Hacker News Reader

Comment by riazrizvi | Hacker News Reader

An outdated service that belongs to the era of encyclopedia. Wikipedia moved us past it. ChatGPT has moved us so far past it, it's become a relic.

View on HN

Isn't it essentially a source for both of those things?

If all the sources dry up then LLM 'facts' will be time constrained.

It aggregates many public sources, so much of it is findable, but not all.

That's the idea, yes. Kill all primary sources, wound all secondary sources (examples: WaPo or "Grokipedia"), convince everyone that they should use this tertiary source whose full control is in the hands of a very few.

It being a technology that inherently has plausible deniability when it for example starts referring to itself as Mecha-Hitler is a feature, not a bug!

Which means resources like real encyclopaedias will again become financially viable.

ChatGPT and Wikipedia are not primary sources of information.

a primary source is not inherently the accurate one, and collab tools like wikipedia allow for more sources -- this makes the difference.

yeah it's game-able, and a bad actor can ruin work, but we're comparing it to a literal singular gospel source of information from a three letter agency.

p.s. I noticed I used an em dash, appropriately or not. i'm leaving it in. I like it. maybe im turning bot. changing the way I speak/type to avoid being taken that way irks me to hell.

I don’t think this is true, some of the data is not clean and is created through estimates and modeling, I’d not trust ChatGpt to get this right, and adding your own uncited models or estimates to wikipedia will get it deleted.

The World Factbook wasn't prone to hallucinations, intentional omissions, the whims of billionaires, or the unstated goals of astroturfing groups.

If the government has somewhere to tell you what it thinks is true, you can use that to double-check another part of the government that's misleading you on that same data. You can also double-check it against other sources of truth to gain insight about potential manipulation in one or more of the systems.

Here's one hot take:

https://tcf.org/content/commentary/a-well-informed-electorat...

LLM’s memory recall is extremely lossy. Facts should not be lossy.

This is so stupid. Wikipedia needs sources and citations in order to construct articles, and chatgpt needs training data to build it's models. The CIA world fact book sits at the core of training and wikipedia citations. It is the inception point of all these other services you use.

It probably also costs nothing to make. The CIA maintains dedicated analysts monitoring the world. Have those guys kick out a public report every once in a while sounds like the cheapest possible program.

Reply on news.ycombinator.com