The idea that someone would go on a multi hour car trip based on an AI-generated route without manually checking where they're actually going is quite funny to me. Some older HN members will remember when the first stories a la 'the GPS told me to drive down that ravine' started coming out. I'm looking forward to "my AI assistant routed me to Kangiqiniq, Canada and border security detained me when I couldn't produce a passport".
replyI sometimes think about the time I was on a road trip with a friend, and for what was generally a 3 hour trip to the mountain, was 5 hours for my friend driving.
replyMidway through the trip I was suspicious of the duration, traffic was fine. He was adamant he knew where he was going.
I pulled up google maps, and sure enough. 3 hours.
Turns out, his mapping app wasn’t aware of an offramp, so instead it wanted us to drive an extra hour, then do a u-turn and drive back to take an offramp.
He was blindly using Apple Maps
In defence of Apple Maps, I find it a lot better in terms of routing and traffic than Google now (especially in the UK).
replyI had a similar experience with Apple Maps. I swore it off forever that day.
replyI often won't look at it's route. If it seems like something's wrong along the way of course I'll double check.
replyThose folk driving off bridges and into ravines is scary as hell, it implies there's people driving around that don't look further than their car bonnet whilst driving (-+5 metres).
About 2 years ago, my then girlfriend, who is a medical doctor (read: likely not an idiot), had ChatGPT plan a trip for us in a foreign country (naturally I had to nip it in the bud). And this was 2 years ago, when accuracy was way worse.
replyYou would be surprised what normies are willing to delegate to AI.
I've used AI to plan entire 5 country itineraries multiple times and it's done better than I or any agent has
replyI haven't seen the livestream, but I just heard that they intend to have their AI automatically change your passwords on websites if it considers them insecure, which sounds to me like the worst idea for AI so far, and that bar is high.
replyThe better use case would be to make AI cancel that damn subscription that lets you jump through 20 dark pattern questions and then tells you to call customer support.
replyAnd you end up with a new subscription the LLM was tricked into accepting
replyGreat news: I was able to score you 10% off the next 12 months on your subscription!
replyI would definitely like an AI that helps me avoid dark patterns and enshitification. Is there a browser that automatically solves annoying captchas for you?
replyYou have to manually start the change, it doesn't do it continuously.
replyFor the vast majority of services, even if this action fails and the wrong password is saved (!?) you're still just a "forgot password" click away.
If it’s so easy, why use AI to do it?
replyBecause it’s easy, but not enjoyable.
replyWe used to just write programs to do those things.
replyIt's essentially impossible to write a traditional program that can go through the full process of logging in and changing a password autonomously, without writing fragile site-specific procedures.
replyBy contrast, an LLM can do it easily.
And if something breaks, and something will break - it's Software+Apple, their support will talk to you for 3 hours very professionally, giving you the scenic route of everything IT support has done in last 300 years and then they will schedule another call, apparently with an expert, on which you will be told to reboot your devices (yeah, all of them), and next stop will be asking you to reinstall your devices clean, of course they will remind you to backup data and how iCloud plans can help. After all that you will be asked to go to a support centre and drop your laptop there (that is, if your device is still under warranty).
replydeleted
reply> if it considers them insecure
replyBut that's all of them though?
Why do you assume Gemini is worse than those two? Especially not for code generation.
replyIMO if it is not absolutely net positive, it is very close to being that
reply