Only last weekend another poor scorecard came against Devin Haney in a world championship event as well no less.
Haney had a solid win on his welterweight debut boxing against Brian Norman Jr to win the WBO welterweight championship.
He dropped Norman in the fight and won the fight on points handily enough but one judge had it crazy close when in fact Haney won the fight widely.
When it comes to professional boxing or any sport where human judging is involved, there will always be occasionally the chance of human error on occasion.
With that being said, it shouldn’t happen as frequently as it does in top flight professional boxing, really though, it shouldn’t.
The Ring have already trialed Ai scoring in boxing recently with some success already to their credit at some events.
What Can Be Done With AI Judging In Boxing
But not perfect either.
If the AI can improved to be totally error free why not roll it out for world title fights next year after more extensive testing?
Important things to note for it though, because again, it is a machine so you have to think like a machine to program it correctly:
- Effective scoring punches landed
- Power shots landed
- Punches thrown
The first two are by far the most important for it to factor in to its tech and sensors, the third not really important but CompuBox and others still rate punches thrown for aggression and work ethic in a round.
Really though, it should be looking at first the punches landed that are scoring shots, whether power punches, jabs, hooks, uppercuts or knockdowns in the round obviously creating an extra point.
The human side of judging can factor in things like ring generalship as well but we don’t really think this is the priority in the unified rules of boxing and as well from a machine point of view, best to stick with effective scoring punches landed (power punches and standard shots).
AI doesn’t always take into account defensive boxing or punches blocked or slipped or parried either.
But effective punches landed and scored, still, at the end of the day, is most important criteria.
Iterate the thing like mad behind the scenes, run maybe 1000 test fights to make sure that the success rate is over 90 per cent at least before letting it go anywhere near a live professional fight to replace a human judge, first.
