June 5 – This week FIFA unveiled a new tech tool box of AI driven ‘improvements’ for match officials and analysts. All to be implemented at the World Cup kicking off next week. Most noteworthy of the AI incursion into the people’s game is the latest version of semi-automated offside technology (SAOT).
SAOT combines player tracking cameras, AI processing, and data from the match ball to identify potential offside situations in real time. No referees, assistants or VAR folk in a room offsite needed in the decision-making.
But before we hand football over to the machines, we asked AI to have a look at whether it thinks AI is any good or not.
Question to ChatGP: Is AI up to the job for the new FIFA World Cup offside rule using VAR and AI?
ChatGP’s answer:
“AI is actually doing a lot of the work in FIFA’s semi-automated offside technology (SAOT), but many people feel it is still “not up to the job” because the hardest part of offside decisions is not the geometry – it’s the interpretation of the law.
What AI does well;
Modern World Cup and top-level VAR systems use AI and computer vision to:
- Track the ball in 3D.
- Track up to 29 body points on every player.
- Detect player positions multiple times per second.
- Automatically suggest an offside line.
- Alert VAR officials when an offside is likely.
The technology can determine player positions far more accurately and quickly than humans.
Where AI struggles
1. Determining the exact moment the ball is played
Offside is judged at the instant the ball is touched by a teammate.
A football isn’t a rigid object:
- It deforms when kicked.
- Contact can last several milliseconds.
- Different camera frame rates can produce slightly different results.
Even a few milliseconds can change whether a player appears offside.
2. The law contains subjective elements
The biggest challenge is not player location.
The law asks:
- Was the player interfering with play?
- Did they obstruct an opponent?
- Did they gain an advantage?
- Did they deliberately play the ball?
These are judgment calls.
AI can identify where players are, but it cannot reliably interpret football intent and context the way referees are expected to.
3. Margin-of-error issues
Fans see offside decisions based on:
- A shoulder.
- A toe.
- A knee.
When the margin is only a few centimeters, people question whether:
- Camera calibration is perfect.
- Body-point tracking is exact.
- The frame selected is truly correct.
Even if the system is technically accurate, it may not appear fair.
4. Football is chaotic
Players:
- Overlap each other.
- Hide body parts.
- Move at high speed.
- Change direction suddenly.
AI models occasionally lose tracking or require human verification.
The new FIFA approach
FIFA has been pushing toward more automation because:
- Decisions are faster.
- Human drawing of lines is reduced.
- Consistency improves.
However, FIFA still keeps a human VAR official in the loop because the system cannot yet handle all the interpretive aspects of the law.
Why many fans think AI isn’t ready
Many supporters expected AI to remove controversy. Instead, it has shifted the debate:
Old debate: “The referee got it wrong.”
New debate: “The technology measured it correctly, but should a player really be offside by 2 cm?”
That is less a technology problem and more a question about how the offside law itself is written.
In other words, AI is already very good at measuring offside positions. The remaining controversies usually come from the rules, interpretation, and the expectation that technology can eliminate every gray area in football – which it probably cannot.”
Our opinion of ChatGP’s response
We don’t think AI is being very honest here!!
- It says … Fans see offside decisions based on:
- A shoulder
- A toe.
- A knee.
It doesn’t state AI can see or do this too.
2. It says “Football is chaotic” because
“Players …
- Overlap each other.
- Hide body parts.
- Move at high speed.
- Change direction suddenly.”
This is not chaos, this is the beauty of a complex game – it’s complexity not chaos.
Just because you can’t deal with it doesn’t make it wrong or chaotic.
3. AI admits … “AI models occasionally lose tracking or require human verification.”
Well video replays don’t do this very often.
4 ChatGPT says it …
- “Automatically suggest an offside line.”
Key word is suggests – and really implies it not the actual offside line but just the link that AI thinks it should be.
5. AI admits …
“A football isn’t a rigid object:
- It deforms when kicked.
- Contact can last several milliseconds.
- Different camera frame rates can produce slightly different results.”
So at first attempt we didn’t really get an answer as to whether AI thinks it is fit for World Cup purpose or not?
We ran all the above through ChatGP again and asked it to come up with a conclusion and a headline. This is what we got back:
“Why AI Still Isn’t Ready to Referee Football’s Most Human Rule
For years we’ve been told that Artificial Intelligence is smarter than us. It can beat chess grandmasters, write poetry, generate artwork, and apparently tell us what kind of toaster we should buy.
Yet somehow, when it comes to deciding whether a striker’s kneecap was offside during a World Cup match, AI suddenly becomes a nervous intern asking for adult supervision.
According to supporters of FIFA’s Semi-Automated Offside Technology (SAOT), the system can track the ball in 3D, monitor up to 29 points on every player’s body, draw virtual lines and alert officials to potential offsides.
Impressive…
The Verdict
Supporters were promised that AI would eliminate controversy. Instead, it has merely upgraded the argument.The old debate was: “The referee got it wrong.” The new debate is: “The computer got it right, but the answer still feels wrong.”
That’s not necessarily a failure of technology. It’s a reminder that football isn’t played on spreadsheets. It’s played by humans.
And despite all the hype, the most difficult part of football remains the one thing AI still struggles to understand football itself.” ChapGPT
And probably FIFA doesn’t understand AI, and still no-one understands the offside rule
Contact the content creator of this piece at… well you can’t. For a human, contact [email protected]