[Replies enclosed]

---What fields would you consider analogous to AI?

Reply: I do not have the technical knowledge to answer that 
directly, however, the source of relevant analogies might so 
broad as to include almost every field (e.g. I wouldn’t be 
surprised if "thesis, antithesis, and synthesis" has an 
instantiation in the field of sculpture). 

---We can observe some general patterns.
---1. Many great discoveries were made by accident. (If the 
results were expected, it wouldn't be great).

Reply: Perhaps this would justify allocating time to random 
trial and error (of course not of anything on a finished 
product stage), although I would not be surprised if the number 
of plausible random trials was so huge and the likelihood of 
benefit so small that it wouldn’t justify resources.

---2. Many great insights are initially rejected by peers. (How 
long does it take to award a Nobel prize?)

Reply: This could indicate that a general principle that at all 
times, the set of ideas put forth a smaller amount of time into 
the past than the amount of time it would take for such an idea 
to overcome peer rejection would have a greater chance of 
containing a "diamond in the rough" than would ideas from other 
times (other things being equal).
-Mark

-- Matt Mahoney, [EMAIL PROTECTED]




----------------
> I imagine the following may have already been considered, 
> nevertheless: It would seem constructive to undertake an 
> analysis of breakthroughs in various fields (e.g. 
engineering, 
> art, chemistry) to search for patterns that might be 
> replicable. A general example of what I mean by a "pattern" 
> would be, "thesis, antithesis, synthesis." Examples of 
> patterns that such an analysis might uncover could include 
> patterns of formal logic, the novel application in one field 
of 
> a structure established in another field, the retrieval of 
> insights from historical theories or practices that were once 
> competitive, but were found inadequate and long forgotten, 
> etc. An analysis of such patterns and the identification of 
> meta-patterns would seem to require broad familiarity with 
the 
> various disciplines in which the breakthroughs occurred; 
> however, the work of identifying the patterns of specific 
> breakthroughs, which might require extensive and deep 
knowledge 
> in the respective fields, could be divided amongst various 
> experts of various fields. Perhaps something like this would 
> be worth adding to the agenda of the AI Impact Initiative or 
> some similar interdisciplinary body.
> 
> -Mark


-----
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=4007604&user_secret=8eb45b07

Reply via email to