Well, honestly, this might be one of the use cases where these “AI” tools could be useful. Pattern recognition is something they are actually great at accomplishing. That said, I am still skeptical and concerned that the algorithms are being designed and deployed mindfully.
we can’t even get self driving cars right yet, I’m not sure ATC is going to be any better. 99% reliability will probably be easy, it’s that last 1% that could take decades to fix.
It’s one of those things that can absolutely be done with care, effort, and money. But we lack a government that will put any of those toward the project. And the private sector will actively undermine it to scrape a few more pennies off the top.
We’ve had self flying and landing planes for years though
Because none of it relies on any kind of AI. It doesn’t have to make decisions, it just flies programmatically. If you commanded it to fly into the ground it would (has, Germanwings crash). Autopilot wont even respond to a collision avoidance resolution advisory.
LLMs are software that turns a computer (an incredibly powerful tool of reliability and precision) into an imitation of a room temperature IQ person who thinks they’re an infallible genius because they can search google.
Well, honestly, this might be one of the use cases where these “AI” tools could be useful. Pattern recognition is something they are actually great at accomplishing. That said, I am still skeptical and concerned that the algorithms are being designed and deployed mindfully.
this will not work with ATC, or even cars in general too many things can go wrong.
we can’t even get self driving cars right yet, I’m not sure ATC is going to be any better. 99% reliability will probably be easy, it’s that last 1% that could take decades to fix.
It’s one of those things that can absolutely be done with care, effort, and money. But we lack a government that will put any of those toward the project. And the private sector will actively undermine it to scrape a few more pennies off the top.
We’ve had self flying and landing planes for years though. You can’t really compare controlled airspace to the chaos that is operating on the ground.
I’m still skeptical they won’t fuck it up, but this should be a manageable problem for ai.
Yet pilots still fly and land manually. And if those systems work so well why haven’t we had self-controlling air traffic for years too?
Because none of it relies on any kind of AI. It doesn’t have to make decisions, it just flies programmatically. If you commanded it to fly into the ground it would (has, Germanwings crash). Autopilot wont even respond to a collision avoidance resolution advisory.
That’s different than AI tho
That’s an example of a computer being utilized in the way ut should be (with predictable outcomes
People don’t realize that LLMs are not computers.
LLMs are software that turns a computer (an incredibly powerful tool of reliability and precision) into an imitation of a room temperature IQ person who thinks they’re an infallible genius because they can search google.
Determinism is one of the greatest properties of a computer and should be the gold standard of any program. Guess what LLMs fundamentally can’t do?
Make a profit?