@marshall_0i

Indeed! It is a profound analogy.

I would like to add a couple of points that I believe are related to Ted Chiang's on knowledge generation, compression, complexity, and communication vis-à-vis LLMs (especially given the role these large and powerful corporations play in our day-to-day lives, and look increasingly indistinguishable from the role of states).

Both are from James C Scott, the great anthropologist who brilliantly captures these ideas in his work "Seeing Like a State: How Certain Schemes to Improve the Human Condition Have Failed"

1) The first of them is what he calls "legibility", and thus a simplification. Simplification (a form of compression) that turned out to be disastrous because of what he calls, "authoritarian high modernism".
This is how he explains legibility:

"The more I examined these efforts at sedentarization, the more I came to see them as a state’s attempt to make a society legible, to arrange the population in ways that simplified the classic state functions of taxation, conscription, and prevention of rebellion. Having begun to think in these terms, I began to see legibility as a central problem in statecraft. The pre-modern state was, in many crucial respects, particularly blind; it knew precious little about its subjects, their wealth, their landholdings and yields, their location, their very identity. It lacked anything like a detailed “map” of its terrain and its people."

2) The second idea is towards the end of the book, a type of knowledge he calls "metis" to contrast it to the knowledge a lot of us at the core of society generate and engage in, "techne".

Techne consists of our legible plans, rules, social codes, principles etc etc., including science and policy.

Metis on the other hand, is the thinking and knowledge process
that has to deal with the messy unpredictable complexity of both the human and natural world in the here and now. Thus practical skills and common sense improvisations to deal with such situations, which by their very nature are know-hows locally rooted, hardly transmittable, or legible, and thus harder or impossible to be either compressed or simplified.

I believe both "legibility" and "metis" along with lossy compression are equally applicable to LLMs.

Loosely speaking LLMs are like the "five year central plans" for organizing knowledge that Scott critiques in his book.

After all, Google does say their aim is to "organize the world's information and make it universally accessible and useful".

#legibility #metis #JamesCScott #chatgpt #LLMs #complexity #critique #mapterritoryrelation #AI #SeeingLikeAState

"A Non-Aristotelian System and Its Necessity for Rigour In Mathematics and Physics", Alfred Korzybski (1931) [pdf]

https://www.semanticscholar.org/paper/A-NON-ARISTOTELIAN-SYSTEM-AND-ITS-NECESSITY-FOR-IN-Korzybski/c29161b8e0971844c0919ea070f6868d28df7adf

Direct PDF link: http://esgs.free.fr/uk/art/sands-sup3.pdf

Paper is source of the phrase "map-terrain relation", a/k/a "the map is not the territory".

See: https://en.wikipedia.org/wiki/Map%E2%80%93territory_relation

#AlfredKorzybski #MapTerritoryRelation

[PDF] A NON-ARISTOTELIAN SYSTEM AND ITS NECESSITY FOR RIGOUR IN MATHEMATICS AND PHYSICS | Semantic Scholar

A very extensive literature shows that the problems of ‘infinity’ pervade human psycho-logical reactions, starting from the lowest stage of human development up to the present and that without some theory of ‘infinity’, modern mathematics would be impossible. Up to date, no satisfactory theory of infinity, on which all mathematicians could agree, has been produced. The results are rather bewildering because what appears to some prominent mathematicians as perfectly sound mathematics is evaluated by other equally prominent scientists as a ‘mental’ disease (Poincaré); or we find opinions that a large portion of mathematics is devoid of proof and has to be accepted on faith; or that some parts of mathematics must be treated as non-sense (Kronecker, Brouwer, Weyl. ,). ‘There are eminent scholars on both sides and the chance of reaching an agreement within a finite period is practically excluded’, says Brouwer, and certainly such a state of affairs does not allow us to have any satisfactory modern standards of proof and rigour; the last thing we should expect in mathematics. The majority of those mathematicians who take interest in the soundness of their science seem to believe that the main difficulty centres around the validity of the ‘law of excluded third’ (‘A is B. or not B’) of the accepted, sharply two-valued, chrisippian form of A ‘logic’. They disregard the fact that we are born, bred, educated, speak a language, live under conditions, institutions. , which still remain desperately A or even pre-aristotelian. If we attempt to reject one of the two-valued ‘laws of thought’ or postulates of the A-system but retain A or pre-aristotelian elementalistic ‘psychologies’, ‘logic’, and s.r, no agreement in ‘a finite period’ can be expected, and the present mathematical chaos would continue.

Meta: The Wikipedia article on "On Exactitude in Science" is not only longer and larger than its subject, but contains its subject within it:

https://en.wikipedia.org/wiki/On_Exactitude_in_Science

#MapTerritoryRelation

On Exactitude in Science - Wikipedia