" A human could do it. Is this so hard?" Yes. A lot of the problems involved - like recognizing faces - can be v. hard.
However, yes this is generally the sort of area where AGI or sub-AGI will start. A robot that can pack any bag with any selection of supermarket goods would b e AGI/sub-AGI. A slightly more modest goal would be a factory robot that can pack any simple box with any set of simple objects WITHOUT having to be specially programmed. That clearly IS where the serious momentum is. That is the goal towards which the latest Baxter robot of Brooks is a partial step. There you have to move the robot's arms to indicate the necessary movements, and it fills in the rest. If you could just gesture to the robot the relevant task and it could work out the best movement implementation - subject to human approval - you would have a serious sub-AGI, and a major breakthrough. As I have repeatedly said, this - simple robotic manipulation and navigation - is where the action will be. All the current would-be AGI projects that come up on this forum, actual or mooted, from virtual world agents to "reverse Turing tests" today are total fantasy, that don't understand the nature of AGI (or language). On 14 October 2013 18:52, Matt Mahoney <[email protected]> wrote: > On Mon, Oct 14, 2013 at 10:54 AM, Bob Mottram <[email protected]> wrote: > > On 2013-10-14 15:27, tintner michael wrote: > >> > >> > >> > http://online.wsj.com/article/SB10001424052702303492504579115310362925246.html?dsk=y > >> [1] > >> > >> > >> [But why can't they barcode the fruit?] > > > > > > > > Some good observations. Also if the checkout machine fails there's the > > embarrassment factor of needing to call for help. I think the checkout > > machines will eventually be perfected but that before that time arrives > > shopping might be mostly done online. Another disruption factor might be > > changes to the business model. Why not have subscription based shopping > > where you just pay some fixed price each week and that entitles you to > up to > > some maximum weight of fresh produce. That would make the fruit a lot > > easier for machines to deal with. > > How about we take this as a concrete goal for AGI? I mean, ideally a > camera should be able to look at your shopping cart from several > angles and calculate the total without you even unloading the items. > Then it would recognize your face and charge your account. A human > could do it. Is this so hard? > > I think shopping online for groceries won't be practical until we have > robots to deliver the groceries for you. Right now, hiring human > delivery people is too expensive. > > I've seen automated checkout machines fail in a dozen different ways. > Walmart has to have a full time person waiting next to the machines to > fix problems. Identifying bananas is the least of the problems. > > > -- Matt Mahoney, [email protected] > > > ------------------------------------------- > AGI > Archives: https://www.listbox.com/member/archive/303/=now > RSS Feed: https://www.listbox.com/member/archive/rss/303/6952829-59a2eca5 > Modify Your Subscription: > https://www.listbox.com/member/?&; > Powered by Listbox: http://www.listbox.com > ------------------------------------------- AGI Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-f452e424 Modify Your Subscription: https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-58d57657 Powered by Listbox: http://www.listbox.com
