What Does AI Say About The Billikens?

[slufan13]

Our Twitter trolls have introduced me to the beauties of AI video. So as part of a new weekly series, I have asked AI a few questions. YOu may need to view these videos on a laptop or in browser mode

First up: How would you describe the Travis Ford era at SLU? Here's how AI responded: invideo-ai-1080 The Rollercoaster Ride of Travis Ford's 2024-08-05 (2) (vimeo.com) 

 

Secondly: What are your predictions for year 1 of the Josh Schertz era? invideo-ai-1080 The Josh Schertz Era at Saint Louis Univ 2024-08-05 (vimeo.com)

And of course, what does AI think about the billiken baseball field? 

 

[CenHudDude]

Not always factual, but entertaining and comical.

[BrettJollyComedyHour]

They called Ford Stuen "Ford's son" lol

 

Also love the irony of showing women's basketball players in a locker room when talking about Sit. 2

They also called Ford's son a "former player"

Edited August 5 by BrettJollyComedyHour

[brianstl]

Let's just hope this AI is a better predictor of what is going to happen than Jim Cramer.

[slu72]

15 minutes ago, BrettJollyComedyHour said:

They called Ford Stuen "Ford's son" lol

 

Also love the irony of showing women's basketball players in a locker room when talking about Sit. 2

They also called Ford's son a "former player"

He was….at Mizzery. 

[billikenfan05]

 

[The Wiz]

And this is why it is called artificial intelligence. 

Recounting an earlier AI story I told on Bills basketball.

When AI doesn't know ...not to worry ...it will give you an answer anyway...When I have asked it questions on Bills basketball it frequently gave wrong answers. AI is good at reporting anything that actually happened(sometimes ). Who won the first Bills game last season and what was the score. AI is on it. But after the last Bills game of the season ...I asked AI who the Bills are recruiting in the portal...it came up with 3 wrong names.  Wrong because the 3 players had already signed elsewhere.  When I told AI they had signed elsewhere...not to worry...Without any other comments the machine came up with 3 "new" names. They too had already signed. I pointed out to the machine it was wrong and guessing. It said it was experimental and under development. I asked if it had any other suggestions...It said ...perhaps you could give me 3 names.  Where upon I told the computer ...if I you give you 3 names then why do I need you.  Its reply...because I am experimental and under development.

A key take away...this is college basketball...What happens when AI is deciding whether to launch a nuclear attack or where to cut on you for surgery.

[Spoon-Balls]

"The FOLKS over at Billikens.com all agree (except Billiken Roy and Hoosier Pal)" 

 

"Gibson JYME-erson"

[TheA_Bomb]

I'm disappointed this isn't about what Allen Iverson thinks. I bet he thinks we talk too much about practice.

[Coach314]

29 minutes ago, TheA_Bomb said:

I'm disappointed this isn't about what Allen Iverson thinks. I bet he thinks we talk too much about practice.

 

[ACE]

Just now, Coach314 said:

 

Ha! abomb not too good at comedy.

[Coach314]

4 hours ago, ACE said:

Ha! abomb not too good at comedy.

he tried 

[Old guy]

9 hours ago, The Wiz said:

And this is why it is called artificial intelligence. 

Recounting an earlier AI story I told on Bills basketball.

When AI doesn't know ...not to worry ...it will give you an answer anyway...When I have asked it questions on Bills basketball it frequently gave wrong answers. AI is good at reporting anything that actually happened(sometimes ). Who won the first Bills game last season and what was the score. AI is on it. But after the last Bills game of the season ...I asked AI who the Bills are recruiting in the portal...it came up with 3 wrong names.  Wrong because the 3 players had already signed elsewhere.  When I told AI they had signed elsewhere...not to worry...Without any other comments the machine came up with 3 "new" names. They too had already signed. I pointed out to the machine it was wrong and guessing. It said it was experimental and under development. I asked if it had any other suggestions...It said ...perhaps you could give me 3 names.  Where upon I told the computer ...if I you give you 3 names then why do I need you.  Its reply...because I am experimental and under development.

A key take away...this is college basketball...What happens when AI is deciding whether to launch a nuclear attack or where to cut on you for surgery.

I did AI for trading for about 8 to 10 years before I stopped. It is neither a time saver nor accurate on its own. What the AI does provide is an approximate solution to a question asked to it (usually entry and exit points). The accuracy or proximity to an accurate answer depends upon the number of iterations the AI runs. The more runs it does the more accurate it becomes but also the more fixed it becomes. If you let it run for as long as it goes it will provide a correct solution, but only for the exact variables of the data you feed it. This solution is useless for trading stocks because it is  rather fixed and the market constantly changes.

To make it adaptable to changes the number of iterations the AI is allowed to run must be limited to a relatively low number. You also must feed it data that contains a wide variation of market conditions and a wide range up and down values for the stock you want to trade. My preferred daily time interval series feed for the AI went from July 2008 to February 2010. From before the market crash 2008 to well into its recovery in 2010. Still with a short number of runs the accuracy of the AI left much to be desired. The way I fixed this issue was by accepting the fact that each run was nothing but an approximation to the desired solution (entry and exit points). The way to solve this issue was to produce a family of close to identical approximations. This was by done leaving all AI parameters used with the initial run as they were and changing the base number used by the AI's random number generator. I would produce 10 or so different runs all with small alterations to the random number generator clock. The 10 approximate results were then fed into a cloud analysis program to find an overall solution to the 10 separate runs. This was a vast improvement in the AI's accuracy but it took a lot of time to do. I finally decided to stop using AI for trading.

You must understand that the market has only 5 variables for each stock every day (open, close, high, low, and volume). You can probably use a similar approach in sports but the number of variables that must be used is much much larger that those for a single stock in the market.

An AI is NOT the automatic process that people believe produces excellent results. It all depends upon what you want the AI to do. If you want the AI to make money for you, you need to learn how to use it, and be able to alter its rules and parameters in order to achieve whatever results you want.

Just for fun, what answer do you get if you feed a general AI a collection of radiuses and circumferences? You get an approximation of Pi, of course.

 

[Dr. Holly Hills]

Little heavy on the title IX stuff if you ask me.