\> statistical learning theory does not adequately model the macro-behavior of very large models.
Might you please elaborate on this? I recognize that "artificial neural networks are lossy de/compression algorithms" does not enumerate the nuances of these structures, but am curious whether anything in particular is both interesting and missing from SLT.
SLT typically uses empirical risk minimization, leading to the bias-variance decomposition and a unimodal extremum as the monotonically decreasing bias supposedly balances against the monotonically increasing variance. We now know this does not accurately model overparameterized models, which exhibit double descent, and other phenomena like grokking. To explain them you have to look past classical statistics to statistical mechanics.
Depends upon the user. Are you fast (and confident) at evaluating the output and discarding the bad suggestions? This is why I think using AI hurts some developers and helps others, and the usefulness is best for those who already have a good deal of experience.
I don't ever use it for fire and forget, but I've been wondering how well that might work in small side projects where hidden bugs aren't a big concern. Like using a fire and forget to spin up a small javascript game. But never in production code that I might get a 2am Saturday incident call on.
Late binding of metadata is critical to my notetaking, and the primary advantage of digital over physical media. I am curious whether any technology like the Nuwa pen will be able to provide a transparent bridge between the two.
If a person could unequivocally understand themselves without understanding their past, then the trivial understanding of mere knowledge of the past should have no impact on their current self-understanding capabilities. Thus, a person with ongoing amnesia, unable to form memories which extend beyond a day's context window, would have no disadvantage in terms of self-understanding.
Further, history isn't just a series of events that happened long ago. It’s the story of how we - communities, nations, and civilizations have evolved over time. The experiences of previous generations shape the social, cultural, political, and economic structures that define a society today. Understanding this history allows a community to grasp why they are the way they are. For instance, the American Civil Rights Movement helps explain contemporary discussions on race in the United States. Without historical context, many aspects of the current social fabric would seem inexplicable or disconnected.-
Collective memory, as Halbwachs would have it, is like the shared pool of information and experiences held by a group of people. This collective memory is essential for maintaining cultural continuity across generations. When a community loses touch with its history, it risks losing the cohesion that binds its members together, which can lead to a sort of cultural amnesia. This concept isn’t just for nations; it applies to any group with a shared identity, including religious communities, ethnic groups, and even professional organizations.-
For example, indigenous communities often emphasize the importance of oral histories and traditions because these narratives carry the accumulated wisdom and identity of their people. If these stories are lost, so too is a sense of who they are as a distinct group.-
An understanding of history also allows a community to make informed decisions about the future. When people understand the causes and consequences of past events, they are better equipped to navigate present challenges. Santayana's quote, "Those who cannot remember the past are condemned to repeat it," might sound like a cliché, but it's rooted in the idea that historical awareness helps societies avoid repeating the mistakes of the past. The European Union, for instance, was built on the memory of two world wars, which profoundly influenced its emphasis on cooperation and integration to prevent similar outcomes. Notwithstanding this whole mess in Ukraine ...
You've just made a whole load of claims this without backing any of them up. But okay, let's take one which does make sense:
> When people understand the causes and consequences of past events, they are better equipped to navigate present challenges
To a small degree. Most of this boils down to the bleeding obvious, like "avoid war when you can because it's hideously expensive", treat people with decency etc. is there really much more to it than that?
And you talk about cultural continuity – some of these continuities are not good, q.v. the Taliban. Also cultures change very rapidly – the kind of homophobia that was acceptable or even encouraged when I was a kid is now seen very much as a serious transgression or crime (speaking as a straight bloke, I'm glad it; I'm very glad things have changed). We haven't achieved gender parity in pay, we getting there. You can't stick your hands all over a woman as you could in the 60s and expect to get away with it (ditto good).
Well I don't know. I can't say I'm convinced by what you've written but I appreciate the answer anyway. Thanks.
> some of these continuities are not good, q.v. the Taliban
Indeed. Good point. As you well point out, in this case the past determines an outcome for the worse ...
PS. That said - I found interesting how, in a recent "live from a Taliban courtroom" video, the Sharia judge made a point along the lines of "We are lucky these aren't medieval times anymore and we have Sharia law with us ..."
They considered themselves an advance upon a certain past. Progress.-
Very cute but that's all. No reason for me to consider it true. This if we are making unbacked claims then let me riposte thus, "the past is another country, they do things differently there". Which if true would mean the past has little bearing on the present – if true.
I'm riffing on bluestein, not defending lo_zamoyski — as far as I'm concerned, the invention of the web has made it far easier to go through the pockets of older cultures and discover what sorts of things they have that are worth taking a five-finger discount on.
(then again, I also find it worth learning how things are done in other countries, so I believe both the Durants' statement and your proposed riposte may be true at the same time)
[as for the truth: from a DBMS' WAL (the past) one may always reconstruct its tables, but for queries it's more convenient to always have them materialised (the present); does that make sense?]
The first order theories of the real numbers using addition and multiplication are decidable [0], while FO theories of the integers under the same operations are not.
The Curry-Howard Correspondence is more of an observation than a theorem, but insofar as it describes the general concept of relating logics and programming languages, what is the CHC for logic programming languages? If strikes me that LP directly instantiates and manipulates logical expressions, in a "self-dual" sense; or, is it that the various "proof procedures" of the logic side map to "proof search" algorithms on the computing side?
Interesting, I have never encountered this initialism in the wild, to my recollection: https://en.wiktionary.org/wiki/f.e.#English