[agi] Measuerabel Fitness Functions?.... Flow charts? Source Code? .. Computing Intelligence? How too? ................. ping
What are the measurable fitness functions that can be built into AI? Dan Goe - Original Message - From: William Pearson [EMAIL PROTECTED] To: agi@v2.listbox.com Sent: Thursday, July 06, 2006 5:21 AM Subject: Re: [agi] Flow charts? Source Code? .. Computing Intelligence? How too? . ping On 06/07/06, Russell Wallace [EMAIL PROTECTED] wrote: On Wed, 05 Jul 2006 15:58:28 -500, [EMAIL PROTECTED] That just gets you a circular definition: if intelligence is the ability to self-improve, what counts as improvement? Change in the direction of greater intelligence? But then what's intelligence? Etc. Basically the problem with all this is that there's no such thing as intelligence in the sense of a mathematical property of an algorithm. I would agree with you here. Intelligence is an informal term for certain types of effectiveness of an algorithm in carrying out tasks in an environment. So the first thing you need to do is figure out what sort of environments you want your AI system to work in, and what sort of tasks you want it to carry out. How would you define the sorts of tasks humans are designed to carry out? I can't see an easy way of categorising all the problems individual humans have shown there worth at, such as key-hole surgery, fighter piloting, cryptography and quantum physics. I'm not saying a human is a general purpose problem solver, just that humans seem to have the ability to mold themselves to many different tasks, that do not seem to be genetically specified. Will Pearson --- To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/[EMAIL PROTECTED] -- No virus found in this incoming message. Checked by AVG Free Edition. Version: 7.1.394 / Virus Database: 268.9.9/382 - Release Date: 7/4/2006 --- To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/[EMAIL PROTECTED]
[agi] Nothing can go wrong... Two draft papers: AI and existential risk; heuristics and biases
The AI system could be built as more of an advisor of actions that we might take. The investment field has already progressed into automated program trading. I would bet that the investment brokers have some human monitors watching and maybe even approving the trading. But you have heard the story many times... Nothing can go wrong... go wrong... go wrong Dan Goe - Original Message - From: BillK [EMAIL PROTECTED] To: agi@v2.listbox.com Cc: [EMAIL PROTECTED] Sent: Wednesday, June 07, 2006 4:08 AM Subject: Re: [agi] Two draft papers: AI and existential risk; heuristics and biases On 6/7/06, Eliezer S. Yudkowsky wrote: snip Because this is a young field, how much mileage you get out will be determined in large part by how much sweat you put in. That's the simple practical truth. The reasons why you do X are irrelevant given that you do X; they're screened off, in Pearl's terminology. It doesn't matter how good your excuse is for putting off work on Friendly AI, or for not building emergency shutdown features, given that that's what you actually do. And this is the complaint of IT security professionals the world over; that people would rather not think about IT security, that they would rather do the minimum possible and just get it over with and go back to their day jobs. Who can blame them for such human frailty? But the result is poor IT security. This is the real world that you have to deal with. You cannot get the funding, or the time, to do the job properly, because there is always pressure to be the first to market. AGI is so tricky a problem that just getting it to work at all is regarded as a minor miracle. (Like the early days of computers and the internet). Implement first, we can always patch it afterwards. A much more worrying consideration, of course, is that the people with the most resources, DARPA (and the Chinese) want an AGI to help them kill their enemies. For defensive reasons only, naturally. When AGI is being developed as a weapon by massive government resources, AGI ethics and being Friendly doesn't even get into the specification. Following orders from the human owners does. BillK --- To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/[EMAIL PROTECTED] -- No virus found in this incoming message. Checked by AVG Free Edition. Version: 7.1.394 / Virus Database: 268.8.2/357 - Release Date: 6/6/2006 --- To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/[EMAIL PROTECTED]
[agi] Best methods of Knowledge Representaion and Advantages Disadvantages?
- Original Message - From: "Ben Goertzel" [EMAIL PROTECTED] To: agi@v2.listbox.com Sent: Wednesday, May 31, 2006 8:25 AM Subject: Re: [agi] Types of Knowledge Representaion and Advantages Disadvantages? Well, the main disadvantage of not representing knowledge is that doing so makes you completely unintelligent ;-) [Of course, whether or not this is really a disadvantage is a philosophical question, I suppose. It has been said that "ignorance is bliss" ... ] Those that choose this path are not likely to achieve success. Seriously: Do you mean to suggest that some intelligent systems *don't* contain any (even implicit) representation of knowledge? My question was more to thedifferent methodology of knowledge Representations (KR) andKnowledge Base (KB) types of designs and their performance at retrieving facts in respect to the computer time/computer instructions required to retrieve factsand storage requirements. I have seen this claim made by some advocates of self-organizing-systems approaches to building and analyzing intelligent systems, but I have always felt it to be a kind of "game with words"... (Feel free to argue otherwise, though!) The product configuration baselineshould be functionally interwoven with the sophisticated software and adds many different trade off considerations. (wordy pun) All I am interested inis what works fast and within the limits of resources. IMO, all intelligent systems represent knowledge internally in some sense, and the right question is what methods are best (in what senses) for doing so What methods are best (concerning fast retrieval, low number ofcomputer instructions, lowmemory requirement) now this can also mean that these facts are somehow zipped/compressed to reduce memory storage requirements. Maybe someone might know of how much (percent) compression can be achieved to help reduce the KB to some more manageable size, yet again there are some computer instructions/time needed to use this methodology. For instance, in an attractor neural net, each piece of knowledge is stored in a wholly distributed way, interpenetrated with other pieces of knowledge. In a traditional semantic net OTOH, pieces of knowledge are stored separately and distinctly without interpenetration. In Novamente's hybrid design there is both a distinct and an interpenetrative/holistic aspect to knowledge representation. The advantages and disadvantages of these different KR strategies may be subtle to understand... -- Ben G On 5/30/06, Danny G. Goe [EMAIL PROTECTED] wrote: Can someone elaborate on the advantages and disadvantages of Knowledge Representation(KR)? Dan Goe To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/[EMAIL PROTECTED]
[agi] Data there vs data not there, Limits to storage?
What are the Novamate limits of storage? Does Novamate look for what is there(data mining) as well as what is not there? How big is Novamate? Reading/Writing data can result in I/O bound systems. Dan Goe - Original Message - From: Ben Goertzel [EMAIL PROTECTED] To: agi@v2.listbox.com Sent: Wednesday, May 31, 2006 10:28 AM Subject: Re: [agi] Best methods of Knowledge Representaion and Advantages Disadvantages? My question was more to the different methodology of knowledge Representations (KR) and Knowledge Base (KB) types of designs and their performance at retrieving facts in respect to the computer time/computer instructions required to retrieve facts and storage requirements. Well, viewing the memory problem as retrieving facts is in itself a serious philosophical statement ... Storing crisp, declarative facts efficiently is not *such* a hard problem; one can use for instance a hypergraph data structure, with multiple indices constructed to make frequent queries rapid. One can even automate the construction of new indices. The space/time tradeoff rears its head here in that more indices means faster access but more memory usage. The subtler conceptual issue, IMO, regards how to store uncertain, context-dependent patterns of knowledge: these may be stored in the same manner as crisp declarative facts, or in a thoroughly distributed way as in an Attractor Neural Net, or via some combination approach... -- Ben --- To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/[EMAIL PROTECTED] -- No virus found in this incoming message. Checked by AVG Free Edition. Version: 7.1.394 / Virus Database: 268.8.0/353 - Release Date: 5/31/2006 --- To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/[EMAIL PROTECTED]
[agi] Types of Knowledge Representaion and Advantages Disadvantages?
Can someone elaborate on the advantages and disadvantages of Knowledge Representation(KR)? Dan Goe To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/[EMAIL PROTECTED]
[agi] Google wants AI for search... The first step..
Fellow AI ... "Seems thatGoogle wants a searchengine that knows exactly what you want"... http://news.google.com/news?ie=utf8oe=utf8persist=1hl=enclient=googlencl=http://news.independent.co.uk/business/news/article570273.ece I doubt once that google gets this far they will stop there. They have the means and the structure to do AI totally. The question remains is who is going to get AI developed first and what will they use it for? We live in interesting times. These future events will have a most profound effect upon societies around the world. Comments? Dan Goe To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/[EMAIL PROTECTED]
Re: [agi] Cell-DG
I think this begs the question of how you factor in the learning cost. CPU-time? Resources? Memory? Total Instruction counts? If you can arrive at the same answer with less instructions executed and or less resources isn't that a better model? Weighing the cost will be based on the availability of those Resources. What resources give the highest rate of learning? CPU-time, memory? Is an Intelligence Quotient a good way to find the learning curve? Or some other method(s) to find the learning rate. I am sure that this will be processed on clusters. Some neural nets might run in background while different mutations are run. Previous arrived computational states maybe either continuing or become fixed at some point in time. If any configuration creates a learning system that the next generation of mutations generates a value greater than 1 from the previous generation you can then start to determine the evolution rates. When you start your process you will have to run a large number of test generated methods and determine if any show promise to learning while some might work better early on others will mutate into higher learning curves as the evolution continues. You will have to run a large number of permutations of all the learning methods to find the optimal mix to obtain a high learning curve. If you decide to add any other learning methods, the new method will have to be tested with all the others. First time runs will generate high learning rates, but level off as the known knowledge gets aborbed by any given configuration. Comments? Dan Goe - Original Message - From: Brad Wyble [EMAIL PROTECTED] To: agi@v2.listbox.com Sent: Thursday, February 10, 2005 10:15 AM Subject: Re: [agi] Cell I'd like to start off by saying that I have officially made the transition into old crank. It's a shame it's happened so early in my life, but it had to happen sometime. So take my comments in that context. If I've ever had a defined role on this list, it's in trying to keep the pies from flying into the sky. Evolution is limited by mutation rates and generation times. Mammals need from 1 to 15 years before they reach reproductive age. Generation That time is not useless or wasted. Their brains are acquiring information, molding themselves. I don't think you can just skip it. times are long and evolution is slow. A computer could eventually simulate 10^9 (or 10^20, or whatever) generations per second, and multiple mutation rates (to find optimal evolutionary methodologies). It can already do as many operations per second, it just needs to be able to do them for billions of agents. 10^ 9 generations per second? This rate depends(inversely) on the complexity of your organism. And while fitness functions for simple ant AI's are (relatvely) simple to write and evaluate, when you start talking about human level AI, you need a very thorugh competition, involving much scoial interaction. This takes *time* whether simulated time or realtime, it will add up. A simple model of interaction between AI's will give you simple AI's. We didn't start getting really smart until we could exchange meaningful ideas. But yes it's true, there are stupidly insane emounts of CPU power that would give us AI instantly (although it would be so alien to us that we'd have no idea how to communicate with it). However nothing that we'll get in the next 100 century will be so vast. You'd need a computer many times the size of the earth to generate AI through evolution in a reasonable time frame. That's not a question that I'm equipped to answer, but my educated opinion is that when we can do 10^20 flops, it'll happen. Of course, rationally designed AI could happen under far, far less computing power, if we know how to do it. I'd be careful throwing around guesses like that. You're dealing with so many layers of unknown. Before the accusation comes, I'm not saying these problems are unsolvable. I'm just saying that (barring planetoid computers) sufficient hardware is a tiny fraction of the problem. But I'm hearing a disconcerting level of optimism here that if we just wait long enough, it'll happen on all of our desktops with off-the shelf AI building kits. Let me defuse another criticism of my perspective, I'm not saying we need to copy the brain. However, the brain is an excellent lesson of how Hard this problem is and should certainly be embraced as such. -Brad --- To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/[EMAIL PROTECTED] --- To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/[EMAIL PROTECTED]
