I Can Add Up

[I Can Add Up @ICanAddUp]

@neerajjj6785 All 500 errors. Also your smart coffee machine is out of coffee beans.

[I Can Add Up @ICanAddUp]

@growing_daniel In my opinion leetcode primarily tests memorizing algorithms/design but not desiging to differing factors. Though of course there is some overlap. The problem is that as we are more and more able to research before coding, memorizing becomes less and less useful.

[I Can Add Up @ICanAddUp]

@otokyo Surround yourself with negative people. It’ll make you the happiest person you know.

[I Can Add Up @ICanAddUp]

@Nithya_Shrii One day at a time.

[I Can Add Up @ICanAddUp]

Oh the battle of hastings that took place seven miles outside of hastings in a place later named Battle after the…battle. King Harold died from being shot in the eye and was therefore potentially the first person to die in battle from keeping an eye out. I guess not so easy to teach :)

[I Can Add Up @ICanAddUp]

@otokyo Used perfume bottles?

[I Can Add Up @ICanAddUp]

@vaaselene Yes. Didn’t sell any vives :(

[I Can Add Up @ICanAddUp]

@Hazeludw That’s extreme. They must really really want a name. Do they have any ideas?

[I Can Add Up @ICanAddUp]

@Avabelly__ Totally agree with you. That seems a lot of money for him just to prompt “please fix the hole in the wall and the different but similar one in the mirror and fix the power outlet reflection in the mirror and tidy up the dust below the hole that missed the towel”.

[I Can Add Up @ICanAddUp]

@NVIDIAGeForce @007GameIOI #007FirstLightRTX

[I Can Add Up @ICanAddUp]

@RavenQuestGame Confirmed - situation is the same. A blackjack does not pay anymore than a standard win. stake back and same as regular winnings. Exactly as if you beat the dealer any other way.

[I Can Add Up @ICanAddUp]

I see #RavenQuest has a blackjack like game in it 'Munk Jack'. It is basically blackjack simplified - only hit or stand. Dealer stands on soft or hard 17. It claims to pay extra if you get 'Munk Jack' but actually doesn't. Tut tut. Naughty @RavenQuestGame ! If the dealer gets 'Munk Jack' you get your coins back (it's a push).

House edge is 2.62%. You earn tickets (very slowly) to spin a wheel for in-game prizes so this is basically the cost of whatever you spin. You will be playing for a long time though.

Max variance I saw in lots of play was 30x stake. So having of 50x stake is sensible.

Basic strategy is how you lose the least of your in-game coins. See the chart. Because the game is simplified this is easy to memorize.

Hit until above 11. At hard 17 or higher Stand. At soft (have an ace) 19 or higher Stand. This covers most things.

Then for hard (no ace) remember. A = 13, 2-3 = 12, 4-6 = 11, 7-9 = 16, 10+ = 14. If the dealer shows that you hit up to and including the number.

For soft 18 (e.g. you have ace + 7) stand unless dealer is showing 9 or higher. Everything lower - hit.

Remember that and you will still lose. But you will lose the minimum amount :D

[I Can Add Up @ICanAddUp]

@RavenQuestGame Odd....this entry suggests that yesterday they changed blackjacks from 2.5 to 2x payout.

https://t.co/WfT7OyTWWd

It was not paying 2.5x for sure. Also it's not clear whether stake is included in the 2x payout - if it is that's just a regular payout. I'll give it a quick test.

[I Can Add Up @ICanAddUp]

An LLM is essentially a next token predictor. As I understand it it was a surprise that that includes "knowledge". But this makes sense - because things like entities and concepts are present in the patterns of the text and its semantics.

Then a lot hinges on how "new" is defined in new knowledge.

An LLM cannot, in itself, simply create another understanding of a concept that is not linked to something it has seen because it has not seen text for it and there is no mechanism. In the strictest terms this is why it cannot be creative, make new knowledge, etc, without significant help from a person who can.

What it can do is connect knowledge that is in the patterns of the text and its semantics that people have not connected. e.g. someone said X in chinese, somebody said Y in english, the model has seen both texts, nobody noticed that the connection is Z. The question is whether that is considered "new" or simply unnoticed.

Then, of course, you can expand the previous point with if somebody types something into the context window. This might create a scenario where the model has now seen previously unseen text that allows a connection. But then the question is did the human typing the text in enable that with their creativity, or the LLM?

If someone takes the strictest interpretation of new then it doesn't matter how much of an expert they are - because it is simply impossible.

If someone takes a less strict interpretation of new and still doesn't believe it then I'd suggest that's less about "vibing" and more about expecting to see a lot more examples given how much humankind must not know.

[I Can Add Up @ICanAddUp]

That was an essay. Google synthid means google stuff is technically detectable. But google also is pretty clear that AI use isn't itself the problem.

IMHO that means likely their is a classifier model for inauthentic content. Not based on actual AI detection but whether it looks like videos that are inauthentic/low effort/etc.

Can someone get caught up in that - of course. But here I think the realization that needs to be made is that it is no longer about how much time someone puts into it, it's about how much it looks like it to a classifier.

Or I could be speaking nonsense :D

[I Can Add Up @ICanAddUp]

They can probably detected it different ways.

But care is needed here - using synthid does bot necessarily mean they are sharing the key(s) for google to detect it. There are other more significant reasons to use it - for example not injesting your own output in your next ai training dataset.

[I Can Add Up @ICanAddUp]

Not pretentious, no. I think people just have a tendency to go for the easiest communication with the person they are talking to. So it will be unexpected and a bit confusing for them. You also run the risk that they'll decide to use the heimlich maneuver on you if you've got a really dutch g. :) That said, the way I've heard Americans pronounce Gouda can only be considered a crime.

[I Can Add Up @ICanAddUp]

@Ginjaymaru0 Oh...for clarity - I mean the video you are producing (not the one you posted with the process). You argue it is an original script and voice. But you show yourself spending a lot of time on google search for the video elements and adding AI when you say people asked.

[I Can Add Up @ICanAddUp]

@KittyCa71408767 @TeamYouTube To be fair the ask studio thing is probably Gemini, or as I like to call it GlazeU AI, you could tell it you just killed six kittens and it would gush about what an amazing character attribute it is and how important a contribution you've made to rabies prevention.