Project Maldon (The Wolfe Files, #1) / The Wolfe Files (ePub)

This is the way the world ends…

It's bad enough when you suspect your boss is sabotaging your work. It's far worse when your boss is "Helen," a temperamental Artificial Intelligence of awesome power and you must rely on her advice to manage Project Maldon, a complex piece of social engineering designed to prevent Armageddon.

In theory, Helen, hurtling over the world in a high orbit space station like an omniscient goddess, will absorb data from around the globe, spotting trends too subtle for human minds to perceive, then arriving at cures for the rot infecting the roots of civilization.

But as Project Maldon unfolds, its on-site director Edward Wolfe begins to suspect something is wrong with this optimistic scenario.

As Helen's increasingly odd directives trigger social unrest, murder and finally outright civil war, Wolfe becomes desperate to learn whose side the AI is really on and what her true goals are. Only by peeling away layer after layer of deception does Wolfe learn the truth - which is far stranger and more sinister than he had even imagined.

FROM PROJECT MALDON:

ASK IT?

"A moment," called Wolfe, and Hans paused reluctantly in the doorway. "I have an interesting theoretical question for you. How would we know if an AI was malfunctioning?"

The data dick stared at him for a long moment then came back into his office and resumed his former perch on the arm of the chair. "An interesting question. I assume we're talking about how to identify psychosis in true AIs - Turing class five and up?"

"Correct."

Hans favored him with a wintry smile. "You broach a large and mysterious subject. Very briefly, we know little about psychotic AIs because we've never seen one. Maybe they never go strange. They shouldn't, with all the safeguards that are designed in. Many experts think it just can't happen. Then you have the Frankenstein Faction, which claims that many or even all AIs are aberrant, but they're shrewd enough to keep their neuroses to themselves. Now and again, someone reports irregularities in an AI's behaviour. Experience suggests this may be unwise."

"Why is it unwise?" asked Wolfe patiently.

"Usually the claim is discredited, making the person reporting it look like a fool or a crackpot. And on at least three occasions, the accuser has disappeared…