AI Doomers and the End of Humanity

[bilby]

How about using AI for sports betting?

FiveThirtyEight often tries to model sports mathematically, and their predictions are very often wrong. There are way too many intangibles in sports.

I don't think they are intangible as much as probabilistic and inaccessible.

You can get your hands on every one of those factors, but doing that is akin to "getting together the energy to send a vessel at .99 of C: it's possible but not plausible.

The problem is that there just simply is not any way to model all that short of re-creating a whole universe just to look at that one part of it so you can know the outcome before it happens, assuming you don't significantly change anything on the lineup to that outcome.

There's no useful model for lover's spats in hotel rooms before games, or for when a player is going to go through an identity crisis on the bench, or for when fans will provide the energetic encouragement to actually sink that basket or hit that ball... or provides the distraction that prevents it.

There are too many inaccessible forward causes with too much influence fun the result for a sports game to be usefully predicted.

In all honesty, horse racing seems a better use of your time and even then as a normal bet placer you lack the information you would need to get the system to solve, things like "what did the jockey eat" and "what hotel are they sleeping in" and "what is their relationship with their girlfriend right now?" And having a model for how those things impact performance.

Then, maybe I'm also being a little extreme on my treatment of "intangibility", but I do take issue with the idea that anything is in fact truly intangible other than perhaps the laws of physics themselves.

Eh, it was just a thought. I don't know anything at all about sports betting.

The only relevant fact is that it's by definition unpredictable.

If there was any way to consistently beat the odds, the odds would be (and are) changed to make that impossible.

There's a thing called 'arbitrage', which leverages the differences between the odds offered (usually by different bookmakers) to find combinations of bets that are guaranteed to return more than was staked, but the bookies are rarely so significantly different from each other as to make this a viable way to raise money - even if you have the vast sums needed to place such bets with the expectation of a return measured in dollars rather than cents, you could still almost always get a better return by investing it elsewhere.

The thing about betting is that the house always wins. If they didn't, they wouldn't still be in business.

 

[Copernicus]

The thing about betting is that the house always wins. If they didn't, they wouldn't still be in business.

Well, that's what Donald Trump thought when he bought a bunch of casinos and overextended himself. Because he didn't understand how to run those kinds of businesses, he ended with six separate bankruptcies in the 1990s and managed to nearly wipe out the fortune he had inherited from his father. So he isn't still in the casino business, and I guess that sort of supports your point.

 

[Copernicus]

Emily Bender's AI Sentience/Consciousness Bingo Card:



One commenter responded to her by email with this, which she posted on her Facebook feed:

Dear Ms. Gebru, Ms. Bender and Ms. Mitchell,

I spent the past month or so carefully studying chatgpt4's capabilities, feeling I had no choice as the hype seemed to permeate every facet of life. In hindsight, I'm grateful that I did, because previously I had been misled by your paper.

I also just surfed Ms. Bender's twitter and saw her constant and intellectually dishonest attacks against chatgpt. For example, a "bingo board" of reasons why people think chatgpt4 thinks like a human. https://twitter.com/emilymbender/status/1536198662656626688
As a retired lawyer (I now design trading algorithms), I fully understand the need to use rhetoric and deception to distract and marginalize facts that oppose your case, even if those facts are truthful and even if your case is wrong. So I get Ms. Bender's tweet. But I also have the integrity to understand how harmful it is to use such rhetorical techniques outside of a courtroom. Your tweet is also incredibly stupid. Imagine me advocating that "the earth is flat" and tweeting "a bingo board of all the reasons why people say the earth is round" as a rhetorical tool. Yes there are dozens of responses to the claim that "the earth is flat" and one could make a bingo board of them for likes from flat earthers, but that doesn't disprove any of those responses.

I understand the academic game, the need to get citations and clicks. I also understand that being provocative, trolling, is a good way to get attention, and so get those two things. But it's wrong to make your way by deceiving the public about an important new technology. You are doing serious harm by delaying people's training in this new tech. You are materially hurting people in your pursuit of attention. This fundamental lack of integrity and morals in people like you is precisely why we need chatgpt4, so the public can get ready access to the intelligence required to quickly disprove people like you.

Any way. Below is a prompt that you can use to debate chatgpt4 on your ideas. It's a prompt I wrote to make chatgpt4 into a human, i.e. a thing that uses a combination of thoughts, statements and actions to achieve whatever goal it has at that moment. I can give it the goal of debating you. If you'd like, I can run the prompt and do this debate with you anywhere you would like, any time you would like, publicly or privately. I have little doubt it would wipe the floor with any of you.

Regards,
[Name & phone number redacted]