Doubleshot vs Ranged Power

[Lacci]

It´s like in Hitchhikers Guide to the Galaxy, where they ask the super computer about the meaning of life, and its answer is 42.
I guess the math there was also correct, and it was all about asking the right question.
Douglas Adams was ahead of his time...

 

[Marshal_Lannes]

If you think AI is some magical shortcut to being able to accurately calculate DPS and other complex things in DDO

Asked and answered. I've already defined AI as practical magic. The belief that DDO is "too complex" for AI to figure out when it's going to be instrumental in sending us to Mars is a relic of old-world thinking - "'ere be monsters!" on the map.

 

[LurkingVeteran]

Magical thinking - untethered from the old world - is clearly the future.

 

[torkz]

Asked and answered. I've already defined AI as practical magic. The belief that DDO is "too complex" for AI to figure out when it's going to be instrumental in sending us to Mars is a relic of old-world thinking - "'ere be monsters!" on the map.

AI will likely be instrumental to future space travel...but it will be a tool wielded by competent scientists and engineers...if not, the astronauts are gonna be soooo f*cked,

 

[Kimbere]

Asked and answered. I've already defined AI as practical magic. The belief that DDO is "too complex" for AI to figure out when it's going to be instrumental in sending us to Mars is a relic of old-world thinking - "'ere be monsters!" on the map.

There is a big difference between "going to be", aka in the future, and "is", aka now.

I don't think anyone is arguing that AI won't eventually advance to the point where it can accurately do the types of things being discussed in this thread. I'm fairly sure it will at some point in the future.

The point being made here however is that AI is not advanced enough to do so accurately right now.

 

[Marshal_Lannes]

The point being made here however is that AI is not advanced enough to do so accurately right now.

And I have admitted as much, responding to Spartan. Someone has to be the pioneer. The Marshal leads the way!

 

[Kimbere]

To add another real-world example of AI's inaccuracies and relative immaturity as a technology to this discussion, consider this.

At work, we've been testing various AIs to see where they can save different teams time on repetitive, day-to-day tasks - i.e. the low hanging fruit that should be relatively easy for AI to do well.

Our BI team fed it a dataset from our ecommerce database and asked it to generate some reports with specific requirements and moderately complex calculations. When they sanity checked the reports by manually doing the same calculations, the AI results were way off. Worse, it gave different numbers each time they requested the same report.

After that, the BI team added a step to the process which asks the AI what its confidence level is on the results. It consistently rates its own confidence level in the 60-80% range.

They added an additional step asking the AI to explain why it's results are inaccurate since it is supposed to be use several strictly defined formulas and datasets to do the calculations. The AI's answers to that are hilarious and different every time it runs.

Long story short, it often just makes stuff up to fill in the blanks - aka it lies.

 

[DBZ]

Did the AI OVER LORDS say rare loot is good too

That does explain some things

 

[SpartanKiller13]

Our BI team fed it a dataset from our ecommerce database and asked it to generate some reports with specific requirements and moderately complex calculations. When they sanity checked the reports by manually doing the same calculations, the AI results were way off. Worse, it gave different numbers each time they requested the same report.

Long story short, it often just makes stuff up to fill in the blanks - aka it lies.

This has been my experience as well. I've seen AI give bad results for even simple math, like doing a whole formula then dividing by a thousand instead of multiplying by it - at a glance it looks reasonable since the right factors are in mostly the right places, but bam off by a million times.

Or in this thread where 72 attacks/minute was just scooted over into (Damage per Hit) x 72 = DPS, instantly off by a factor of 60. Attacks/minute is the usual input for most calcs in DDO (since it's testable) but the accurate DPS calcs all have a /60 to convert to attacks/second (which is not typically listed in DDO), but since that's not directly linked to the attacks/minute scraping a lot of data can easily result in being off by x60.

Many people do not sanity check their results (even a vague estimate helps a lot) and then promote it as truth since it came from AI.

This is my biggest issue with trying to use AI to solve the initial question in this thread, if DS or RP is better point per point.

For some more random reference, for a VKF thrower I have a spreadsheet for:

• Unboosted, each point of DS adds 0.313% DPS. Each point of RP adds 0.180% DPS.
• Boosted, each point of DS adds 0.208% DPS. Each point of RP adds 0.165% DPS.

For that build, each point of DS is 1.73x more valuable than RP unboosted, or 1.26x more valuable when boosting.

Per AI (Prompt: In DDO, estimate the DPS increase from adding +1 RP vs +1 DS to an endgame VKF thrower build) I got the following:
DS is +0.6%, RP is +0.20-0.25%, therefore each point of DS is ~3x as valuable as a point of RP. It's within the ballpark, but using AI here would cause you to weight DS > RP as roughly double the actual value (~3x instead of ~1.5x).

 

[vryxnr]

Blind trust is not pioneering anything. It is not leading anything either. Blind trust is not how to advance into the future.

Going the extreme in the other direction is no good either. Blindly rejecting anything new does not progress humanity into the future either. (omg not nuance, nuuuuuu, the bane of internet discourse, aaaaahhh!)

The key is to not do or believe things blindly. The key is making informed decisions, to become more educated and understand what is going on. Not jumping on the new shiney bandwagon to try and be amongst the earliest adopters and hoping that this time it's something that ends up with staying power; but instead investigating (which is more than just googling and reading the first headline, and more still than skipping to a summary as well, and is definitely more than using an AI summary to think for you), removing as much personal bias as possible, and actually understanding what is going on so you can make informed decisions and possibly even build on the thing to make it even better.

AI is not magic. It is a tool created by people that can be (and is) modified by, and manipulated by people.

 

[Mindos]

Grok, should this thread be closed?

 

[Lopnel]

Whenever I plan to play a ranged character I've always assumed that Double shot and Ranged Power were of equal value when working out the build. But I wanted to ask if this was actually the case. If they aren't of equal value, could you also explain why you think one is better than the other?

Having everything as best as it can with attack speed and ranged builds start to pack a punch. You will also want deadly, seeker and deception. It takes a lot of work to get a great gear set together so be sure to get involved in hard Raiding and build your currency of Threads of Fate.

 

[Marshal_Lannes]

Long story short

That is an Anecdotal fallacy and has no bearing on anything except what you witnessed. It is not reflective of any AI sample size. It's your n of 1 experience.

 

[Kimbere]

That is an Anecdotal fallacy and has no bearing on anything except what you witnessed. It is not reflective of any AI sample size. It's your n of 1 experience.

Of course it's anecdotal. What do you think real-world example means?

However, that was just one "example", as in there are more where that came from even at just the small "sample size" company I work for. We've had similar results testing different AIs.

There are many more examples of AI failures at many other companies but I assume nobody here is interested in reading 50 more examples of the same basic plot.

It's also not just my personal experiences. Google "AI slop", "AI hallucinations", "AI failures", or any other number of similar terms and your results list will be littered with more examples than you can count.

If you want to read some actual fact-based stuff with citations of actual studies done showing that AI tends to be unreliable, you could start here: https://spectrum.ieee.org/ai-failures.

I assume you've at least heard of the IEEE organization before? They're a bit more than an "anecdote".

As I've said, if you want to blindly trust AI results while knowing you lack the skills to verify the results, you do you boo. People believe in all sorts of things they can't/won't verify every day.

 

[Tsutti]

DMG = [1+ (DS/100)]x [(Imbue dmg x MRR scalar x (RP scalar for imbue- either 1.5 or 2)x(1+Rp/100) )+( sneak dmg x PRR scalar x 1.5 x (1+RP/100))+ (Weapon damage x PRR scalar x crit chance x multi x (1+Rp/100))+ (weapon dmg x PRR scalar x chance for non crit x (1+Rp/100))]

Just wanted to ask about that RP and sneak scalar on the ranged power--is that correct, having the scalar outside the (1+RP/100)? I assumed that "150% RP scaling" meant (1+1.5 x RP/100) but have never tried testing.

 

[Marshal_Lannes]

Of course it's anecdotal.

Then it has no factual basis on anything but your own experience. Your "AI slop" reference is almost always a comment made by people afraid of AI, trying to denigrate it and cling to or justify their place in a rapidly changing world. AI is far more functional than people realize. I am undetered by anyone's oh look, gotcha moments with AI. Gotcha journalism is a relic from an age when we agreed on facts. That time no longer exists. But, I accept, there will be many slogging away in the dark ages for a long time as AI takes over. The Marshal won't be one of those. You sound like you are. As you say, you do you, boo.

 

[Kimbere]

Then it has no factual basis on anything but your own experience. Your "AI slop" reference is almost always a comment made by people afraid of AI, trying to denigrate it and cling to or justify their place in a rapidly changing world. AI is far more functional than people realize. I am undetered by anyone's oh look, gotcha moments with AI. Gotcha journalism is a relic from an age when we agreed on facts. That time no longer exists. But, I accept, there will be many slogging away in the dark ages for a long time as AI takes over. The Marshal won't be one of those. You sound like you are. As you say, you do you, boo.

Says the guy who blindly believes AI's results even though he can't verify them.

Technically, it did have facts. We got results from AI. We checked them with known, good math and the AI results were wrong. It wasn't just my experience. It was the experience of multiple teams in multiple scenarios. Those are facts, even if you don't like them.

Secondly, you seem to have misunderstood. The post with my example which you quoted wasn't directed at nor was in response to you. I was participating in the discussion with others who are more fact and evidence oriented.

It's been obvious for a good while you're not open-minded about this. If math, facts, and logic were going to convince you, it would have happened many posts ago.

You've essentially reached the point of using "it's fake news" as a defense, so obviously you've exhausted all other options for ignoring math and facts. There's no point in continuing the discussion with you.

 

[Matuse]

Asked and answered. I've already defined AI as practical magic. The belief that DDO is "too complex" for AI to figure out when it's going to be instrumental in sending us to Mars is a relic of old-world thinking - "'ere be monsters!" on the map.

AI can't even figure out how to spell "Dungeons" when I asked it to create an image for Dungeons & Dragons,, and will only kill everyone who goes to Mars. There is no AI. It's a large language model. It cannot think. It cannot innovate. It cannot design. It does nothing new. It does nothing original. It regurgitates what it has been told in slightly different ways, and more often than not screws that up catastrophically.

 

[Buddha5440]

Today I learned people would prefer to use AI rather than think for themselves. AI is a useful tool, not the answer to everything. Using AI to justify an answer, while also admitting to not knowing how to do math, is a recipe for disaster. Lacking the skills to identify when AI is wrong will make for some spectacular outcomes down the road. Least commendable being a supremely subpar shuriken build, let alone actual real world consequences.

Imagine trusting AI to do your taxes and books, and not having the skills and oversight to verify its results. That's gonna be aces!

This past week saw the release of articles pointing out how software developers who use AI are not saving time nor increasing efficiency, and newer developers are suffering for lack of critical thinking skills and genuine programming knowledge and experience. Give the AI hype train enough time and we will see it derailed.

As for RP vs DS -> I'm usually trying to get as much RP as possible, then DS. The maths always came across as too arcane. The top players who share build theory with me tend to lean towards doubleshot over RP, usually b/c there are fewer sources of DS.

^This 100%. AI is a tool that is meant to decrease the time it takes to do things that you could do yourself. If you aren't able to do it yourself AI could tell you that 1+1=3 and you would just blindly accept it as fact.

 

[Buddha5440]

AI can't even figure out how to spell "Dungeons" when I asked it to create an image for Dungeons & Dragons,, and will only kill everyone who goes to Mars. There is no AI. It's a large language model. It cannot think. It cannot innovate. It cannot design. It does nothing new. It does nothing original. It regurgitates what it has been told in slightly different ways, and more often than not screws that up catastrophically.

It could be the smartest machine in the world but, if it's using invalid data and opinions... It reminds me of what my multi-variable calc prof told us... "You will never do this by hand outside this class but my job is to make sure you can."