If Fingolfin is greatest swordsman, why does AI say Turin is greater?

[Ent]

For decades This is how you "googled" something:

youths these days..! (sigh.)
having a passel of elven grandkids can be a challenge at times.

 

[Ent]

Fingolfin fierce and elvish-eared,
grey eyes as deep as Mirrormere,
once taunted Bauglir Insincere,
and cut his foot from there to here!

But Túrin too had great career,
in hist'ry dark did interfere,
how many orks did disappear!
But oops his marriage - did you hear?

O who is greater - please be clear!
and may opinions, far and near,
here reappear from year to year.

🐾

OK I gotta ask, as I don't recognize this...
Is it an Elthiran original? Paws and all?

 

[Glorfindel Goldtress]

@Glorfindel Goldtress
For decades This is how you "googled" something: you went to a library and used its card catalog (aka Library Catalog)
Each drawer probably contains at least 100 cards.
Every book in the library has at least one card associated with it, but usually more:; one for locating a book by Title, one for locating it buy the Author, and one for locating it by its Subject.
Go HERE


View attachment 44810


View attachment 44811
Card catalog at Yale's Library

Actually, this sounds delightful. I kind of wish we had one locally. Libraries are plentiful... but not with this feature.

 

[Deimos]

Actually, this sounds delightful. I kind of wish we had one locally. Libraries are plentiful... but not with this feature.

Past tense....use the past tense.
Card catalogs are dinosaurs...extinct.

 

[Ricardo]

We have to understand that AI really is a series of very efficient algorithms run on powerful serves and what it does is data mining.

The accuracy will depend on many factors, but primarily on what sources the agent has access to as well as how those sources are prioritized.

As @Ent said, they are great for a start on a topic. Sometimes you know something but can't remember where you saw it. The AI will likely put you in the right direction.

I do find in general that Gemini produces better results, probably because it leverages Google search engine to find better sources. Google has decades of experience with ranking topics.

Also, Gemini is currenttly integrated on the search engine, which means it probably receives more requests than any other AI agent and with that it can improve its LMLs.

Yet, All AI agents need human (adult) supervision and AI is incapable of feeling, or being creative (except by basing on existing work). It's also unable of rationalizing, inferring and other aspects of human intelligence, even though sometimes it looks like it does, but it doesn't.

 

[Elthir]

OK I gotta ask, as I don't recognize this...
Is it an Elthiran original? Paws and all?

Yes. Tday I awoke all stuck 'n syllables.

I'm btter now. 'Cept for som mssing vowls.

 

[Ent]

Yes. Tday I awoke all stuck 'n syllables.

I'm btter now. 'Cept for som mssing vowls.

Well. It was quite well done.
And, Ent was quite surprised, not having observed Elthirian magic of this kind in the past, from his shelf outside his cave.
Now, having seen delivery of rhyme from yet another unexpected source, of course - expectations abound.

But I WILL remember - what will I remember? (often very hard to say).
Ah yes... I will remember to 'hold my expectations loosely.'

 

[Deimos]

We have to understand that AI really is a series of very efficient algorithms run on powerful serves and what it does is data mining.

The accuracy will depend on many factors, but primarily on what sources the agent has access to as well as how those sources are prioritized.

As @Ent said, they are great for a start on a topic. Sometimes you know something but can't remember where you saw it. The AI will likely put you in the right direction.

I do find in general that Gemini produces better results, probably because it leverages Google search engine to find better sources. Google has decades of experience with ranking topics.

Also, Gemini is currenttly integrated on the search engine, which means it probably receives more requests than any other AI agent and with that it can improve its LMLs.

Yet, All AI agents need human (adult) supervision and AI is incapable of feeling, or being creative (except by basing on existing work). It's also unable of rationalizing, inferring and other aspects of human intelligence, even though sometimes it looks like it does, but it doesn't.

And (as I have repeatedly stated) depending on how a question is worded, and despite using precisely accurate vocabulary with impeccable grammar, AI at one time will give you one answer, then you reword the question (still maintaining precise vocabulary and grammar) and it gives you the exactly and precisely opposite answer.
So it's useful as a starting point (as others have stated), but for me that's its only use. Caveant lector et scriptor.

 

[Ricardo]

And (as I have repeatedly stated) depending on how a question is worded, and despite using precisely accurate vocabulary with impeccable grammar, AI at one time will give you one answer, then you reword the question (still maintaining precise vocabulary and grammar) and it gives you the exactly and precisely opposite answer.
So it's useful as a starting point (as others have stated), but for me that's its only use. Caveant lector et scriptor.

you are not wrong, the use natural languages queries and the agent can't really understand anything unless it's taught, different wording or word order can mean different queries.

I think AI is very useful in some very specialized scenarios, like writing computer code, the content is less ambiguous, agents like Gifhub Copilot are very valuable to programmers (like me), but you always have to use your judgement to decide if what is suggested by the agent is correct.

 

[Glorfindel Goldtress]

you always have to use your judgement to decide if what is suggested by the agent is correct.

A solid choice in any circumstance. To "test the spirits" is applicable anywhere.

 

[Ricardo]

A solid choice in any circumstance. To "test the spirits" is applicable anywhere.

Yes, but this is where the problem is.

In the example of Github Copilot, I am qualified to judge the outcome because I am a programmer.

If a doctor ask AI about an illness, they are qualified to evaluate the response.

But the real issue is when someone ask a question about a topic they don't understand. It's perfectly ok as long as they are not making decisions based on that, and using AI for health diagnosis is unfortunately common and very dangerous.

 

[d4rk3lf]

The real issue with AI (at least for me) is that confident tone when AI is providing you some false information, and then trying to defend it, and then finally apologies when you prove it wrong.
I got that many times for various topics, and with creation of some images or videos for projects.

It would be much better if AI could respond with something like "As far as I know... ", or "As far as I can research this topic...".

 

[Deimos]

Yes, but this is where the problem is.
In the example of Github Copilot, I am qualified to judge the outcome because I am a programmer.
If a doctor ask AI about an illness, they are qualified to evaluate the response.
But the real issue is when someone ask a question about a topic they don't understand. It's perfectly ok as long as they are not making decisions based on that, and using AI for health diagnosis is unfortunately common and very dangerous.

THaaaaaank Youuuuuu.....🙂

This is (analagously speaking) what happens when someone uses the GPS in their car for a longish trip through an unknown area.
GPS directs them the shortest route which happens to take them on , say, unimproved dirt roads up the side of a mountain and down the other side which is damn near a sheer vertical drop. People (usually retirees) have died following their car's GPS instructions.
True occurrance ... "Death by GPS"

I can see the same thing happening mutatis mutandis with the "medical advice/directives" version of AI

 

[Ricardo]

The real issue with AI (at least for me) is that confident tone when AI is providing you some false information, and then trying to defend it, and then finally apologies when you prove it wrong.
I got that many times for various topics, and with creation of some images or videos for projects.

It would be much better if AI could respond with something like "As far as I know... ", or "As far as I can research this topic...".

sounds like a politician talking.

 

[Ricardo]

THaaaaaank Youuuuuu.....🙂

This is (analagously speaking) what happens when someone uses the GPS in their car for a longish trip through an unknown area.
GPS directs them the shortest route which happens to take them on , say, unimproved dirt roads up the side of a mountain and down the other side which is damn near a sheer vertical drop. People (usually retirees) have died following their car's GPS instructions.
True occurrance ... "Death by GPS"

I can see the same thing happening mutatis mutandis with the "medical advice/directives" version of AI

Yes I agree, I love technology and I used it a lot in different formats, but it's not for everyone.

In my city for example, they removed the bus schedules and instead put signs saying you should use the apps to get the schedules, but it fails to acknowledge that not everyone has a smartphone or know how to use it.

Same happens with checkins at airports, some airline would charge a fee for in person check in but not everyone is comfortable with doing it online.

And I think we have veered way out of topic 🤦‍♂️

 

[Deimos]

Yes I agree, I love technology and I used it a lot in different formats, but it's not for everyone.
In my city for example, they removed the bus schedules and instead put signs saying you should use the apps to get the schedules, but it fails to acknowledge that not everyone has a smartphone or know how to use it.
Same happens with checkins at airports, some airline would charge a fee for in person check in but not everyone is comfortable with doing it online.
And I think we have veered way out of topic 🤦‍♂️

1) what city? I have a dumbphone that doesn't do apps (forewarned is forearmed 😉)

2) Oh, that never happens on our forum 🙄😁

 

[Ricardo]

1) what city? I have a dumbphone that doesn't do apps (forewarned is forearmed 😉)

2) Oh, that never happens on our forum 🙄😁

Toronto

 

[Deimos]

Toronto

OK ... thanks.
I won't be going to Toronto for a while, certainly not in the coming months.....brrrrr!....

 

[d4rk3lf]

sounds like a politician talking.

HA!
The funny thing.. on some other forum.. some time a go... I asked a question like "why (western) people always have this sort of polite speech".
Like somebody post a crap of picture, or some vfx animation, and everybody says "Hey it's great, just keep working on it", and I say "No, it's not great. It's not good. You miss this and this, and you can't skip fundamentals on that, because it show's in your work... etc.. try this... try that to improve".
And then somebody jump in and say: "Yeah. But that's only your opinion".
And I am always like: "What? But of course it's my opinion. I said it. Have I not"?
So I was always confused with this "this is only your opinion" on a sentences that I personally said. Who's else opinion is it, then if not mine?

Then I realized that in their eyes my comments actually sounded rude (and I had no idea).. and they were afraid that I don't hurt the artis (again, not my intention at all)... so these days I am trying to be more careful... but you know.. it's a matter of culture I guess.
For example, what I find rude (in my culture) is Western use of term "artist".
In the west.. artist is everyone... I am vfx artist.. industrial artist... dev design artist... makeup artist... plumber artist...
In the east, 2 decades a go... it was like... even the best artis would say publicly "I have no right to call myself an artist".. so when someone was publicly recognized like artist, it was much more then phd status, for example.
And even agnowledged, these artists would refuse to call themself artist, because of admiration of "true artist" they admire, and they don't feel their equal.

Sorry for Off Topic.. now let's get back to first Off Topic.. about AI.

So, my friend Ricardo, while I don't really like people being overpolite, and I encourage them to be much more open, I think that AI must be programmed to be very very careful about the stuff it claims. To not claim it with that confidence. Not because me or you. But there will be someone (or many) who will take AI confident answer as a fact, and that is not good.

 

[Ricardo]

So, my friend Ricardo, while I don't really like people being overpolite, and I encourage them to be much more open, I think that AI must be programmed to be very very careful about the stuff it claims. To not claim it with that confidence. Not because me or you. But there will be someone (or many) who will take AI confident answer as a fact, and that is not good.

Yes, but that's not just a problem with AI. The problem isn't the technology but people. Look a social media, way before AI. Anybody can post anything and affirm it as true and a lot of people believe it, even people that are supposed to be more educated (I don't want to say names here) but there are several high-profile people who have embraced dubious or unproven facts and had made policies out of it.

So, we don't need to change AI, we need to change people's brains.