Read on Twitter
Good morning! Here are a few more thoughts on #Fakhrizadeh.
1. Just because an explanation is possibly, technically feasible doesn't mean there aren't a million simpler—and thus more likely—ways to achieve the same effects. Occam's razor, people.
2. This is fairly obvious, but "remote killing" here couldn't possibly mean that all ppl involved were hundreds/thousands km away. For one, they would have needed to get the vehicle into place, along roads, and that would have required ppl in-country.
3. It's hard to see how investigators could have found an AI component simply by studying the burned wreckage of the vehicle. In theory they could look at the onboard computers, if they were even intact, but that would require some very sophisticated forensics....
h/t @IranWonk
h/t @IranWonk
4. ...That being said, a remote-control or satellite link could potentially be more easily discoverable from the wreckage, since these capabilities involve hardware (antennas).
5. Satellite communications can be laggy—like long-distance phone calls in the 90s—which would likely make it very challenging for accurate remote control targeting in such a quick and dynamic setting.
h/t @mgubrud
h/t @mgubrud
6. Direct "line-of-sight" remote-control wouldn't have lag issues, but it would require an operator to be within, well, line-of-sight of the vehicle.
h/t @missy_cummings, @David_Hambling
h/t @missy_cummings, @David_Hambling
7. If there was any "AI" component—a big 'if'—it was more likely aiding human operators in a supporting capacity, rather than fully autonomous targeting. Such targeting support features are fairly routine, but they're more on the 'automated' end of the autonomy scale.
8. It's always tempting to declare a new revolution in remote warfare. But if any of the claims about the attack are true—again, a big 'if'—that doesn't necessarily mean they have a wide-ranging applicability, especially at their current state of tech maturity....
9...more likely, there may be a very narrow set of conditions under which such attacks may now be possible. And as such this is simply one more small step in a decades-long trend toward greater remoteness in warfare, enabled by advances in comms and autonomy.
10. Put another way, just because a remote weapon works in one instance doesn't mean there aren't many other simpler, easier, more reliable alternatives that would be more effective in every other instance. But yes, perhaps the envelope of remote warfare just got a little bigger
11. Finally, I think the broader takeaway here is that as long as the tech remains immature, the main strategic impact of mil AI will derive more from its mythology, not its capability or utility: if it's claimed that an op was autonomous, the narrative balance always shifts...
