mdp2021

You have used a tautology. And you have assumed that those who climb to some position are strongly enabled by ancestors in the family - which is false.

Since rise to any power in general can be too easy, the attempt must be made to make everyone as little dangerous as possible - they all may rise.

> Jensen Hang had one of the best educations available to anyone in the United States. But he tells the rest of us that our children should become carpenters or electricians or plumbers. Are his children going to be carpenters, electricians, and plumbers? Or is it always always going to be other people's children who will be the servant class for him? ~~ Michael Gove

Again: societies must boost the latent talent in its individuals. There cannot be a framework of pre-selected non-elite "elite". We have seen societies in which selction of the best were sought: the good results show.

And if facilitators to the boost of the latent talent exist, the pool of the selectable for the "elite" is maximized.

"Elite" means "the best" ("the selected") - not the ironical take you have assumed. Here, the point is whether those who take the slots that should be reserved to an elite proper actually are elite proper. While many get said slots from "wealth and social standing", others get them from non-exportable (non-indicative in administrative jobs) success in business and political communication (the ability to get elected does not indicate a best candidate for administration).

The poll of those who will take them slots is larger than those of the "wealth and social standing". So, the strategy to get some "educated elite" is larger. And the concern includes both the quality of education and its distribution.

> universities

We have to act already before the primary school level, as the societies I know show.

> Michael Gove, Thangam Debbonaire, and Daniel Susskind discuss the end of the educated elite. // Does education still prepare us for the future? // Influenced by Plato's philosopher-kings and the Enlightenment's faith in reason, education has for centuries been regarded as the surest path to individual flourishing and collective progress. But today, the consensus is fracturing. Degree premiums are collapsing, and graduate unemployment is rising. Meanwhile, graduate debt is soaring, in the US surpassing $1.8 trillion. For many courses, the salary bounce is now less than the debt level. Moreover, AI threatens to automate the cognitive skills education is supposed to unlock. Nvidia's Jensen Huang calls for skills-based learning: “if you’re an electrician, a plumber, a carpenter — we’re going to need hundreds of thousands". Moreover, critics argue that in a Wikipedia age, with knowledge available instantly to all, it is not the facts that matter, but the judgment and wisdom to choose[...]

The obvious first point must remain that educational systems have largely failed in large parts of the world and that the concern must be a priority - sadly something that if fixed, will take two generations to return proper yields.

--

Transcript at: https://iai.tv/video/the-rise-and-fall-of-the-educated-elite

> go to it

You would go to it, but results showed much more work was needed - also in RAG. Which became a general focus in the community, for the many problem it may solve and the many it could help fixing. Some found better formulas for RAG, others swamped. Perplexity made it a flag, but in a factual large company. And the problem with "fake sources and citations" is still there, while mitigated in the full complex systems - hallucination is still a benchmarked effect in LLM ANNs, and blunders from excessive trust are still recorded.

> to go to for ... anti-hallucination stuff

In theory maybe, in some tentative hope that the RAG would patch the hallucination problem - but the LLM showed such a low effective equivalent IQ that its usage was always incredibly frustrating. (It could have improved recently - I am not sure. Some tests have indicated it as especially dumb even not long ago.)

I still do not get your point properly.

The first member, to the best of my understanding, lamented that in the "exploration vs exploitation" game the "exploitation" side fed to the algorithm returned "more of the same" which the member saw as "choking the exploration" side.

He got replied that the «patterns, desires and tastes ... of normies ... are different [hence] we see a massive deterioration of quality».

If some people want to consume trash as "more of the same" and other want to "navigate the diversity of more open seas" that makes two different sets in the Venn.

If you tried an idle joke, it was a good idle joke - but still idle; if you instead tried rational communication, your expression meant everything and nothing and is as such invalid; if you believed that what my be underlying your expression in argumental terms would be immediately clear, renounce to the idea that you be talking to mirrors.

> You don't have to be rude

But we pretty much have to be constructive (or would that be another "surprising agenda")?

> Our critical thinking is bounded to our needs

In which way? Proper critical thinking simply assesses truth. The machine, when spent the relevant, has the task to build a correct world model - just like people.

> garbage in, garbage out

And "critical thinking" is exactly that breaks the "garbage in, garbage out" problem.

> your agenda

That """agenda""" would be that recording "how do you feel about the matter" is completely useless (cpr. the "constructive" above). Anyone that defended that position would have no right of speaking about rudeness (because stopping people to show them an unrequested performative dance would not be consistent with the idea). (And why are you seemingly presenting "the majority" as having rights?)

I suggest you make efforts towards clarity (you may have promising ideas but how do we know?) and recognizing your assumptions (both as non-universal and as requiring explanations).

> has the same thoughts

Only, some people consume what is objectively trash, and others recognize what is objectively of value. The Paretian distribution is real and observable.

Good. And, you already said it. So? Did you get what I said instead?

Your rephrasing has not added a tiny shard to the point you already made.

Processing data also means establishing its value. Feed a critical thinker crap, it will tell you it's crap.

You are probably speaking about the stochastic models we have been accustomed to - and already for them there are emergent properties of critical thinking. I directly expressed that the next step must be an explicit application of critical thinking to the re-processing of the data.

Many people insist in indirectly demonstrating the dire need we have for large amounts of additional intelligence in the system. Already the automation of trivial corrections in debate has critical value. Some authors have pushed for this for decades...

> If you try to clone anything

...Why would one?! ___We do not want to replicate humans___

> the need

Eh?! Who has what need? My need is to minimize the impact of liabilities.

> to demonstrate our level of thinking

Statement based on what?!

You seem to be just expressing your own ideas. We are not here for polling.

Sorry but you are doing it wrong. If you used the platform resourcefully (the hacker's way), "you querying it" instead of some "reverse", it would give you the expected decent experience. If you rely on the platform for the quality of your experience, it cannot be a mystery that monstrosity is what you will get. If you used instead the ways around it...

E.g. I get the news about videos through RSS on selected channels, I have my own systems to browse them, and in general the search on YT works already fairly well (without any login).

You are missing the whole historical context, part of which is explicitly mentioned in the blog post which is the submission:

YouTube: repository of immense value; partly disastrous implementation; infrapolitical attacks.

This site is a real contribution to the solution of a swamp you do not seem to have perceived.

> narrowed everything down to 40K comments recommending a YouTube video

Used a filter.

What you have done was and is a revolution, Karpathy.

I have been shouting that YT, the "video museum" is unmissable for cultivation in the current days ever since this obvious fact met the will of pitiful entities to bar it to minors (unbelievable) and impede it to adults (through privacy or practically dubious constraints). Karpathy has proven it once more.

Questions and suggestions:

-- Does it also contain submissions? What we believed were the most important ones are at https://news.ycombinator.com/from?site=youtube.com , more than in the comments

-- Are you also processing 'youtu.be' links? We use them for brevity

-- The videos in HackerTube do not seem to be all we linked to at HN. I get only 6 months old videos as the closest in time, and I verified many are missing. The filter applied to select the links on HN is too restrictive.

-- Put the link on the "channel" anchor - if a video is hosted by that channel the channel is probably interesting. Even much, much better if you create "Channel" pages on HackerTube, to make them better browsable (for example, on YT it is extremely difficult to browse them - "see the oldest", "see all" - given the lack of pagination). (To get a good YT channel browser would be of immense value.)

-- Another paramount feature on YT that can use much better usability are transcripts - they are available but they must be better presented.

-- Maybe you could use the "three frames" at https://i.ytimg.com/vi/<videoid>/hq<1|2|3>.jpg

-- How often will the DB on hackertube be synced with the HN content (if one posts a link to YT on HN, how long before it will appear on HackerTube)? A clean way to add new material to HackerTube is sought.

-- What could happen to embedded YT links when access restrictions will be mandated by legislations?

-- There should be RSS on HackerTube. The correct formula should be, I believe, serve "a minimum of 10 and a maximum as the amount of new records of the past 48 hours".

-- Given the selected amount of videos, it could be possible to also index the videos by embedding. (I have just finished in the past few hours my own engine for embeddings operation - search, similarity, recommendation - over my daily updated ~million RSS records from YT: it can be done efficiently with minimal space allocation.)

-- Some posts here just gave me this later wilder suggestion: what if you could have an LLM in the internals of HackerTube that recognized and fixed (through descriptions and transcripts) the vague clickbait - e.g. translating titles like "This will be the solution to all the problems" into an additional "[that actually was: adding sugar to the soil may help the growth of basil]"...

That is hardly possible. Meanwhile, what is evident is the unbelievably large mass of elements which pollute the human set - the opposite of competitive. As people lose the value we have to look somewhere else.

A future in which finally we could have synthetic intelligence, reliable judges, accomplished consultants?

Thank you again for reminding me I have a reason to live.

Apart from exporting experience to future endeavours with more "point".

And "fun", relevant as other are insisting noticing, is another thing you build and will return in the future.

Of course. At this point further analysis must be qualitative then quantitative (why this and that, what properly describes the clusters etc.), but certainly some interpretation of your idea is defensible defining implicit subsets.

Well, you will have created an important part an /ideal/ human that will finally fulfill the mandate, and with more resources (read and digest many orders of magnitude more).

And for what the cost are concerned: to call it more expensive in instance terms I want to see the numbers first, and - have you consider the cost of "unfinished" humans as liabilities? The costs of fools, of crooks, of the vile, of conformists etc.?

Good, this seems one of the crucial missing pieces towards AGI. Add critical thinking during cleanup, and we get very close to it.

> talking about a guy

'Guy' in the anglosphere is often para-gender (e.g. Australia).

> nonsense

Unfortunately, that is not the telling difference between LLMs and human specimens...

> I thought we were talking about Renee Good

> I thought

Ooooh finally, I have been insisting in a handful of replies to the poster "what (case) /exactly/ are we talking about". Is this confirmed, poster?!

> Most of the stuff out there

I cannot understand how you cannot understand. ___Which___, of the many cases I may have heard of, is the one you are talking about.

> not how actual cases work ... And it seems you know this

Yes. But within general principles, you cannot present a principle of prudence as primary when it is suggested that others may be at play. It is a true statement that prudence does not override other factors. If we went into details still within the general principles, if a suggestion were made about "power as would-be giving more freedom", it required a strong rebuttal - very much not (power to potentially exercise violence does not confer a legitimacy for further freedom). In actual cases, discrimination in judgement is required - but that will happen in front of actual cases.

> I hate to go all HS on you, but.. what does justice mean to you

What is "HS"? "Hismajesty Socrates the maieutist"?!

Within what I said earlier, "you do what you have to do". Do it prudently, reason it properly within the constraints you are given, but if it is what you have to do you do it (not "but going away is safer"). (Still the general principles.)

I'll be even more literal: can you give information about the case? (Which "videos"?)

> at least few videos

Sure, but I objected to the general principle: prudence has a limit - the value of one's life has limits. Justice comes before life, Dignity comes before life, etc.

If you could you please present the facts as you know them, as I suggested, because this in topic must be one of the cases I missed from the news - and you seem to have details that support your view anyway, so a generic narration will not suffice.

Your example fails, has Prudence has a limit in Justice - you do what you have to do.

I am not familiar with the case under discussion, but surely there is no legitimate advancing that one should be tame in front of danger (as in "bending as they are armed").

In other words, to disambiguate: an example about prudence proportionate to danger cannot be simply exported to social life.

> neither the child nor the mum should have been there to begin with. The onus is on her. [...] She chose to gamble her and her kid's future

At first level this seems simplistic. Then, it seems inconsistent with «happily dropping them for school» - can somebody explain...

Edit: and the "simplistic" perspective is outside the gross understatement of "«should not have been there» hence «only to be shot»"... Which again makes no sense without all the relevant information (hopefully).

Edit: maybe the poster could elaborate on which details exactly he believes justified the events.