Watch: chess-playing robot grabs child opponent's finger and breaks it

midian182

Posts: 8,336   +103
Staff member
WTF?! In what could be the first warning of a machine uprising, a chess-playing robot grabbed and broke a seven-year-old child's finger during a tournament in Russia. This wasn't a case of an AI gaining sentience and throwing a violent tantrum, but an accident—at least that's what the robot's makers are claiming.

The incident took place on July 19 during the Moscow Chess Open that ran in the capital city from July 13 to July 21. The child, whose name has been reported as Christopher, was competing in the under-nines category and just had one of his pieces taken by the robot. According to Sergey Smagin, vice-president of the Russian Chess Federation, the boy responded before the robot had completed its move.

As you can see in the video below, the robot, which had already played three games that day, moved forward and grabbed Christopher's finger. It crushes the digit for several seconds before a woman rushes to his aid. Three men also tried to help, and the boy is eventually freed before being led away by what is presumably an event official.

Smagin seemed to put most of the blame on the victim. "There are certain safety rules and the child, apparently, violated them. When he made his move, he did not realize he first had to wait," he said. "This is an extremely rare case, the first I can recall."

Sergey Lazarev, president of the Moscow Chess Federation, said, "The robot broke the child's finger. This is of course bad," showing a flair for understatement. Lazarev noted that the robot had been playing chess for around 16 years and competed in many previous exhibitions without breaking competitors' bones or inflicting other acts of violence upon their person.

"The robot was rented by us, it has been exhibited in many places, for a long time, with specialists. Apparently, the operators overlooked it. The child made a move, and after that we need to give time for the robot to answer, but the boy hurried, the robot grabbed him. We have nothing to do with the robot," Lazarev added.

Christopher, said to be one of the thirty best chess players in Moscow in the under-nines category, had his fractured and scratched finger put in a cast. He showed a remarkable amount of grit for someone assaulted by a robot by playing the next day and completing the tournament. He was also able to attend the awards ceremony and sign documents.

Christopher's parents have reportedly contacted the public prosecutor's office, but the Moscow Chess Federation said it is having conversations with the couple and trying to help sort out the situation (I.e., convincing them not to press charges). "It [the robot] has performed at many opens. Apparently, children need to be warned. It happens," Smagin added.

It's not been the best weekend for artificial intelligences. Yesterday brought news that Blake Lemoine, the Google engineer who claimed its LaMDA (language model for dialogue applications) chatbot is sentient, had been fired from the company. He'd been on paid leave since publishing transcripts of his conversations with LaMDA, something Google said was a breach of its confidentiality policies.

Masthead: Don DeBold

Permalink to story.

 

m4a4

Posts: 3,105   +4,150
TechSpot Elite
Certainly looks like the kids fault for not following the safety rules (or maybe it is the organizers fault for not explaining those rules). As a programmer, I can't say I'm surprised it mistook his finger as a chess piece.

That said, you'd think there might be some id!ot proofing in the code or on the machine to avoid this (like a safety gate that opens/closes as it does it's move). But again, not surprising. A lot of coding is meant to be precise and within expected boundaries.
 

sreams

Posts: 359   +507
Certainly looks like the kids fault for not following the safety rules (or maybe it is the organizers fault for not explaining those rules). As a programmer, I can't say I'm surprised it mistook his finger as a chess piece.

That said, you'd think there might be some id!ot proofing in the code or on the machine to avoid this (like a safety gate that opens/closes as it does it's move). But again, not surprising. A lot of coding is meant to be precise and within expected boundaries.

I suppose I would ask: Why is it necessary for a robot that does nothing but pick up super-lightweight chess pieces to have the strength/torque to break a bone?
 

Entrylevel

Posts: 24   +19
Certainly looks like the kids fault for not following the safety rules (or maybe it is the organizers fault for not explaining those rules). As a programmer, I can't say I'm surprised it mistook his finger as a chess piece.

That said, you'd think there might be some id!ot proofing in the code or on the machine to avoid this (like a safety gate that opens/closes as it does it's move). But again, not surprising. A lot of coding is meant to be precise and within expected boundaries.

If a system is not tolerant of human error, then it is not a safe system.

And that tolerance should be tailored to the users it's exposed to: can we expect a trained, adult operator to stick to rules? Probably most of the time, so the risk might be acceptable. Can we [reasonably] expect the same of a child?
 

m4a4

Posts: 3,105   +4,150
TechSpot Elite
I suppose I would ask: Why is it necessary for a robot that does nothing but pick up super-lightweight chess pieces to have the strength/torque to break a bone?
Who said they're super light? I think it would be better to set it so it will work with heavier pieces (which, those would place better anyways).

They're also probably using a basic, reliable motor and mechanics for the fingers (which will be pretty powerful). And I definitely wouldn't expect any pressure sensors for it to know it picked up a piece (at least, not on a relic like this thing).
 

m4a4

Posts: 3,105   +4,150
TechSpot Elite
If a system is not tolerant of human error, then it is not a safe system.

And that tolerance should be tailored to the users it's exposed to: can we expect a trained, adult operator to stick to rules? Probably most of the time, so the risk might be acceptable. Can we [reasonably] expect the same of a child?
Then nothing is truly safe if you apply that "logic" far enough. But no one is claiming it was, or is meant to be at a certain safety level.

As stated, it's been in use for almost 16 years. Which means, that people have been following the safety rules up until this point.
Either the kid forgot, or wasn't told. The next version can be more human friendly (maybe shake hands at the start?) lol
 

Entrylevel

Posts: 24   +19
Then nothing is truly safe if you apply that "logic" far enough. But no one is claiming it was, or is meant to be at a certain safety level.

As stated, it's been in use for almost 16 years. Which means, that people have been following the safety rules up until this point.
Either the kid forgot, or wasn't told. The next version can be more human friendly (maybe shake hands at the start?) lol
I do apply that logic for a living ;)In industries like aviation, nuclear and oil & gas there are regulations enforcing it.

But no, I'm not really saying that random chess robots should be held to the same standard as these industries, only that we shouldn't blame the victim for something that could have been reasonably mitigated through system design.

Hopefully the lesson learnt from this incident is an improvement to the robot, not harsher application of the procedural mitigations.
 

eTheBlack

Posts: 32   +111
I suppose I would ask: Why is it necessary for a robot that does nothing but pick up super-lightweight chess pieces to have the strength/torque to break a bone?
You never worked with robotic arm did you? Its simply NOT possible to stop moving arm or whatever in instant it hits something. It will stop, but you will feel pain. I think in video robot do stopm but when it got info that something is wrong when it tried to pick it up, the clamps or whatever went to far in, thus breaking finger. Thats why most robots work in cages, so people dont put fingers where robot can reach it.
 

sreams

Posts: 359   +507
Who said they're super light? I think it would be better to set it so it will work with heavier pieces (which, those would place better anyways).

They're also probably using a basic, reliable motor and mechanics for the fingers (which will be pretty powerful). And I definitely wouldn't expect any pressure sensors for it to know it picked up a piece (at least, not on a relic like this thing).
No chess piece is heavy enough that it requires a level of force that could break a bone... unless, I suppose, you are using the set from Harry Potter.
 

m4a4

Posts: 3,105   +4,150
TechSpot Elite
I do apply that logic for a living ;)In industries like aviation, nuclear and oil & gas there are regulations enforcing it.

But no, I'm not really saying that random chess robots should be held to the same standard as these industries, only that we shouldn't blame the victim for something that could have been reasonably mitigated through system design.

Hopefully the lesson learnt from this incident is an improvement to the robot, not harsher application of the procedural mitigations.
Well, they'll either over-react (which, I don't see why they'd need to), or keep going (making sure kids and other people know not to interrupt it when it's moving).

I'm sure in the 16 years this has been working, they've also made more human friendly versions :p
 

m4a4

Posts: 3,105   +4,150
TechSpot Elite
No chess piece is heavy enough that it requires a level of force that could break a bone... unless, I suppose, you are using the set from Harry Potter.
Maybe they should make it so the pieces are feather lite. The gentlest force to move them.

Or maybe you want focus on something else other than this pointless tangent that has nothing to do with what happened. Because the weight of the piece doesn't chance the simplicity of the mechanism setup 16 years ago...
 

captaincranky

Posts: 19,184   +8,328
Much in the same way Russia invaded Ukraine, "for its own good". ("to deNazify it"), the robot probably was probably following a similar directive. After all, the child didn't follow directions, and had his deserved fate meted out.
 
Last edited:

texasrattler

Posts: 1,345   +644
More important is that why did it even react? Why would it even think or care to act like a human?
Maybe it just seems it acted a certain way and for some reason it mistook, misunderstood or simply didn't know what to do in that moment.

It all just seems strange to me. I bet the diagnostic reveals something that we will never hear of. I'd be interested if that can be reproduced or if it ever was. Stuff I doubt we will ever hear about.
 

Raytrace3D

Posts: 357   +433
I use to work for the worlds largest industrial robotics manufacturer and there were a ton of safety emphasis around robots and people. Having people around a robot is dangerous, but having children only increases the risk exponentially. Obviously, the kids injury was a freak accident but that's what safety is suppose to help to avoid - freak accidents. The robots we had in place for people who were to work in close proximity were specifically designed to immediately pause motion as soon as they encountered resistance in any axis outside of rated electrical values measured with the motors. In addition, grippers were replaces with soft, flexible components that would bend and all surfaces were padded with dense foam. Using a full robotic pinching gripper with kids around is stupid and injury would be inevitable. That kid's family is going to have a paycheck coming their way.
 

nismo91

Posts: 1,255   +304
It's just an accident. of course it doesn't help that the victim is 8 years old while the perpetrator is a 16 years old machine that was built before windows vista even came out.

I guess the lesson here is to always look after your children, whether in a theme park or in a goddamn chess tournament hall.
 

Fearghast

Posts: 585   +518
Will the robot be punished?
It will. They will call him "Bad robot, BAD!" and it will have to sit in the corner and face the wall for at least an hour.
Could be an issue to locate a face to face the wall in this case, but they will find the way.