|
^ that sincerely.
|
#
?
Jun 23, 2017 13:24
|
|
|
|
| # ? May 8, 2024 14:43 |
|
|
McDonald's is making a huge success of the automated kiosks: http://www.cnbc.com/2017/06/20/mcdonalds-hits-all-time-high-as-wall-street-cheers-replacement-of-cashiers-with-kiosks.html
|
|
#
?
Jun 23, 2017 14:27
|
|
|
CommieGIR posted:McDonald's is making a huge success of the automated kiosks:
|
|
#
?
Jun 23, 2017 14:37
|
|
|
Cicero posted:I'm kind of surprised they didn't start a widespread rollout of these earlier. When I worked at Amazon, one of the local deli chain branches (Specialty's) used iPads for ordering and it seemed to work great. That was 2013. They had one in Macon, GA that I got to use, I really liked it. Its a sad death knell for low income jobs, but with the recent job losses at Carrier and others due to outsourcing and automation, its the push we need for the public to realize that change has to happen.
|
|
#
?
Jun 23, 2017 14:46
|
|
|
CommieGIR posted:They had one in Macon, GA that I got to use, I really liked it. Its a sad death knell for low income jobs, but with the recent job losses at Carrier and others due to outsourcing and automation, its the push we need for the public to realize that change has to happen. Legalize gunning down the poor you mean
|
|
#
?
Jun 23, 2017 14:47
|
|
|
Nah it won't be on the moon. East side burbs of Seattle.
|
|
#
?
Jun 23, 2017 15:35
|
|
|
Lightning Lord posted:Legalize gunning down the poor you mean
|
|
#
?
Jun 23, 2017 16:40
|
|
|
Having a bunch of auto turrets riddling the bodies of people who's only crime is that they're not rich with flechettes is illegal... for now.
|
|
#
?
Jun 23, 2017 16:44
|
|
|
Lightning Lord posted:Having a bunch of auto turrets riddling the bodies of people who's only crime is that they're not rich with flechettes is illegal... for now. That went well for the French Royalty.
|
|
#
?
Jun 23, 2017 17:42
|
|
|
CommieGIR posted:McDonald's is making a huge success of the automated kiosks: they had these at every mcdonalds i visited in europe (shut up sometimes you just want familiar garbage) and it was legit amazing. they had just as many staff in back but they were 100% focused on just fulfilling orders. it was amazing how efficient and non-stressed they looked compared to the mcdonalds' here in the us. one odd thing, though, was that they all had to wear either black jeans or black denim skirts with the mcdonalds arches embroidered on the butt pockets which was p lol.
|
|
#
?
Jun 23, 2017 20:48
|
|
|
CommieGIR posted:McDonald's is making a huge success of the automated kiosks: I'm too lazy to verify but I could have sworn there were a couple goons who didn't think this was happening. I hope they still read this thread.
|
|
#
?
Jun 24, 2017 01:42
|
|
|
Freakazoid_ posted:I'm too lazy to verify but I could have sworn there were a couple goons who didn't think this was happening. lol what really? It's been in every McDonald I've been into for... what four years?
|
|
#
?
Jun 24, 2017 04:04
|
|
|
Nevvy Z posted:Elysium was real. They are gonna have robot guards and doctors on their moon base while the rest of us madmax it down here. CommieGIR posted:That went well for the French Royalty.
|
|
#
?
Jun 24, 2017 06:26
|
|
|
StarMinstrel posted:lol what really? It's been in every McDonald I've been into for... what four years? This thread attracts people that think automation for any task can only take the form of androids with human levels of intelligence and dexterity.
|
|
#
?
Jun 24, 2017 09:17
|
|
|
AI is exciting and all but there are still vast productivity gains available from automating current business processes. I work in a fairly IT friendly industry and we still have some counterparties generating and sending us daily document drops manually. I dread to think what it's like in healthcare or government work. Bringing these processes into a digital framework can also significantly increase transparency and enable new use cases by other business units. I suspect that even with 0 improvement in technology we would see a huge increase in automation over the next 25years as organisations slowly adopt and adapt to the possibilities of current technology (+ hire a load if highly skilled eastern Europeans./
|
|
#
?
Jun 24, 2017 09:32
|
|
|
Kekekela posted:This thread attracts people that think automation for any task can only take the form of androids with human levels of intelligence and dexterity.
|
|
#
?
Jun 24, 2017 09:38
|
|
|
https://www.youtube.com/watch?v=VE0dPpewEqI This film from the 1950's practically bragged about how efficiency experts could use automation to eliminate workers. Three jobs in a mailroom were eliminated by reorganizing the room layout and installing an automatic letter opener, an automatic stamper, and a spring scale. A machine that typed out orders onto standardized templates replaced the jobs of several workers who previously copied down the info by hand onto multiple forms. And three file keeping departments in two different cities were consolidated into a single department by doing things like buying a wheeled ladder so workers could use taller cabinets while also reducing the number of years files were kept. Even so, the offices were still filled with women collating paper forms, typing paper documents, sending paper documents, and finding paper documents. Jobs which were made redundant by largely paperless word processors, email, spreadsheet software, database software, and accounting software.
|
|
#
?
Jun 25, 2017 06:21
|
|
|
Kekekela posted:This thread attracts people that think automation for any task can only take the form of androids with human levels of intelligence and dexterity. There are plenty of reasons to be skeptical that this will actually end up replacing front-end jobs in the medium term at least, and isn't simply a stunt to appeal to investors. They have them here, people still greatly prefer ordering in person, and each of the few times I've visited a McDonald's in the past five years or so, I've had to watch a cashier get pulled from the front to help out someone at the kiosk anyway. It could just be a cultural thing that they feel they can brute force themselves past by making it the only option, but that's still a pretty substantial assumption. I'm a technically-minded young person and ordering from them is still a little overwhelming-- there's this pressure to look through all of the options presented to you rather than just finding what you 'want' and tapping it, and even if you're quick, ordering a number of different items from different categories takes significantly longer than just listing them to someone who instantly understands you and queues it up. I think that feeling will only get worse if there's actually a line of people behind you thinking you're an rear end in a top hat for taking the time to get a pie with your custom-ordered burger, even though they're likely going to do the same thing. If anything, they work best as a complement to the cashiers, to service extra capacity and be there for the people who don't want to wait in line to place some highly specified order, especially considering the majority of their customers are in their cars, rather than on foot. Ormi fucked around with this message at 06:19 on Jun 26, 2017 |
|
#
?
Jun 25, 2017 07:16
|
|
|
Kekekela posted:This thread attracts people that think automation for any task can only take the form of androids with human levels of intelligence and dexterity. This misconception won't go away until Machine Learning becomes better communicated to the public. Traditional AI approaches are out folks!
|
|
#
?
Jun 25, 2017 07:49
|
|
|
Ormi posted:There are plenty of reasons to be skeptical that this will actually end up replacing front-end jobs in the medium term at least It already works and regardless of your personal anecdote lots of businesses are moving that direction, here in Seattle most retail stores from Home Depot to Safeway you can buy whatever you want and never interact with a human. Also since you clearly don't understand how business works they don't show off technology to impress investors, they save/make money to impress investors. The fact that it's becoming more common means it's an investment that is paying off, it's not just for show and your assumption that people prefer a human is wrong. ElCondemn fucked around with this message at 18:10 on Jun 25, 2017 |
|
#
?
Jun 25, 2017 18:07
|
|
|
The lean, mean machine of the modern multinational corporation, renowned for being free of graft and always planning for the long-term rather than fabricating and cashing in on speculation.
|
|
#
?
Jun 25, 2017 18:13
|
|
|
Ormi posted:The lean, mean machine of the modern multinational corporation, renowned for being free of graft and always planning for the long-term rather than fabricating and cashing in on speculation. It's more that these kiosks are probably just going to end up being a stopgap/fallback option until there's more widespread adoption of ordering through an app on your phone. That solves most of the problems (being unfamiliar with the interface, taking too long, etc.) since you can just sit at your table and browse for as long as you want on a device that you're familiar with. It's also never going to be the only way to get food. There will always be human staffed restaurants because that's definitely a thing people want, but fast food is primarily about price and convenience.
|
|
#
?
Jun 25, 2017 18:40
|
|
|
Doctor Malaver posted:Quoting from a month ago but... Okay, I've only skimmed the paper, but basically, the way they've defined curiosity is kinda clever. Basically, they've defined curiosity as exploring the portions of the world that are less understood. So, say you have the following systems: 1) You've got your policy that finds the most profitable move. 2) You've got a system that predicts what is the state world likely to be if I make a particular move - this should be linked with 1. 3) You've got a system to predict the error of system #2 for each possible move - this is your curiosity. Then, when training the entire thing you use both 1 and 3 as your goals - you want moves that are likely to be good and you want moves that are likely to have unexpected results. So one place it'll work better than a fixed percentage choice between explore and exploit is - if it's in a state which it believes it fully understands (that is it believes it can always predict the outcome of any possible move), curiosity is no longer a goal and it just does the best move that it can find. Edit: Thinking about it a bit more, I think this is only a training time thing, which means I probably should go into how neural networks are trained. Basically, during training time, each time a neural network makes a prediction/produces some output, you need some way to say how good or how bad the output is (this can be as simple as the prediction is correct/incorrect), and then you'll have an update function that causes the neural network to either be more likely or less likely to make the same prediction on the same input. So the problem they're trying to solve is, what happens if you don't have a way to say if the prediction is correct or incorrect - for example, if you've got a game, you might only know the correctness of a series of predictions when you win or lose the game. And then you have the problem of figuring out which prediction in the series is responsible for losing or winning the game, and the problem of games where you may be able to put it into a cycle and never win or lose. So what this does is, it expands the prediction that the neural network makes - it both produces a next possible move, and it tries to predict the state of the world after making the move. And now, you have two predictions whose goodness you can judge: the move you made (good if you win, bad if you lose, neutral otherwise) and your prediction of the state of the world (good if you're wrong - you're in new territory, bad if you're right - you already understand the move you just made). That second part of the goodness judgement is what they're calling curiosity. It's kind of nice because it allows you to do the training with a single neural network. I think your other options are: just use a neural network that tries to pick good moves, and then add some noise so it's less likely to get stuck in a cycle; or use a neural network that just tries to understand the world (call it pure curiosity), and then use transfer learning (somehow) to use what it learned to train the neural network that picks good moves. moebius2778 fucked around with this message at 00:02 on Jun 26, 2017 |
|
#
?
Jun 25, 2017 23:44
|
|
|
Let's try making the fast food automation example more concrete. Instead of thinking of apps and kiosks as immediate employee replacements, think of them as streamlining tools. These specific tools do a few things, but two of the big things would be reducing the total time employees deal with customers and reducing the number of errors. A customer describing their custom order to a cashier can take over a minute. Even a regular order can take half a minute. But a kiosk can let the customer place the order directly without talking to the cashier. The order might be placed on company equipment, but it's not being placed on the cashiers time. This won't eliminate cashiers entirely. But it can relieve some congestion during an unexpected rush. And it can provide a way for a customer to place a custom order directly without fear that the cashier will mess it up. In fact, it places the onus on the customer ordering correctly the first time. These tools aren't just going to be faster and more efficient. They also have a cheaper long-term cost than hiring an extra cashier to handle online orders (something they couldn't do before anyways), special orders, and to work unexpected rushes. These benefits add up really quickly in fast food restaurants that serve thousands of people a day. And over time, more people will use the kiosks and the apps. This will result in downtime. Downtime will result in fewer people working per shift. This isn't about putting on a show for investors. Investors don't like finding out that tech bought last quarter reduced profit margins for the current quarter. These moves come from a desire to have comparable levels of productivity with lower labor costs. It's the same reason efficiency experts were so popular in the 1950's and 60's which is what that short film from a few posts back was about. The film showed how electric letter openers and pre-addressed envelopes helped reduce a mailroom from six people to three. And how three people eventually lost their jobs because a department made two simple changes: they reduced how long documents were kept and they bought a wheeled stepladder so the department could use all of the shelves on their tall filing cabinets. People were transferred or fired as a direct consequence of buying new equipment and changing how things were done, but it wasn't always immediately. That'll happen here too, even if it doesn't happen overnight or at every fast food restaurant.
|
|
#
?
Jun 26, 2017 00:03
|
|
|
Kekekela posted:This thread attracts people that think automation for any task can only take the form of androids with human levels of intelligence and dexterity. Up to and including defense against the guillotine building hordes. On their moon base, where they legally rule as the investors who make breathing possible. We've already got a laser that can target female mosquitoes based on the frequency of their wingbeat to kill them.
|
|
#
?
Jun 26, 2017 00:08
|
|
|
Fast food places have been using kiosks for decades and automats are older than you are
|
|
#
?
Jun 26, 2017 00:08
|
|
|
RandomPauI posted:Let's try making the fast food automation example more concrete. Instead of thinking of apps and kiosks as immediate employee replacements, think of them as streamlining tools. My parochial anecdote might have obscured the fact that I do believe automation here will reduce labor costs in certain areas (McDonald's themselves are presenting this as a way to free up labor to focus on other areas of their restaurants), and that I understand the mechanisms involved. My point is more that in the arena of fast food, conceiving of these kiosks (and not necessarily apps, or pre-delivered food e.g. automats) as a direct and absolute efficiency gain in service delivery over human interaction seems to involve a significant degree of wishful thinking, or being sold a bill of goods by corporate PR. They merely have a different cost profile, in that they often take longer, have upkeep costs, are unable to be flexibly replaced when they malfunction, are quite expensive up-front, but do work around the clock without a wage. So they work best in tandem with humans for delivering the whole package: a meal that takes as little as time as possible between ordering, receiving, eating, and leaving for as many people as can be serviced. They can be and are profitable investments for franchisees to make. But the idea that they're suddenly going to reduce McDonald's total labor costs by whatever ridiculous percentage by downsizing the front-end to a skeleton crew or eliminating it entirely is a fiction which makes certain people who know a little but not enough of the internal economics involved very excited and loose with their praise and money.
|
|
#
?
Jun 26, 2017 03:45
|
|
|
I think we have different concepts for kiosks. I'm picturing something like a touchscreen that apes a companies app, with a way to accept payment and a way to print receipts. Established technologies. The screen could even be a tablet that can be swapped out if it gets broken. This will be an added expense, not every restaurant would want the hassel, but it wouldn't exactly be hand-made bespoke devices.
|
|
#
?
Jun 26, 2017 04:34
|
|
|
Bespoke Automated Terminal would be a good user name.
|
|
#
?
Jun 26, 2017 04:54
|
|
|
RandomPauI posted:I think we have different concepts for kiosks. I'm picturing something like a touchscreen that apes a companies app, with a way to accept payment and a way to print receipts. Established technologies. The screen could even be a tablet that can be swapped out if it gets broken. This will be an added expense, not every restaurant would want the hassel, but it wouldn't exactly be hand-made bespoke devices. This already exists, services like square even send your receipt to your email without having to sign in or anything. Places like Red Robin or Olive Garden around here use this service kiosk called "ziosk" at every table. You no longer have to interact with your waiter to pay and tip and you can even order beverages and food through them, so you could theoretically run a sit down restaurant with only bussers and no actual wait staff. The kiosks at places like McDonalds and other fast food places work too, they wouldn't be used and continue to be expanded if they were negatively impacting business. This isn't a long term strategy, it literally only exists because it's a way to save money. I have a feeling that the people who have negative experiences are not the average user, either that or they're trying to find excuses for why the technology can't work. Or they just haven't been exposed to systems that work well, which in my city are used pretty ubiquitously.
|
|
#
?
Jun 26, 2017 05:31
|
|
|
RandomPauI posted:I think we have different concepts for kiosks. I'm picturing something like a touchscreen that apes a companies app, with a way to accept payment and a way to print receipts. Established technologies. The screen could even be a tablet that can be swapped out if it gets broken. This will be an added expense, not every restaurant would want the hassel, but it wouldn't exactly be hand-made bespoke devices. I've posted about Eatsa a couple times but they basically have a bunch of iPads in special mounts with a card reader where people can order their food. They don't take cash and they don't do paper receipts. It's not too hard to imagine a future where a place like McDonalds does something similar while leaving one cashier as a legacy option/iPad watcher. It'd basically train their customers how to use an app that mirrors the functionality of the kiosks and, if you place enough of them in the store, it wouldn't be a catastrophe if one or two malfunctioned.
|
|
#
?
Jun 26, 2017 05:58
|
|
|
Hmm. Walmart has patents on self-collecting shopping carts. Are there any dead-end jobs that aren't being automated?
|
|
#
?
Jun 26, 2017 06:29
|
|
|
Malcolm XML posted:Fast food places have been using kiosks for decades and automats are older than you are You have correctly deduced that technology usually exists in a half-baked, debatably-useful form until it advances far enough to be indisputably useful.
|
|
#
?
Jun 26, 2017 08:44
|
|
|
New tech emerges Goons: Who cares? It's too expensive and impractical, this'll hardly do anything Tech matures and deployment increases Goons: Who cares? This stuff has been around forever
|
|
#
?
Jun 26, 2017 08:50
|
|
|
We just don't want to face the fact that droves of unemployed poor people are going to be slaughtered by near-future overlords
|
|
#
?
Jun 26, 2017 09:09
|
|
|
Cicero posted:Computerphones existed for a a while as a dumb niche until suddenly they were actually practical. VR was utterly useless until suddenly it got decently competent. Uh dude its a pure roi thing its still cheaper to deal with humans than kiosks and the human Cicero posted:New tech emerges The tech isn't new its old as heck and the cheap cost of labor has kept it from spreading. Look at automated checkouts -- they are barely functional and usually need a human to kick them into shape But some local fast food places have been successfully using tablet kiosks for ordering for years so it's nothing new to the industry
|
|
#
?
Jun 26, 2017 09:48
|
|
|
Like I'm totally on board with auto ordering since queuing behind 20 people to get my 2am nuggz is real bad but its a) not new b) not going to result in humanlesa restaurants You're gonna see modern automats and full service places but nothing in between
|
|
#
?
Jun 26, 2017 09:53
|
|
|
Lightning Lord posted:We just don't want to face the fact that droves of unemployed poor people are going to be slaughtered by near-future overlords
|
|
#
?
Jun 26, 2017 10:10
|
|
|
Malcolm XML posted:Like I'm totally on board with auto ordering since queuing behind 20 people to get my 2am nuggz is real bad but its a) not new b) not going to result in humanlesa restaurants Is not about a humanless restaurant. Is about saving 1 job in every shop. With 10 shops it means 10 less dudes. Maybe 1 less HR dude. Si it may end killing 100+10+1 jobs. Every unemployed person is one more person you pay to do nothing. Enought of these and our system dies.
|
|
#
?
Jun 26, 2017 10:11
|
|
|
|
| # ? May 8, 2024 14:43 |
|
|
Tei posted:Is not about a humanless restaurant. Is about saving 1 job in every shop. With 10 shops it means 10 less dudes. Maybe 1 less HR dude. Hmm yea just like the automobile destroyed civilization when it put the buggy whip makers out of work I look forward to mad max but every thing is done by kiosk ordering That said online shopping is killing retail so who knows https://mobile.nytimes.com/2017/06/25/business/economy/amazon-retail-jobs-pennsylvania.html?_r=0&referer=https://www.google.com/
|
|
#
?
Jun 26, 2017 10:17
|
|