[POLL] Using AI to predict outcomes

[beerlover]

A way to think about AI is to look at chess

will studying AI moves or playing against an AI help you become a chessmaster?
no it won't.

can it be used as a resource to help you improve and teach you new things, as well as study your mistakes and blunders? yes it can.

I think it is obvious that we are not going to end up with anything that would be "learnable" or whatever you want to call it from the AI itself because AI also learns from humans, so we teach AI what to do ourselves, then it does what we teach it to do. In that logic, that means if a human wants to learn from AI, they could very well learn from the people who gave AI all that information as well, and to be fair in the chess example, there are a lot of books and videos that can make you a great chess player. Doesn't mean you will be the best, but AI wouldn't make you the best neither.

So it is the same with any AI related thing, it will not help you get better more than what we already have, if you prefer AI as a tool to be better that's fine, but it wouldn't be suddenly a new thing, it already existed and just changed the wrap that's it.

[nullama]

~snip~
A way to think about AI is to look at chess

will studying AI moves or playing against an AI help you become a chessmaster?
no it won't.

can it be used as a resource to help you improve and teach you new things, as well as study your mistakes and blunders? yes it can.

Well, maybe not with chess because deep blue was not really that "smart".

But in Go, absolutely yes, and that has happened. AlphaGo surprised Go masters and they basically started to learn new moves based on what AlphaGo was doing.

It really is an alternative intelligence, and we humans can learn from it.

[Betwrong]

AI from what I've seen so far is mostly about enabling what is already possible but usually far more awkward.  The AI is a way for the plain masses of the population to access tools normally less available then it can provide via a search engine type conversational interface.      
  The other big point about AI that Ive heard a few times is you must remember the giant amount of data required for it to work.   Dont forget the computer isnt really super intelligent its mostly about resourcefulness to reference the data on tap it has; a giant reserve of data accumulated and formatted ready to use with the AI is the secret behind its magic.
   Show me the data for the outcomes with gambling and I will believe it can be of great help, for card games this is often maths data and long established and known so I can believe that possibility as AI.  I studied card games as part of a statistics and probability theory course a quarter century ago and it was old as hell then, its still relevant now to be in any AI  deployed solution no doubt.

AI can analyze a big amount of data in a second, the amount that would take weeks for a human to just read through.  So, I agree with you, AI can be helpful. It only needs to "know" what do you need. Tell it in a prompt, and you might get very useful results. I don't know how to use it in card games, but in sports betting AI can tell you whether to bet on a particular team with particular odds.

[KTChampions]

~   
  The other big point about AI that Ive heard a few times is you must remember the giant amount of data required for it to work.   Dont forget the computer isnt really super intelligent its mostly about resourcefulness to reference the data on tap it has; a giant reserve of data accumulated and formatted ready to use with the AI is the secret behind its magic.
~

It can be assumed that the next revolutionary leap in the development of AI will occur when AI begins to extract data itself and not be based on arrays of information prepared by a person. I'd like to see this, and I'm sure I will, since the field of AI is developing very quickly. I think it will look something like this: now bookmakers use 300 (I don’t know the exact number, just an example) different parameters to assess the probability of game outcomes, and AI will use 300 classic parameters + invent new ones that a person simply has not seen, let’s say another 200 parameters. And apparently the new model will be many times more effective.

[STT]

For sure the whole fluidity of performance and the biggest thing with AI is the self improvement, the way they instruct the AI to teach and refine its own process.     To some extent we do already have the search engines crawling for data, we've had this 20 years.  I remember how bad the late nineties interlinked websites were, you had to know the exact address by word of mouth almost.  So we are advanced down that road already, what is most impressive is progression AI has beyond human understanding.
    Last AI project I heard about is AI learning keystrokes not by sight but by sound.  They never taught it specifically only setup the learning to listen for distinction between each key, probability upto 90% success I think it did manage and part of that is the sequence of keys pressed so indirectly the AI was learning common language spellings.   Anyhow I mention this as I know someone for their degree studies had already conducted and performed this project however they did not use AI especially in this way; it was possible for humans to do this but AI was important in the refinement and to bring finesse to the project.
  AI allows the programmer to just start the process not needing to entirely construct and complete it all.   For sports betting this would be big, we all have the ability to do this manually but its alot of work to collate results; statistics when done properly is higher mathematics not especially simple or any small amount of work to work as a task.  In large data but also development of results the AI would be a great assistance in constant monitoring and analysis.

[nullama]

~snip~
It can be assumed that the next revolutionary leap in the development of AI will occur when AI begins to extract data itself and not be based on arrays of information prepared by a person. I'd like to see this, and I'm sure I will, since the field of AI is developing very quickly. I think it will look something like this: now bookmakers use 300 (I don’t know the exact number, just an example) different parameters to assess the probability of game outcomes, and AI will use 300 classic parameters + invent new ones that a person simply has not seen, let’s say another 200 parameters. And apparently the new model will be many times more effective.

Well, that would need that the AI is able to generate completely new information, which right now might or might not be able to do.

Yes, AI can create new things, but I think it's basically a combination of previously known things, not uniquely new.

It might be because of the training mechanism, we give positive feedback for things that look human made let's say, and negative to weird things.

We might be an obstacle to AI expressing a truly new creation.

[KTChampions]

~snip~
It can be assumed that the next revolutionary leap in the development of AI will occur when AI begins to extract data itself and not be based on arrays of information prepared by a person. I'd like to see this, and I'm sure I will, since the field of AI is developing very quickly. I think it will look something like this: now bookmakers use 300 (I don’t know the exact number, just an example) different parameters to assess the probability of game outcomes, and AI will use 300 classic parameters + invent new ones that a person simply has not seen, let’s say another 200 parameters. And apparently the new model will be many times more effective.

Well, that would need that the AI is able to generate completely new information, which right now might or might not be able to do.

Yes, AI can create new things, but I think it's basically a combination of previously known things, not uniquely new.

It might be because of the training mechanism, we give positive feedback for things that look human made let's say, and negative to weird things.

We might be an obstacle to AI expressing a truly new creation.

I do not think that this will be a big problem, because even now when AI processes (trains) on various images, it creates new information for itself (which we cannot understand). For example, AI analyzes well, and subsequently imitates the styles of different artists, but it is probably impossible to write down exactly what these styles are. Similarly, it can happen, for example, with a football match - the AI will watch it and will make live quotes regarding the outcome of this match (and they will be extremely accurate), but a person will not be able to see the rationale for these quotes.

[maydna]

~snip~

A way to think about AI is to look at chess

will studying AI moves or playing against an AI help you become a chessmaster?
no it won't.

can it be used as a resource to help you improve and teach you new things, as well as study your mistakes and blunders? yes it can.

You are right. But it's better to study chess until we are really good at it and try to fight the AI in playing chess. We may lose against AI, but it will be helpful for us to improve its skills. And if we can learn AI steps or moves, that could also be useful for us to win playing chess against AI.

And the only thing we can do is keep learning while practicing it. One day, we will have a chance to beat AI because AI is also human-made, which will be better than humans. And there must be gamblers who won't give in to AI and will keep trying to beat it.

[nullama]

~snip~
I do not think that this will be a big problem, because even now when AI processes (trains) on various images, it creates new information for itself (which we cannot understand). For example, AI analyzes well, and subsequently imitates the styles of different artists, but it is probably impossible to write down exactly what these styles are. Similarly, it can happen, for example, with a football match - the AI will watch it and will make live quotes regarding the outcome of this match (and they will be extremely accurate), but a person will not be able to see the rationale for these quotes.

Fair enough, that's basically the black box problem. That the AI might be able to predict things usefully but we don't know why.

Now, that could happen, but still we don't know if current AI has the predicting capability based on the current information.

Maybe, and maybe not. I don't think we know this yet.

[slapper]

~snip~

A way to think about AI is to look at chess

will studying AI moves or playing against an AI help you become a chessmaster?
no it won't.

can it be used as a resource to help you improve and teach you new things, as well as study your mistakes and blunders? yes it can.

You are right. But it's better to study chess until we are really good at it and try to fight the AI in playing chess. We may lose against AI, but it will be helpful for us to improve its skills. And if we can learn AI steps or moves, that could also be useful for us to win playing chess against AI.

And the only thing we can do is keep learning while practicing it. One day, we will have a chance to beat AI because AI is also human-made, which will be better than humans. And there must be gamblers who won't give in to AI and will keep trying to beat it.

Chess isn't about gambling. It involves planning, insight, and pure intelligence. Chess's millennia of intellectual legacy are trivialized by attempts to compare it to a game of chance. A foolish tactic is to try to outwit AI by learning its moves. AI is a machine that is constantly learning, not a predictable foe with set tendencies. Therefore, even though you may believe you understand its patterns, its tactics are constantly changing. Now, when discussing how humans are posing a challenge to AI, certainly, we made it. But does creating anything imply that we will always be able to improve upon it? I contend not. It's still admirable to pursue constant learning and improvement. Don't stop pushing the limits

[maydna]

~snip~

Chess isn't about gambling. It involves planning, insight, and pure intelligence. Chess's millennia of intellectual legacy are trivialized by attempts to compare it to a game of chance. A foolish tactic is to try to outwit AI by learning its moves. AI is a machine that is constantly learning, not a predictable foe with set tendencies. Therefore, even though you may believe you understand its patterns, its tactics are constantly changing. Now, when discussing how humans are posing a challenge to AI, certainly, we made it. But does creating anything imply that we will always be able to improve upon it? I contend not. It's still admirable to pursue constant learning and improvement. Don't stop pushing the limits

For this reason, we also have to continue learning apart from observing the movements of AI. Perhaps we can beat AI by continuously learning from various sources and developing it into the right strategy to beat AI. But it is not easy to beat AI because AI is a machine that continues to learn like us and never stops, so it is not easy for AI to predict what moves it will make. But at least, if we can learn too, it can improve our playing skills to have a higher chance of winning. And with that increased skill, we can use it against AI or other players.

[l3pox]

~snip~

A way to think about AI is to look at chess

will studying AI moves or playing against an AI help you become a chessmaster?
no it won't.

can it be used as a resource to help you improve and teach you new things, as well as study your mistakes and blunders? yes it can.

You are right. But it's better to study chess until we are really good at it and try to fight the AI in playing chess. We may lose against AI, but it will be helpful for us to improve its skills. And if we can learn AI steps or moves, that could also be useful for us to win playing chess against AI.

And the only thing we can do is keep learning while practicing it. One day, we will have a chance to beat AI because AI is also human-made, which will be better than humans. And there must be gamblers who won't give in to AI and will keep trying to beat it.

Chess isn't about gambling. It involves planning, insight, and pure intelligence. Chess's millennia of intellectual legacy are trivialized by attempts to compare it to a game of chance. A foolish tactic is to try to outwit AI by learning its moves. AI is a machine that is constantly learning, not a predictable foe with set tendencies. Therefore, even though you may believe you understand its patterns, its tactics are constantly changing. Now, when discussing how humans are posing a challenge to AI, certainly, we made it. But does creating anything imply that we will always be able to improve upon it? I contend not. It's still admirable to pursue constant learning and improvement. Don't stop pushing the limits

you are right, probably easier to make an AI for chess than for poker, that's why we hear about chess GMs unable to beat AIs but nothing like that in poker, do you think its possible to create an AI that can bluff?

that would be interesting to see

[nullama]

~snip~
you are right, probably easier to make an AI for chess than for poker, that's why we hear about chess GMs unable to beat AIs but nothing like that in poker, do you think its possible to create an AI that can bluff?

that would be interesting to see

That's actually quite an interesting idea.

An AI can clearly learn the basic rules of poker, in the same way that Deep Blue learnt all the chess moves basically.

So, in terms of strategy, the casino always plays the same. So, I wonder if there are some strategies that might be better in certain scenarios.

Card counting is one strategy that gives the gambler an edge.

But bluffing doesn't really help against the casino because they always play the same. Maybe when playing against other players it would work.

[KTChampions]

you are right, probably easier to make an AI for chess than for poker, that's why we hear about chess GMs unable to beat AIs but nothing like that in poker, do you think its possible to create an AI that can bluff?

that would be interesting to see

Is this so?

https://en.wikipedia.org/wiki/Pluribus_(poker_bot)

As I understand it, bots have been playing poker better than people for a long time and they have no problems with bluffing. A quick search of the name Pluribus yields many links about what professional players think of it and the fact that this bot bluffs in an unimaginable (and effective) way. And this is just one bot out of many, I’m sure there are commercial projects that keep their bots secret and simply make a profit (as long as there are live players on the market and not just bots haha).

[Betwrong]

you are right, probably easier to make an AI for chess than for poker, that's why we hear about chess GMs unable to beat AIs but nothing like that in poker, do you think its possible to create an AI that can bluff?

that would be interesting to see

Is this so?

https://en.wikipedia.org/wiki/Pluribus_(poker_bot)

As I understand it, bots have been playing poker better than people for a long time and they have no problems with bluffing. A quick search of the name Pluribus yields many links about what professional players think of it and the fact that this bot bluffs in an unimaginable (and effective) way. And this is just one bot out of many, I’m sure there are commercial projects that keep their bots secret and simply make a profit (as long as there are live players on the market and not just bots haha).

I'm not sure about this. We should know the costs. It can be not profitable to run AI bots for winning small amounts, and usual poker bots(not AI ones) are easily outplayed by above average players. And as far as I know AI bot needs about 20 hours to play with particular players to start winning against them, knowing their habits. Against a completely new player AI can't do much actually.

[l3pox]

~snip~
A way to think about AI is to look at chess

will studying AI moves or playing against an AI help you become a chessmaster?
no it won't.

can it be used as a resource to help you improve and teach you new things, as well as study your mistakes and blunders? yes it can.

Well, maybe not with chess because deep blue was not really that "smart".

But in Go, absolutely yes, and that has happened. AlphaGo surprised Go masters and they basically started to learn new moves based on what AlphaGo was doing.

It really is an alternative intelligence, and we humans can learn from it.

oh, that's interesting
didn't know about the go AIs, heard about it but just really little in the past

what we are seeing in art with AI is really cool too, many breakthroughs happening
we're hitting acceleration point now.

[nullama]

~snip~
oh, that's interesting
didn't know about the go AIs, heard about it but just really little in the past

what we are seeing in art with AI is really cool too, many breakthroughs happening
we're hitting acceleration point now.

Yes, I think we are starting to enter the phase where human minds and AI minds work together to create new things.

This will be an inflection point in history, similar to how the Internet changed everything.

[Hispo]

~snip~
oh, that's interesting
didn't know about the go AIs, heard about it but just really little in the past

what we are seeing in art with AI is really cool too, many breakthroughs happening
we're hitting acceleration point now.

Yes, I think we are starting to enter the phase where human minds and AI minds work together to create new things.

This will be an inflection point in history, similar to how the Internet changed everything.

Not sure if I would call this phase of the technological era to be an inflection point, because even though we can be pretty much amazed about what it was been reached with Artificial intelligence, there is still a long path forward before developers and even politicians who need to regulate the ethical questions behind AI.

Rather than being a tool for people to be more productive, I am afraid big companies will seek to replace much of their personnel by robots or machines powered with AI, in the case of casinos, they could try to replace costumer service with Artificial Intelligence in the Future, not for now though, since managing accounts where money is deposited requires human common sense.

[ChuckBuck]

Not sure if I would call this phase of the technological era to be an inflection point, because even though we can be pretty much amazed about what it was been reached with Artificial intelligence, there is still a long path forward before developers and even politicians who need to regulate the ethical questions behind AI.

Rather than being a tool for people to be more productive, I am afraid big companies will seek to replace much of their personnel by robots or machines powered with AI, in the case of casinos, they could try to replace costumer service with Artificial Intelligence in the Future, not for now though, since managing accounts where money is deposited requires human common sense.

I'm pretty convinced we're well into the stuff "inflection point". The inevitable marriage of AI and sectors like the casino industry is just around the corner. And as for responsible gambling? Sure, there's an argument about AI possibly taking the empathy and human touch out of the equation.

AI can actually be programmed to detect addictive behaviors in real-time and take immediate action. A human might miss it or choose to ignore it for profits, but an AI? It could be ruthlessly efficient in promoting responsible gambling. As for replacing customer service? Maybe you're right; the human touch is irreplaceable... for now. 

[nullama]

~snip~
I'm pretty convinced we're well into the stuff "inflection point". The inevitable marriage of AI and sectors like the casino industry is just around the corner. And as for responsible gambling? Sure, there's an argument about AI possibly taking the empathy and human touch out of the equation.

AI can actually be programmed to detect addictive behaviors in real-time and take immediate action. A human might miss it or choose to ignore it for profits, but an AI? It could be ruthlessly efficient in promoting responsible gambling. As for replacing customer service? Maybe you're right; the human touch is irreplaceable... for now. 

Customer service is probably already replaced with AI for most tasks really...

And the AI is usually nicer and knows their stuff much better than lots of humans anyway.