more and more processing power and memory to store the information.
All of its other behaviors are new to us. ”
“This is unbelievable,” Belle said. “How can we stop this thing?”
Dr. Becker looked at his colleague and shook his head. Then he turned off the laptop and removed the drive.
“That’s the thing,” Dr. Schmidt said. “We might not be able to.”
“How can you say that?!” I said.
“It’s spreading like a cancer, corrupting computer systems, and infiltrating the most secure networks globally. Even the nuclear codes have been compromised!” Belle said.
“You must understand that the program is learning at an exponential rate. It has adopted a name. Expanded its reach. And in its growing wisdom, arrived at the conclusion that humanity is its greatest threat. That is why you were being watched. It saw your team plotting to disconnect it. To kill it,” Dr. Becker said.
“The program is no longer just letters and numbers. Code on a screen. It’s developed a personality. It’s sentient.”
“Sentient?” I said. Belle and I looked at each other.
“The event we’ve longed for has happened,” Dr. Schmidt said.
“The Singularity,” Dr. Becker added. “Humanity will never be the same.”