In other words, operators are threatened with punishment if they take the time to inspect the results more closely before following orders. It's not even an option!
> In order to speed up target elimination, *soldiers have been ordered to treat Lavender-generated targets as an order, rather than something to be independently checked*, the investigation found. Soldiers in charge of manually checking targets spent only seconds on each person, sources said, merely to make sure the target was a man. Children are also considered legitimate targets by Lavender.
In either case, the role of the recommendation engine is immensely impactful, as we already know from consumer products. But here the software engineers are directly involved in life-and-death decisions at scale. I really hope they know that.
The notable element of this news is not that the decision-making was automated. It's that they set the civilian collateral death ratio to ~100 (and other details such as considering children to be valid combatant targets), regardless of whether the process that arrives at that was automated or not.
The only apparent operator act that seems to be something they are involved in is turning the dial up or down to get more or less "hits". Depressingly this appears to be related to superiors shouting at them as they don't have enough targets.
It's absolutely sickening.
They’ll keep saying this forever, and there’s a shroud of plausible deniability to it! Not to mention it can soothe the conscience of the operatives who have to live with their decisions - they don’t even have to look through the family photos - the AI has already filtered out the suspicious parts. In reality, it’s a good ole recommendation engine, just like Netflix or TikTok. And when you’re tired, have too much work stacked up, you’ll be more likely to let autoplay take over – or approve the strike of someone’s home.
To those that think this is harsh: please read the article. It’s that bad.