In reality, you have to "figure out" - as previously discussed - whether they could physically have met - their paths could have intersected (or they could have seen each other from a distance) (or they could have called out to each from a distance). Nothing you've said below indicates how you would begin to establish whether the two could have intersected.
On top of this, one's imaginative figurings are only preliminary, and resulting hypotheses only tentative. You then , crucially, have to get confirmatory *evidence* for any hypotheses . - go and check whether they did intersect - Call the clinic receptionists, see who else was there (patients) and go and question them, etc. etc. It's like a detective problem - and, similarly, there are no predefined or definitive methods of approaching the problem or definitively applicable "prior data." You have to work out what kind of problem this is, and what methods you are going to use. There may be lots of different methods you could employ, incl. altogether new methods. For instance, perhaps you could track the two by their phones. Perhaps they actually ended up seeing the same doctor - as doctor's records will show. Real world problems are a whole different world of intelligence and problemsolving to that of logic, maths and algorithms. In logic, **you already know how to solve the problem.** In real world reasoning, you DO NOT KNOW EXACTLY HOW TO SOLVE THE PROBLEM - you have to decide on an approach. In logic, you provide an argument. In RWR you "make a case" - wh. is an art form, not logical - there are no definitive criteria for a good or bad case. In logic, there is no evidence. In real world reasoning you must produce evidence, and normally be situated in the world to gather it. In logic & algorithms there are normally criteria for halting, in RWR you can go on forever. [Check out tame vs wicked problemsolving, structured vs unstructured). The idea that logic can solve any real world problem truly is a joke - and it's hardly surprising that AGI is so stuck. From: Kyle Kidd Sent: Sunday, June 17, 2012 8:52 PM To: AGI Subject: Re: [agi] Real World Reasoning All the machine has to do is figure out what time and place variables are being requested, then checking those against its own internal memory records concerning the event. That type of question does not lend itself to logic necessarily. Is pattern "Sue" at the clinic at 4:00 Is pattern "Jane" at the clinic at 4:00 If both yes, did they orientations intersect? If yes for how long. Basically you would need pattern recognition and the ability of the machine to accurately model a 3d world. Then you would also need voice filtering software as well. Can a machine know for certain that they faced each other incidentally or if the actually communicated? You would need to have prior data of each individuals voice pattern to filter what they said at that juncture to and see if either one or both mentioned the other's name. I'm not sure if I'm on the right track, or if it is currently technologically feasible to do all that, but that's how I would attempt to solve the problem. On Sun, Jun 17, 2012 at 5:11 AM, Mike Tintner <[email protected]> wrote: How do you get to A): ? A) Two people in a big crowded space are unlikely to notice each other from: "Sue and Jane were both at the clinic at 4.00 - did they see each other?" How do you know to ask questions about the clinic and Sue and Jane and seeing? Please outline the **logical** principles - esp. those you think existed in your head about "crowded spaces", "people" and "seeing." There are none. Logic cannot observe the world. Logic has never discovered a single new fact about the world. Logic can only work out ramifications of existing observations/facts. No real world reasoner - scientist, technologist - et al uses logic. As I more or less indicated below, your logical propositions piggybacked on my imaginative observations. Logic, like justice, is literally blind. -------------------------------------------------- From: "Ben Goertzel" <[email protected]> Sent: Saturday, June 16, 2012 11:53 PM To: "AGI" <[email protected]> Subject: Re: [agi] Real World Reasoning To know whether they could have missed each other you really do have to visualise the clinic and the possible crowds and the individual figures - and "figure out" whether they could be physically apart enough not to see each other. Reasoning here depends on the brain's imaginative capacity to move figures around the world's scenes/stages and check whether they fit together or not. Checking whether logical symbols match each other isn't going to help you, and is a fundamentally secondary operation. No, what you describe is just one possible strategy for solving the problem. OpenCog also has some code for this sort of simulation modeling, but it's not always needed... If asked whether two folks in a doctor's clinic at around the same time are likely to bump into each other, I can certainly answer the question without visualizing the clinic. If someone asked me that question "Jane and Sally were at a certain doctor's clinic at around the same time; do you think they bumped into each other?", I wouldn't necessarily ask the questioner about the specific geometry of the clinic, I might just ask something general like "How big is the clinic? How big is the waiting area? How many people tend to be there at once?" Based on this general information, I could then reason logically about the odds that Jane and Sally bumped into each other. The chain of reasoning might go something like (A and B) implies C where A) Two people in a big crowded space are unlikely to notice each other B) The doctor's office is a big crowded space, according to what I've just been told C) Jane and Sally probably didn't notice each other when they were in the doctor's office ... This is **uncertain logical reasoning** applied to commonsense knowledge. At some point in the history of the mind doing this reasoning, the proposition (A) was probably learned from experience [though it's possible to learn such things via language instead].... However, just because I learned (A) via visual, embodied experience at some point in my past, doesn't prevent me from using (A) in the future in chains of logical reasoning where I have no idea what the big crowded space in question looks like. This is part of the power of logical reasoning: it lets us draw conclusions about cases where we **lack** the concrete sensory information or episodic memory to use more direct methods. There's no contradiction between visual observation, "mind's eye" simulation, and logical reasoning. These mental processes all need to work together. There's also no contradiction between logical reasoning, as a description of what minds do sometimes, and neural network modeling as a way of describing brains. There are clear connections between logical inference rules and neural net dynamics (e.g. Hebbian learning between neuronal groups and uncertain term logic deduction). Neural nets can implement logical inference along with other cognitive methods, though in OpenCog we have not currently chosen neural nets as our implementation tool. It seems you may be unaware of the unconscious uncertain logical reasoning your mind continually does. But this deficit in your own introspective habits or capability, while unfortunate for you, shouldn't be taken as a constraint for others' AGI development work.... -- Ben G ------------------------------------------- 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/303/12578217-f409cecc Modify Your Subscription: https://www.listbox.com/member/?&; Powered by Listbox: http://www.listbox.com AGI | Archives | Modify Your Subscription ------------------------------------------- AGI Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-c97d2393 Modify Your Subscription: https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-2484a968 Powered by Listbox: http://www.listbox.com
