This isn’t new. Check out Yasha Levine’s Surveillance Valley. It’s a nice primer. Most of our internet tech was built for the military or funded by the military for military ideas (no matter what MIT or Berkeley theoreticians might try to convince you of).
No way in hell would I do that if I had that kind of knowledge. Look what happened to Snowden for doing something like that.
He would still spend the remainder of his life in federal prison or be executed if he ever steps back on US soil or the soil of someone with an extradition treaty that is looking to get some brownie points.
That wouldn't happen to all of them, but I bet you there are some working on some classified mess that would be found and made an example of in short order to shut the others up.
In 2017, I played a part in the successful #CancelMaven campaign that got Google to end its participation in Project Maven, a contract with the US Department of Defense to equip US military drones with artificial intelligence.
Today a similar movement, organized under the banner of the coalition No Tech for Apartheid, is targeting Project Nimbus, a joint contract between Google and Amazon to provide cloud computing infrastructure and AI capabilities to the Israeli government and military.
If a strategically placed insider released information not otherwise known to the public about the Nimbus project, it could really increase the pressure on management to rethink its decision to get into bed with a military that’s currently overseeing mass killings of women and children.
It certainly wasn’t a spontaneous response to an op-ed, and I don’t presume to advise anyone currently at Google (or Amazon, Microsoft, Palantir, Anduril, or any of the growing list of companies peddling AI to militaries) to follow my example.
Back then, the company responded to our actions by defending the nature of the contract, insisting that its Project Maven work was strictly for reconnaissance and not for weapons targeting—conceding implicitly that helping to target drone strikes would be a bad thing.
Today it maintains that the work it is doing as part of Project Nimbus “is not directed at highly sensitive, classified, or military workloads relevant to weapons or intelligence services.” At the same time, it asserts that there is no room for politics at the workplace and has fired those demanding transparency and accountability.
The original article contains 1,257 words, the summary contains 258 words. Saved 79%. I'm a bot and I'm open source!
If there is one thing the Chinese government does well it is law enforcement. If you make a deep fake that isn't in line with the public narrative you disappear.
I wonder how long it will be before we see people purposely create these so they can influence their descendants. There might be a future where people have "relationships" with a collection of long dead people.
AIs controlling money is the application that scares me the most, honestly, not weapons. It's flexible and attached to every section of life by design; there is no such thing as sandboxing.
Depending on how it's implemented, it might very well be one of the non-creepy uses. As long as it isn't taken to extremes, being able to have a chance to say goodbye to your loved one could be therapeutic. Or to see them once a year during the Ghost Festival to help keep their memory alive.
It does have high creepy potential (all deepfakes do), but I can understand the appeal. Make sure you take some nice videos of the people you love, and let them take some of you. Death doesn't always come with a warning, and you don't want to be wishing you had some recordings.
technologyreview.com
Hot