On Nov 7, 2007 8:46 AM, Edward W. Porter <[EMAIL PROTECTED]> wrote: > > It is much easier to think how superhuman intelligences will outshine us > in the performance arena, since all one has to do is take known human > mental talents and extrapolate. It > seems to me it is more difficult to > understand what new mental competences superintelligences will have > because many of them involve abilities, that because they are outside our > competence, we have not yet thought of. > > Ed Porter
Yes, that's the main concern here. But there appear to be 3 separate varieties of "new-ness", one of which could be (and perhaps should be ??) addressed more easily than the other two. 1. abilities that currently exist which we simply haven't noticed 2. concepts that haven't yet been discovered 3. abilities that haven't yet been expressed For example, imagine if we lived in an environment where no one ever got injured, so that there would be no scrapes, bruises, paper cuts, etc... We wouldn't know of the body's ability to heal, that it could stop blood loss via clotting at the location of the injury. This would be an ability that exists but that we are unaware of. (type-1) Fields/inventions such as those mentioned by Russell, below, - "experimental mathematics, quantum chemistry and video games" would be examples of new concepts discovered using existing mental & artificial machinery. (type-2 -tough to predict) Examples of type-3, would depend on certain genes being activated to produce new competencies such as .... the ability to consciously detect and correct brain haemorhages ... or the ability to detect and eliminate cancerous growths ..... or the ability to detect and correct unhealthy mutations in an embryo. (type-3 - tough to predict) So the question is are there type-1 competencies that we are overlooking? This is relevant if we want our AGI partners to add to our abilities and not give us what we already have (but are unaware of). No matter whether AGIs help to extend our performance or our competence, they would have work with the blueprint of human intelligence. That is they would need to have a sufficiently detailed understanding of what the human brain/mind can do and how it does that. Eg: pacemakers and hearts .. etc. -Monika > > > > > -----Original Message----- > From: Russell Wallace [mailto:[EMAIL PROTECTED] > Sent: Tuesday, November 06, 2007 6:22 PM > To: agi@v2.listbox.com > Subject: Re: [agi] Questions > > > On 11/6/07, Monika Krishan <[EMAIL PROTECTED]> wrote: > > So when speaking of augmentation, a clarification would have to made > > as to whether the enhancement refers to human competence or human > > performance. ..... and hence the related issue of "discovering human > > competencies". > > Ah. *nods* Well, literally millions of volumes have been written on the > subject, so you'll need to ask a more specific question :) > > Are you asking whether computers have enabled any completely new human > competencies, anything we didn't in principle know beforehand how to do > even the tiniest bit of? There aren't a lot of examples of that (and more > or less by definition, we can't foresee future examples). Depending on how > you define the terms, experimental mathematics, quantum chemistry and > video games might mostly/almost qualify. (Programming itself, ironically, > is a mostly/almost; the world's first programmer never did get her hands > on a working computer.) > ----- This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244&id_secret=62732879-628ae0