00I0
Long --°--'N Lat --°--'E
“So, run this by me once more,” she prompted.
“Well, I gotta offload this hacked laptop––to the injection point in Eritrea,” he begun. “That’s a central hub where the flight recording data is being ingested––direct from the US military drones. Statistics, coordinates, mission parameters––you’ve got visuals, audio… basically a complete history of every mission. There’s hundreds of collection relay points sitting out there––pumping in a feed from every different kind of weaponised tech that you can imagine.”
“Okay, so the system is pulling together detailed notes on everything. And that’s an unreasonably risky number of eggs to keep in one basket, right?” she had put together.
“Yes, exactly. The machine has been hooked up like that to keep tabs on all the active players on the battlefield. Stitching everything all together; watching for the tell-tale signs of war crimes and so on. It can identify any military behaviours that need to be corrected out there––to minimise injury and causalities. Its central algorithm will crunch all that data all the time––run simulations and predictions, learn what works, how best to self-correct. Endlessly playing catch-up, in truth, but it’s got lightning-fast processing and continually makes itself more efficient bit by bit. Standard AI stuff it seems. But here’s the rub: Countless missions are kept top secret. Reams of facts redacted––left offline indefinitely. That basically pokes a hole in its every theory and risk calculation. It's a bug they call the overgrown known-unknown.” Neither of them smiled.
“The system’s gaps in knowledge are endemic and bloody gaping wide by now. Clearly that makes for very shitty decisions in the AI’s central controller. Its kill counts are skewed far too low––that’s just the tip of the iceberg. So yeah, this is where my guys come in. We have a copy of those missing files. We know how to convertly inject them into the machine. From out there––from Eritrea.”
“So, run this by me once more,” she prompted.
“Well, I gotta offload this hacked laptop––to the injection point in Eritrea,” he begun. “That’s a central hub where the flight recording data is being ingested––direct from the US military drones. Statistics, coordinates, mission parameters––you’ve got visuals, audio… basically a complete history of every mission. There’s hundreds of collection relay points sitting out there––pumping in a feed from every different kind of weaponised tech that you can imagine.”
“Okay, so the system is pulling together detailed notes on everything. And that’s an unreasonably risky number of eggs to keep in one basket, right?” she had put together.
“Yes, exactly. The machine has been hooked up like that to keep tabs on all the active players on the battlefield. Stitching everything all together; watching for the tell-tale signs of war crimes and so on. It can identify any military behaviours that need to be corrected out there––to minimise injury and causalities. Its central algorithm will crunch all that data all the time––run simulations and predictions, learn what works, how best to self-correct. Endlessly playing catch-up, in truth, but it’s got lightning-fast processing and continually makes itself more efficient bit by bit. Standard AI stuff it seems. But here’s the rub: Countless missions are kept top secret. Reams of facts redacted––left offline indefinitely. That basically pokes a hole in its every theory and risk calculation. It's a bug they call the overgrown known-unknown.” Neither of them smiled.
“The system’s gaps in knowledge are endemic and bloody gaping wide by now. Clearly that makes for very shitty decisions in the AI’s central controller. Its kill counts are skewed far too low––that’s just the tip of the iceberg. So yeah, this is where my guys come in. We have a copy of those missing files. We know how to convertly inject them into the machine. From out there––from Eritrea.”
Comments
Post a Comment