You would think this is what people pay Amazon to do. What is it they do pay them for, really?
In other news: Divers want shark cages; sharks do not
.. og når de så endelig accepterer at det er menneskeskabt, så gør det ingen forskel hvad Danmark gør fordi vi er et lille land.
Den slags typer er tabt. Vi må bare glæde os til de dør en naturlig død og prøve at løse problemerne uden deres hjælp i mellemtiden
There’s nothing to “learn”. Using one of these is in no way different than using the other.
I’ve got one hand gesture for them!
Cruise control lets me relax my legs. It’s purely mechanical.
Other automations like lane assist and emergency break system saves my ass if I’m not paying attention. FSD (if used as intended) does exactly the opposite: I have to save it’s ass if/when it overlooks something. I have to be constantly alert without almost ever having a reason to. The human mind is not made for that. I would 100% find it harder to pay attention in that scenario than if I’m also doing the driving and not because Elon is an asshole.
If I could check my email or read a book and have the car say “take over in 10 seconds time” I would get it, but if I’m constantly aware, hands on the ready and eyes on the road I would literally much much mucho much rather not have it at all.
But of course that’s not how it’s being used. Tesla drivers are willing to run the risk on behalf of themselves and anyone they are sharing the road with.
This is basically automation bias you’re describing and it’s what scares the hell out of me with these “FSD” teslas on the road.
Even if you were able to keep constantly alert during the 99% of the time where this works (which I think is close to humanly impossible) why would you want a system that doesn’t offload you at all? The only value of this system is if people ignore the limitations and allow themselves to zone out - the rest of us are at risk when it goes wrong!
I cannot begin to describe how many times I’ve had to re-type the same info in some input form because I hit esc when I had first finished it!
Formentlig, ja. Men det er formentlig en fin idé at fortsætte med at undervise som hidtil indtil vi finder ud af hvor meget.
På trods af at jeg har en kandidatgrad i molekylærbiologi og en phd i biofysisk kemi har jeg ikke lært at folde proteiner. Der var ingen mennesker der kunne folde proteiner før alphafold.
Alphafold er et godt eksempel på en ekstrem god brug af sprogmodeller som ét af flere elementer i et specialiseret system. Det har meget lidt at gøre med Sam Altman og venners idé om en generel kunstig intelligens.
Jeg kan vitterligt ikke komme i tanke om en eneste ting jeg har lært i løbet af 20 års skolegang som maskiner kan bedre end mennesker i dag. Tekst-til-tekst oversættelse måske…? Hovedregning selvfølgelig, men det er ligesom ikke noget nyt og den diskussion tog de vidst tilbage i 80’erne