How much influence are you happy for AI to have?

[Tarqs]

Nobody is suggesting so. It's gradually replacing management decisions, and being pushed by the owners into the management. That's not debatable, or the point or the debate (certainly not mine). I'm questioning. And asking for opinions, on how far you'd let that process go.

Id like to know how far you think it's being used.

You're telling me that it has final rule on team and tactical selection over what the manager sees with his eyes?

 

[Tarqs]

Quite possibly so. I find it very alarming, but others won't and there's no doubt it's where the world's heading.

What's alarming about it?

 

[raul]

AI is a tool. Doesn't mean it's good or bad it's how you use it.

So far the evidence is we've accidentally drilled a hole in our hand with the Jefferson one but that doesn't mean the drill is bad.

Maybe him and the rest are the crash test dummies to evalute how useful or otherwise this change of direction is/isn’t?

 

[alcoblade]

What's alarming about it?

I'm personally uncomfortable with an AI model thrusting (enforced by non footballing roles) a player into a position that a footballing professional would have chosen someone else for.

You don't have to be alarmed by that.

 

[alcoblade]

Maybe him and the rest are the crash test dummies to evalute how useful or otherwise this change of direction is/isn’t?

Very true. The owners certainly aren't dumb and will be constantly re-evaluating the situation.

 

[Carlton Blade]

It depends on what relationships the ML model identified when it was trained - it’s very possible for it to find correlation where there’s no causation and that’s where having good data scientists and footballing domain knowledge are key when building these models.

I'm unable to believe that bias (concious in this case) didn't inform this decision from all parties - I would hope that their experience with medical data would have cured them of any conscious or unconscious bias but I'm sure the Cannon signing has given them an expensive wake up call to the cost of only going half in with data led decision making in football.

There was nothing in the way United played that led Cannon to be a logical decision. He's a horrible lone forward and does not suit a team who lack control of the ball. That would only get worse in a higher division so the 'one for the future' stuff is even more nonsensical

 

[10-0 win]

AI is a part,too much been put into it. Its the work done after AI detecting a possible target that really decides anything. Built on the Brighton model. AI found Goykoeres the management sold him off cheap to coventry

 

[Tarqs]

I'm personally uncomfortable with an AI model thrusting (enforced by non footballing roles) a player into a position that a footballing professional would have chosen someone else for.

You don't have to be alarmed by that.

Football, particularly recruitment has been going down the data and analytics model for years.

Again, this is no different.

Brighton pulled caicedo from indy del valle.

Was sold for 115m.

 

[alcoblade]

Football, particularly recruitment has been going down the data and analytics model for years.

Again, this is no different.

Brighton pulled caicedo from indy del valle.

Was sold for 115m.

No, understood. But it begs the question - if he initially appeared 'miles off' how much the team managemt would be been forced to persist?

 

[Tarqs]

No, understood. But it begs the question if he initially appeared 'miles off' how much the team managemt would be been forced to persist?

Same as any other u21 is miles off? Being peruvian makes very little difference.

 

[Snake]

It's just a natural progression in the way we as a species use tools and learned/shared processes to advance ourselves. That's why we are top of the food chain - we have a yarning to constantly innovate and push boundaries. Ok, so today it's AI and advanced data collection/analysis models. Years ago somebody said .... hey why don't we use this round thing, lets call it a 'wheel' to move our heavy objects. Then somebody said..... forget horses, let's use this steam stuff to get us from a to b. And it goes on and on and on.

In my very humble opinion, cos I know very little about AI, all it is is the latest methodology of doing what we have always done........... collect information on a player to help with the decision making process. In the good old days that used to be by observing and forming mental opinions, then putting that on paper to share with others. Then we advanced to scouting teams and emerging foreign markets. The natural progression here was then to have data companies to gather and store a myriad of metrics etc etc. And on to today, where we are using advanced algorithms and AI to further these developments.

Chaps, nowt has fundamentally changed! We've just advanced in the way we collect and analyse data. Nothing (atm) can replace a manager/mentor/coach putting his arm around a player and giving him that mental lift that leads to that extra 10% needed to win a game. Or that kick up the arse, when needed, to get somebody to think about their issues, given themselves a good shake, and get themselves back on track. That was, and still is in the hands of we mere mortals - we human beings! And that will always be the difference between real success and failure.

 

[Nick Jansky]

Nobody is suggesting so. It's gradually replacing management decisions, and being pushed by the owners onto the management. That's not debatable, or the point of the debate (certainly not mine). I'm questioning, and asking for opinions, on how far you'd let that process go.

The question is too vague. If you could define your parameters, maybe with some hypothetical examples, we would have a better chance of answering it.

 

[alcoblade]

Same as any other u21 is miles off? Being peruvian makes very little difference.

I'm not sure what relevance his nationality has?

He's miles off under 21 level, not the first team. Someone else isn't getting that place or prospect of development as a result. Again, time may prove that the model knows best.

 

[Loco Blade]

I am no AI expert but on the face of it I get it can filter out crap and risk . Pretty sure both Brighton and Brentford have mastered it signing players for less than a million and selling for £50m + . With PSR affecting everything income from player sales is key . We’ve hardly mastered that have we . Also getting the average age of the squad to below 24 also helps . My only concern is AI cant measure character and heart unless I am missing something. We need a more dynamic team full of energy minute 1 to 90 along with good tactics and flair . It’s the hope that kills you

 

[alcoblade]

The question is too vague. If you could define your parameters, maybe with some hypothetical examples, we would have a better chance of answering it.

A very specific - Jefferson Caccares would not be getting a place in the Under 21's without it.

 

[Tarqs]

I'm not sure what relevance his nationality has?

He's miles off under 21 level, not the first team. Someone else isn't getting that place or prospect of development as a result. Again, time may prove that the model knows best.

Why is he miles off the u21 level? Wasn't he playing regularly?

 

[alcoblade]

Why is he miles off the u21 level? Wasn't he playing regularly?

Yes. That's the point!

 

[Tarqs]

Yes. That's the point!

So is this you deciding he's miles off then and what are you basing this on?

 

[Carlton Blade]

A very specific - Jefferson Caccares would not be getting a place in the Under 21's without it.

In that situation, AI is filling the role of a DoF, identifying elite traits and taking a 'gamble' on a low cost player from a (perceived to be) undervalued market.

The biggest generator of value in football is player trading. The ideal woukd be to have a loan destination for these players, see them gain a bit of value and sell them at a profit. Then, when one really pops you bring them back to play here. It's insodious and corporate but it undeniably works if you have the capital to keep investing

 

[Nick Jansky]

A very specific - Jefferson Caccares would not be getting a place in the Under 21's without it.

Our very first signing. Experimental. Low risk. And obviously the system will continue to evolve. We can’t use that signing to judge what the eventual success will be, far too early.

And if a minor European team ends up paying £1.5m for him, he’ll have been a good signing.

I take your point about blocking the route of others into the U21s, but the vast majority of those players won’t make it no matter where they come from. The U21s only exist to provide a team to play in for the small number who will make it. Most are just there to make up the numbers.

 

[Tarqs]

Also, any playing in the u21s is blocking another player, irrespective of circumstances.

If he's blocking another player, chances are the other player isn't good enough.

 

[Tarqs]

Just to wrap this up from my perspective. If we're going to have any issues or worry with this seasons xfers, Cacares isn't the one that's flagged.

It's the non ai, human signing of Tom Cannon.

 

[alcoblade]

Our very first signing. Experimental. Low risk. And obviously the system will continue to evolve. We can’t use that signing to judge what the eventual success will be, far too early.

And if a minor European team ends up paying £1.5m for him, he’ll have been a good signing.

I take your point about blocking the route of others into the U21s, but the vast majority of those players won’t make it no matter where they come from. The U21s only exist to provide a team to play in for the small number who will make it. Most are just there to make up the numbers.

We definitely shouldn't judge based in this, for all the reasons you've suggested. I'm more questioning how much we trust the model to go beyond what the human brain is suggesting.

 

[alcoblade]

Just to wrap this up from my perspective. If we're going to have any issues or worry with this seasons xfers, Cacares isn't the one that's flagged.

It's the non ai, human signing of Tom Cannon.

Though not flagged up initially by, Tom Cannon (and our willingness to reach the lofty fee) was enormousoy influenced by AI. I'm 100 percent certain of that.

 

[Tarqs]

Ai -Jefferson 100k
Human - cannon 10m

Case study closed

 

[Nick Jansky]

He definitely shouldn't judge based in this, for all the reasons you've suggested. I'm more questioning how much we trust the model to go beyond what the human brain is suggesting.

I think it’s impossible to answer that question objectively until we begin to see outcomes from a more mature and evolved system, and that’s still a long way off.

 

[Tarqs]

Though not flagged up initially by, Tom Cannon (and our willingness to reach the lofty fee) was enormousoy influenced by AI. I'm 100 percent certain of that.

And I'm 100% certain he was persued before any mention of AI involvement at the club

 

[alcoblade]

If he's blocking another player, chances are the other player isn't good enough.

According to the model yes, not according to the football professionals. The argument should be about if that's right. Not if it's happening.

 

[bornablade]

This morning I'd suggest Jack Lester sits down at his computer and keys in : -

'How can I turn Brewster, Cannon and Moore into Mo Salah?'

 

[alcoblade]

And I'm 100% certain he was persued before any mention of AI involvement at the club

We were using data and AI long ago like most were. Rak Saki was signed in the back of it. Not because of Wilder.