I'm fine delegating my programming work to a clanker, even accounting, but come on. The clanker isn't going to have to deal with the consequences of booking $40,000 airplane trips to hotels full of bedbugs.
95% chance they're lying
4.5% chance they're that foolish
0.5% chance they're so rich they don't care.I certainly did not have AI do the bookings for my vacation though. Is that what you meant?
I use AI to plan stuff all the time, but then I book it all myself.
Planning != booking.
But yeah im not using to book flights or hotels.
(reusing bullets lists you carefully handcrafted for yourself would probably be better but that is another topic)
Agents seem like they might be helpful for high-friction tasks, with a lot of data. But summoning Ubers and ordering food aren't really that.
The whole point of agents for this kind of stuff is to not have to think about it so you can focus on something more important than a mobile app failing to compute its core function alongside the other nonsense it's doing. And no it doesn't screw up the address either, this isn't 2021 anymore.
This one is not law. That absolutely exists ... but airports demand money for taxis to join that line of customers and so these got more expensive and are never the cheap option.
The rideshare race to the bottom model of pricing has some bad aspects to it, but choice for the driver before accepting the trip is one positive.
I don't have any kind of agent set up to do that now, but to me that sounds like removed friction even if I use it just a few times a year.
I constantly use the Ask AI thing in Google maps to tell it a complex route involving biking and transit and needing to be at a certain place at a certain time. It hasn't messed up on me yet. I would much rather specify the question in natural language than type anything into a rule based UI. In fact I built a toy app years ago that handled a few of my most common queries with the Maps API and I still never used it because of the friction.
siri is horrific
Maybe. But we need to invent a SOTA model that can reliably count to four first.
It's the behaviors of humans when they fight over scarce money that you are describing as "evil".
Humans are the root of all evil.
There are lots of evil behaviors that have nothing to do with convenience.
I don't see money as convenient, honestly money adds a lot of complexity and inconvenience onto my life. What it adds in a good way, though, is power - power to trade hours of my time for stuff I want to buy or services I want to psy for.
That seems incredibly inconvenient, though; you then have to sort it out.
Perhaps you had experience with some old models and crap setups like OpenClaw? But things don’t work like that any more.
I think it's pretty common to actually not want to delegate and just do simple things yourself. It can be satisfying, aside from any of the practical problems it avoids.
Maybe it's a personality thing, maybe it's a learned behaviour thing, but I'm not sure I've met many people I'd imagine being happy just speaking these sorts of things into a phone and seeing what shows up.
This seems really strange to me? I have preferences on seats too, but they largely boil down to “I want to be at the front to deplane faster” and “I want a bulkhead or emergency seat for leg room” and “I prefer aisle seat”. I can communicate those preferences and a human booking my flight can balance them regardless of specific airframe.
As someone who flies between Australia and the US 3-4 times a year, not at all.
I want leg room and the ability to sleep with my head against the wall of the aircraft, but I want to be away from galleys and toilets and not in a seat where the overhead locker closest to me is reserved for crew equipment.
There's so much nuance to these things
"I prefer to be in the back half of the first class cabin if it's a 757, but the front half if it's not, but never book the first or last row of first class (bulkhead) unless it's an A320-family (or another type that has a curtain between first and coach), and I prefer an aisle seat unless it's the last row of first class and the A/C has asymmetric seating in either first or coach (IYKYK), and I don't mind sitting in the back of the cabin if it's on the upper deck of an A380 or 747 (so please don't 'helpfully' move me to a row with a lower numerical value) but otherwise I don't want to sit in the back of the plane, and never book an exit row or the row ahead of an exit row, and only book the row behind an exit after confirming it has the normal amount of leg room and recline, and never book the first row of coach behind premium economy and preferably not the last row of premium economy unless the other rows only have middle seats; and I'm more willing to pay for first if it's a 757 and less willing if any other narrow-body, and don't book a first class seat if the flight is shorter than 1 hour and the premium is more than $25, and never book a middle seat in premium economy unless the only other available seats are in the last 5 rows of the plane."
It is obviously possible to encode that in a decision tree, but, for some of us, it's easier to just look at the seat map and pick what we want.
This proves, yet again, how accurate the AI labs' talk of "money won't matter" is.
It’s kinda vibes based. “You’re at Costco. You last bought X”.
A noteworthy proof of this is the failure of Amazon Dash buttons [1] that blindly ordered your consumables for you. Amazon gave away these devices essentially for free ($5 off with your first order, or some such offer), and people still didn't seem to have been comfortable to use it.
I really liked the dash buttons for two reasons:
1. It made ordering something you regularly and only buy online anyway easy for an entire household, like for my use case, hookah coals roughly every month.
2. You could place the button next to where you store said consumable and most folks would prefer using it over asking for a link or reminding someone else to re-order.
I do think it was a mistake to launch with an initial limited set of branded buttons for common household goods most people pickup at the store anyway (like detergent) instead of a generic button for my point 1's usecase but thats just plain-old BigTech thinking.
I can imagine an AI agent grappling with all that but the first time a Dash button price gouged me that way I'd throw the thing straight into the trash.
There is still the issue of who is paying for said item and building an open platform that allows another individual to use my funds at the press of a button would require a lot of investment (identity, trust, limits, etc) though I do see the value you are proposing showing a "mini market" to select from.
I never got one, but they certainly made sense in my mind for something like laundry detergent, etc. You notice you are getting low, hit the button, forget about it and 3 days later it shows up and you go "oh, right perfect".
Envisioning the gigolo opening up their nightstand drawer, shocked they're down to their last couple rubbers... slapping the button and getting back to some other slapping.
Yes, you'd need to sort out pricing. I'm unusually price insensitive, but too many sellers play stupid games and win stupid prizes (me shopping around instead of blindly buying from them every time) by changing unit prices too much too often (e.g. my sons preferred shower gel litre price regularly rotates between prices that adds up to a factor of 3 difference, with any given size container at any given time potentially being a massive rip-off). You could fix that by letting users configure price bands and make it pick from a set of products based on price parameters or give an error. Or only include products where the sellers agree to not play stupid games.
But now your UI is massively complicated and you have error scenarios you didn't have before.
Even then, though, the bigger problem is that there just aren't enough product categories where we frequently order the same thing often enough separately from e.g. the weekly grocery run, to be worth it.
In my household I think it basically boils down to shower gels - because I have a teenage son - toilet rolls, dishwasher tablets and washing machine tabs.
That's not nothing, but it's not going to make me order materially more from Amazon to be able to press a button to reorder vs. pull out my phone, press "orders" in the Amazon app, and find the last order I made and reorder the product.
And I say this as someone who did order a couple of Dash buttons when they came out and promptly forgot to even use them (and I don't even remember why).
It's just such a very minor convenience that to work it'd need to have absolutely zero friction, and even then the revenue it'd bring would likely be totally marginal.
> Except that regular people compare 13 different options to pick the cheapest one.
Which takes time... and the AI can do that for you, presumably? Do I trust AI _right now_ to do that _for me_? No. Will people try it out and invariably work out the kinks, for a stable set of use cases? Definitely. Screens are just one UI for ordering stuff. There's definitely a market for doing the same thing with minimal or no screen time.
Beyond time, I will go out of my way to make larger portions for leftovers because I know what's in the food I make, and I try to make healthy food. Even "healthy" options from places that deliver aren't always that healthy. There's an element of knowledge about what I eat that is equal to cost savings in my mind.
Yes but you have to spend time making it beforehand dumbass. (Not saying it's not fiscally irresponsible to get delivery for lunch every day, but you're obviously missing the point of the convenience)
I understand the convenience, and I'm explaining that if you are smart enough to plan ahead once in a while, convenience is overblown.
It is few taps in an app and going out for it when it arrives. Literally. And while you are at work anyway. Cooking, shopping and packing take more time - and at the time when you could have been doing something more pleasurable (like chilling with netflix).
For me cooking is meditative enough that it can be included in "chilling with netflix". :D
If you think about the logistics of running a decent kitchen, the set is quite limited in the first place.
Ah, this is the new marshmallow test, except instead of "can you leave a marshmallow uneaten on the table" it's "can you avoid summoning a taxi to take a single marshmallow to your house".
(Bluesky calls this "burrito taxi discourse"; I do notice that a lot of the people who claim that delivery is absolutely essential to them give an impression of privileged learned helplessness while at the same time claiming to be poor. Perhaps that's unfair, or perhaps not. I'm just a higher earning middle aged professional like the rest of HN, who rarely orders delivery food)
Are you arguing against observed reality? People have not been using AI assistants to buy things. Remember all those smart devices they sold trying to do exactly that? None of them are being used that way.
Really shows why companies salivate over people using these features.
- prevent them from being able to access without going thru the company
— cost them more money
The disconnect was almost comical. All the engineers and other regular employees had "eeew, get it away from me" on their faces, and the sentiment was "how much do I have to pay to not get this?"
The managers, on the other hand...I don't know how to describe it. What they exuded was something along the lines of a religious or probably more accurately cultish experience. They had been on the mountain, they had seen the light, the future had spoken to them, and they were preaching the gospel.
And it didn't seem like they were faking it, either.
This is why you can finance Chipotle orders with Klarna
I want to have 30' time to spend cooking my effing dinner, not save 5' on my DoorDash order. Is that too much to ask these days?
My agent routinely does comparison shopping across various sites to filter out according to my criteria, and even finds coupons - something I never bothered to do.
Certainly feels less gameable than the kind of stuff we already see so much of (like referral bonuses, free sign-up balance, "unlimited" promos, etc)
E.g quick commerce. If I order groceries and a charger and it all shows up in 7 minutes, I will just order it and receive it synchronously, don't need an agent for that anymore than I need an agent to pick it up at my doorstep, that is to say, I don't need one.
Food is async (30+min), so it's useful. Most AI consumer users are young people, and many young people especially living in college, order out every single day, and mostly rotate the same ten options. So it becomes easy to automate. Wherever there is a pattern, you abstract. If it's a one off order for a gathering, sure an agent is pointless. I don't use it when I'm at home either and we order the rare dessert.
Anything which I need to replenish at a certain cadence, and is not available in quick commerce, e.g medicines, I absolutely have my agent order. Once every two weeks the same medicine shows up on my doorstep and I don't have to think about it anymore.
Exactly.
So I have a specialized AI agent that comparison shops for me and tracks prices when I'm placing component orders for my company.
Even the first run isn't too painful, but afterwards you can ask it to write memory files with preferences; and any scripts and tools you write for an earlier search remain usable for later ones of course.
I don't let it place the orders themselves, I'd prefer to do that by hand yet. But recently it finds spec matches and prices better and quicker than I can.
This is kind of handy in today's wild IT market with a customer breathing down your neck asking why it's all so expensive.
(of course the reason it's all expensive is because the datacenters out-bid everyone. But hey, every downside has an upside)
Been in plenty of meetings where lunch and drinks are catered. Either you order enough variety that everyone can find something they'll like, or you take individual orders.
"AI, bring me ${food}" is too nonspecific to ever get you what you want. And as soon as you have it present options, you're now in a flow that would have been faster done manually.
"You can call an Uber with a voice command"
"You can ask alexa to buy you things"
"Your AI will order food for you"
Seem like magical ideas that will remove friction and increase sales. Except that regular people compare 13 different options to pick the cheapest one.