Free Rath's Gambit (The Janus Group Book 2) by Piers Platt

potential.”
    “Don’t undersell it,” Dasi chided him.
    Lizelle glanced over at her, then back at the programmer. “AI? Fair enough. You’re not going to tell me you broke the Groenveld Barrier, are you?”
    Khyron bit his lip. “I think I did, sir.”
    Lizelle put his fork down, eyebrows raised. “Really?”
    “Really,” Khyron nodded. “I think.”
    “You better start at the beginning,” Lizelle said.
    Dasi smiled. He’s interested now. I know that look.
    “It started as my senior thesis project, a few years back,” Khyron related. “As you know, artificial intelligence has made notable advances over recent years, particularly in large dataset applications, but programs always seem to reach a threshold where they are no longer evolving or thinking on their own without significant human intervention.”
    “Sure. The Groenveld Barrier,” Lizelle agreed.
    “Yes, sir. They’ll learn everything they can given the data they have, and solve the problems set before them, but they don’t seek out new problems to solve. They don’t take initiative, as it were.”
    “And yours does?” Lizelle prompted.
    “Not at first, no,” Khyron admitted. “But I’ve been tinkering with the code for the last few years, and several weeks ago it had some surprising results. I’ve been using open source data and stock-trading as a test case – but not actually trading with it,” Khyron reassured him.
    “That would be illegal,” the senator agreed.
    “Right – just feeding the data in, and letting FiveSight make hypothetical trades on a virtual account.”
    “Five-what?” Lizelle asked.
    “FiveSight, the name of the program,” Dasi clarified. “Because it’s better than foresight ….”
    “Cute,” Lizelle chuckled.
    “It was late at night and my thesis was due,” Khyron shrugged. “Anyway, I started with standard datasets: weather, census, publicly available stuff. Then I connected it to market trading data, and started showing the AI how to correlate events and predict outcomes. When storms damage coastal communities, the regional insurance company’s costs rise, and their stock goes down, for instance. As it started to learn, FiveSight began to create its own investing rules. One of the first things it figured out is that solar activity has a positive effect on the market on certain planets – human mood is affected by those magnetic waves washing over them.”
    “Fascinating. So I should schedule my speeches for high solar activity days, when everyone’s in a good mood?”
    “Yes, sir! People would be more receptive. Anyway, I made some tweaks to FiveSight’s code, funded the virtual trading account and gave FiveSight discretionary authority. I started with ten thousand dollars. And FiveSight lost over nine thousand dollars in about twenty minutes of trading.”
    Lizelle grimaced. “Ouch. That doesn’t inspire me with confidence.”
    “No, sir. But by the close of trading, FiveSight had the volatility figured out, and the balance was north of a million.”
    “You turned ten thousand dollars into a million dollars in one day?” Lizelle asked.
    “Well, after the losses, we actually started with just about eight hundred dollars, and FiveSight did all of the work, but … yes.”
    Lizelle whistled. “Okay, I’m impressed.”
    Khyron nodded. “But honestly, that’s nothing extraordinary, Senator. Other programs can perform similarly well in the right market conditions. And all of them would have fallen short of the Groenveld Barrier – they’re all trading with high efficiency, theoretically making money … but none of them are taking initiative. But FiveSight did.” He shifted in his chair, pushing his shirt-sleeves up to his elbows. “I left FiveSight running overnight, while the markets were closed. When I checked on it in the morning, it had composed a message for me. Usually the messages are things like ‘I discovered a new correlation between this variable and that variable, and