The AI illusion in gambling

[Taskford]

AI use had become so prevalent in everyday use cases that people keep finding new things to replace with it without being very caring about the potential consequences. Many people have now replaced their sports betting routine with an LLM. The issue here is that people don't consider that any of the LLMs available to the general public isn't optimised for sports betting. They might be able to search the internet now and may sound very convincingly and what a sports punter would want τo hear but still, it's not a system that looks into real data about current matches.

This is mostly insider info or it's not quantifiable in a way that a public model would be able to access. So... Many people are having the illusion that they have assistance to their betting, when jn reality it's just some word salad. Have you and your friends considered that before using AI for your sports betting?

I think that many people believe that Ai can help them towards creating fast decision towards their bets. Since somehow that technology is helpful towards gaining quick knowledge and also for gathering data.

Also we can't avoid that there are some people got misled by thinking that Ai could help them to get passive income. Maybe this occur to newbies and they have this illusion on which they can potentially earn passive income by the help of this technology.

But for sure once they experience using that tool for long time, they figure out even f they use Ai this will not offer any guarantee to earn 100% win to them.

[Sandra_hakeem]

When it tells me that "I can leave the house without an umbrella", I try to understand how it came to this decision.... sometimes he told me this just because it did not rain on the same day in last year

That doesn't sound like a protocol that can, even in a decade, get even a quarter inch closer to human imperfection.
Things will never stay the same way they were, even the inflation rate gets higher every single year.

In the same way, it has already ruined a good code that I had created just because I asked he to make a new and simple implementation.
And with sports betting I have also done the opposite of what it advised me to do just because my intuition was telling me otherwise....

You're trying to tell me that something has totally been useless to you, but somehow, you figured that it can help you anyway? No, fuck your intuition. Just do what the bot says!

[lionheart78]

This is mostly insider info or it's not quantifiable in a way that a public model would be able to access. So... Many people are having the illusion that they have assistance to their betting, when jn reality it's just some word salad.

Using Ai in sports betting is not just some word salad.  Ai is able to gather information, competing team data, the strengths and weaknesses of each team, and the possible outcome if given a scenario or simulating a possible outcome in given different situation.

I can consider it a good assitance in determining the possible outcome of the game, but obvously it is not 100% accurate due to possible changes of statistics in the ingame match.  There is also this variable that is undetectable by any research until it happens in the match.

Have you and your friends considered that before using AI for your sports betting?

Why not?  I believe there are also some members in this forum who tried to rely on AI in some of their bets, and if I am not mistaken won some.

[alegotardo]

....

Simple... either way, an AI will never make a decision for me.

[Darker45]

....

Simple... either way, an AI will never make a decision for me.

LLMs are useful to a certain extent if you want to avoid reading one sports analysis after another just to have an idea whether the experts agree with you or not, or to confirm whether your pick is solid or not.

Rather than manually researching what individual sports analysts and commentators have to say about an upcoming match, you could just ask an LLM for a nice summary. That's quite handy and efficient. You're just gathering second opinions, anyway. Your own prediction is the first, of course.

But you don't just ask an AI and bet on its answer. There's no fun in that. That's throwing money away.

[tread93]

AI use had become so prevalent in everyday use cases that people keep finding new things to replace with it without being very caring about the potential consequences. Many people have now replaced their sports betting routine with an LLM. The issue here is that people don't consider that any of the LLMs available to the general public isn't optimised for sports betting. They might be able to search the internet now and may sound very convincingly and what a sports punter would want τo hear but still, it's not a system that looks into real data about current matches.

This is mostly insider info or it's not quantifiable in a way that a public model would be able to access. So... Many people are having the illusion that they have assistance to their betting, when jn reality it's just some word salad. Have you and your friends considered that before using AI for your sports betting?

Yes a word salad too that runs on hype and many people are probably getting majorly scammed by hackers who make something look great and then hack the heck out of them and steal all of their coins

[m2017]

Many people have now replaced their sports betting routine with an LLM. The issue here is that people don't consider that any of the LLMs available to the general public isn't optimised for sports betting. They might be able to search the internet now and may sound very convincingly and what a sports punter would want τo hear but still, it's not a system that looks into real data about current matches.

It’s their money and their risk. How is AI any worse than, say, match information from a search engine, or rumors and arguments picked up from an acquaintance? In my view, even if you went to a fortune teller for a betting prediction, it wouldn't make a fundamental difference, because betting always involves randomness. The outcome isn't predetermined (unless it's a fixed match), and no matter what "tools" you use to try and ensure accuracy, the result remains unpredictable.

This is mostly insider info or it's not quantifiable in a way that a public model would be able to access. So... Many people are having the illusion that they have assistance to their betting, when jn reality it's just some word salad. Have you and your friends considered that before using AI for your sports betting?

Any analytical data used for betting is just as much of an illusion as AI-generated results. Ultimately, it all comes down to randomness. Regardless of whether you have a four-leaf clover in your pocket, a coin you toss will land on tails exactly as often as statistics dictate - not because of the clover.

[Tungbulu]

The level of trust people, especially bettors place on AI models should be really moderated, people really need to be careful about this, because I’ve seen people who do much depend on these AI models and end up exaggerating its capabilities. Yes, AI may be a great tool in processing statistics and also in providing useful insights and analysis, but people should never make the mistake of considering what appears to be convincing and logical to be a real edge.

[jcojci]

I realized that thing before using AI although AI help collecting the data without too long. Maybe AI gives the option to pick but we can't let AI decide and pick the choice. We should pick by ourselves by analyzing more the data to finds which could be the right choice. But I realize that some people rely on AI and let them follow through AI to pick their option. We are free to use AI and how we will pick the option but we must realize the risk behind that.

[viljy]

I’ve used regular AI for betting several times. The results were quite mediocre. The AI simply aggregates information from betting websites, including publicly available forecasts, and produces the result of its “analysis,” which is no different from what’s written on betting websites. To make predictions, you need a specialized model that isn’t publicly available.

[abaeze]

AI use had become so prevalent in everyday use cases that people keep finding new things to replace with it without being very caring about the potential consequences. Many people have now replaced their sports betting routine with an LLM. The issue here is that people don't consider that any of the LLMs available to the general public isn't optimised for sports betting. They might be able to search the internet now and may sound very convincingly and what a sports punter would want τo hear but still, it's not a system that looks into real data about current matches.

This is mostly insider info or it's not quantifiable in a way that a public model would be able to access. So... Many people are having the illusion that they have assistance to their betting, when jn reality it's just some word salad. Have you and your friends considered that before using AI for your sports betting?

We have been very excited and enthusiastic about the positive aspects of AI and its advantages, but most people have not understood its terrible aspects. It is time to stop its excessive use. Currently, it is causing more harm than good to the common man, especially in the crypto sector, betting and hacking, AI has made it very easy. For experiment in the case of prediction, I have tried it a couple of times but its results have disappointed me every time. So it can be used for data collection, but depending on it, it is a kind of madness.

[Iranus]

So... Many people are having the illusion that they have assistance to their betting, when jn reality it's just some word salad. Have you and your friends considered that before using AI for your sports betting?

Over reliance on AI can sometime lead to bad outcome. If AI gives you 80% confidence in any match, but you don't have any audited record of whether that 80% has historically hit 80%. Then that 80% number has no betting value. So along with such types of AI's confidence, if your overconfidence is also be added. On top of that if you don't have any track record, then your results won't be very good.

So I would say that AI is in no way replacing sport betting. If used well, it can be a tool in bettor's workflow. But AI itself is not the edge. Where the edge lies need to be proven measurably against market odds.

[Betwrong]

AI use had become so prevalent in everyday use cases that people keep finding new things to replace with it without being very caring about the potential consequences. Many people have now replaced their sports betting routine with an LLM. The issue here is that people don't consider that any of the LLMs available to the general public isn't optimised for sports betting. ~

I think this might create another illusion, like, if AI was optimized for sports betting, people using it could be in profit in the long run. Then various scammers would start popping up like mushrooms, selling access to LLMs  "optimized for sports betting". What I'm trying to say is that it's impossible even in theory to create an AI that could make correct predictions all the time. And that is because we are not talking about some particles here moving in this or that direction, we are talking about real people, and real people are unpredictable.

[Nrcewker]

AI use had become so prevalent in everyday use cases that people keep finding new things to replace with it without being very caring about the potential consequences. Many people have now replaced their sports betting routine with an LLM. The issue here is that people don't consider that any of the LLMs available to the general public isn't optimised for sports betting. They might be able to search the internet now and may sound very convincingly and what a sports punter would want τo hear but still, it's not a system that looks into real data about current matches.

If you dont want to use or trust publicly available LLMs then why cant you develop your own supervised machine learning model? You could train your model using all the historical data or create an automation to retrain it every day and then test the model using the current match information.

Anyone can easily build a model like this by following YouTube tutorials and having basic Python knowledge. I could have done it myself as well but what I think is that I am not ready to trust AI with my money yet. AI is good for analysis, but I still really want to double check everything because my hard earned money is at risk.

In the current situation, I dont think many people will rely on LLMs or AI for this. But yes, after 4 to5 years, I believe this will be offered by casinos themselves to make gamblers lives easier.

[free-bit.co.in]

AI use had become so prevalent in everyday use cases that people keep finding new things to replace with it without being very caring about the potential consequences. Many people have now replaced their sports betting routine with an LLM. The issue here is that people don't consider that any of the LLMs available to the general public isn't optimised for sports betting. They might be able to search the internet now and may sound very convincingly and what a sports punter would want τo hear but still, it's not a system that looks into real data about current matches.

I agree with you. But I wouldn't rule out AI completely. I can use it as research assistant. For example, organizing data, cross-checking injury/news sources, challenging my hypothesis or using AI to find out the assumptions of a model can easily do the work for me.

But when it comes to making final bet, I definitely won't take LLM's prediction as final proof. Before making any final decision, I will look at historical out-of-sample performance, calibration, closing-line comparison, actual ROI/CLV etc. Because I know that even sophisticated ML models fail to make consistent profit against bookmaker odds.

[swogerino]

AI use had become so prevalent in everyday use cases that people keep finding new things to replace with it without being very caring about the potential consequences. Many people have now replaced their sports betting routine with an LLM. The issue here is that people don't consider that any of the LLMs available to the general public isn't optimised for sports betting. They might be able to search the internet now and may sound very convincingly and what a sports punter would want τo hear but still, it's not a system that looks into real data about current matches.

This is mostly insider info or it's not quantifiable in a way that a public model would be able to access. So... Many people are having the illusion that they have assistance to their betting, when jn reality it's just some word salad. Have you and your friends considered that before using AI for your sports betting?

I have used AI lately extensively, cheap free models for the general public like ChatGPT non premium and it has behaved well, it really depends on what prompt do you feed it, the better the prompt the better the response that you will get. Most people recklessly rush without giving much thought to the prompt and then they are surprised by the results. AI is just a very good virtual assistant to everything, we just have to harness its true power.

[Bryan jessy]

I have always considered AI as a supportive tool for research i make use of AI a lot but not for gambling but for other things that I want to structure very well, or looking for an additional knowledge, I have never once thought of AI as something that can help in gambling games because I already understand that gambling is a game of luck so using AI will not make it better, if it's a winning game the bettor is definitely going to win and if it is a lost game the bettor will still lose the game at the end, except there are other things that AI help with in gambling games aside that it can not predict a winning game and this is what everyone will like to make use of AI for.

[Ayers]

So... Many people are having the illusion that they have assistance to their betting, when jn reality it's just some word salad.

It would not be right to call LLM a useless word salad for sports betting. Raw LLM and properly designed betting model are not same thing. You have to understand that odds themselves are very powerful public forecast. And just predicting match winner correctly is not enough for profitable betting.

Calibration is especially important.

Read this report by University of Bath and you will understand that selecting a model with calibration instead of accuracy gives much better betting return. That is, AI said Team A will win, is too little information for me. I will need how much probability the AI ​​is giving and what the bookmaker's odds are saying. If you can compare these and find out the implied probability and make your decision, you can profit well in long run.

[TopTort777]

I agree with you. But I wouldn't rule out AI completely. I can use it as research assistant. For example, organizing data, cross-checking injury/news sources, challenging my hypothesis or using AI to find out the assumptions of a model can easily do the work for me.

That research assistant is useful only if you ask direct questions and ask for direct info. Then it might give you info about news or injuries. If you do not ask it or ask for something specific, then AI will give you a plain and obvious answer based on odds. Kinda: favorite is more likely to win, underdog is likely to lose. AI still is more an instrument to structure info you give or ask, than assistant who will search info for you.

[Kelward]

Been there, done that.

Where people get this wrong is that they trust whatever output that comes from the AI or LLM. I use AI a lot and I think where it excel is research and helping to build things that would otherwise be impossible to build without extensive coding experience. Any gambling system will fail if it relies on AI for 100% of its decisions; plain and simple.

You're correct AI or LLM are not specifically designed for prediction games like gambling, they are basically to gather stats for research and helping in solving complex problems very fast that would stress the human brains, AI is effective in many areas of research and analysis but not very effective in gambling. I've never relied on LLM for my sports predictions, I prefer the traditional way of doing my own research as much as I can and bet with amount that I can easily give away. If the rate of AI aided predictions were always correct I'm sure that most of us would switch to it but winnings in gambling still depends on luck no matter the high intelligence that we rely on for winnings.