TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

How AI is changing gymnastics judging

111 点作者 mjwhansen超过 1 年前

18 条评论

t_mann超过 1 年前
It&#x27;s not just the <i>je ne sais quoi</i> performance aspect that I&#x27;d be concerned about, it&#x27;s something I&#x27;d call <i>Wii-Sports</i>-ification: the first time you try out movement-controlled games like they made the Wii console popular, you&#x27;ll likely put a lot of effort into the movements as if you were wielding a real tennis racket or a golf club. After a while, you realize that the motion sensor is much more coarse, and you do much more efficient, bastardized versions of the original movements. While I acknowledge that this system is supposed to be much more sensitive, I&#x27;d expect a similar effect to be happening once athletes get their hands on those machines in training: they&#x27;ll figure out how to best trigger the detection of each move, and how much flexing of the toes, bending in the back and overall effort is just enough for the machine, and it&#x27;s going to change the look and feel of the sport.
评论 #39131597 未加载
评论 #39128623 未加载
评论 #39128631 未加载
评论 #39131448 未加载
评论 #39131814 未加载
评论 #39132612 未加载
评论 #39132299 未加载
评论 #39131482 未加载
divan超过 1 年前
There is a deeper and largely overlooked problem with sports judging, especially in the sports with artistic component (like gymnastics or figure skating) – Goodhart&#x27;s effects. Goodhart&#x27;s effect stands for &quot;when the metric becomes a target it stops being a good metric&quot;. As soon as you make measurement part of the reward system, it changes the whole system.<p>Take figure skating as an example. Since 2006, ISU has introduced a similar judging system - IJS – that uses points for elements and an increasingly complex code of rules on how to assign those points. Each element has a so-called &quot;base value&quot; and judges determine GOE value in the range from -5 to +5 for each element. The computer averages that and takes trimmed mean to produce the final points for the element.<p>Those base values and points are logically assigned by the complexity of elements. More complex elements cost more. Great skaters can perform more complex elements. Sounds logical, right?<p>What has happened, though, is that using these points as a target has changed the way how skaters train. Why spend time learning &quot;easy&quot; elements that give little points? All time must be spent now learning &quot;complex&quot; elements that give more points. In a way, before, IJS skaters were chasing overall greatness; now, they chase points.<p>&quot;Skater&#x27;s greatness&quot; – whatever that means – is hard to define, let alone measure. The ability to perform complex jumps was a consequence of it, not a cause. Using points as a metric changes the whole system of training, and a new generation of skaters can perform insane quads (often because of pre-puberty body sizes, though), but no one would call them great skaters.<p>I&#x27;m not even starting with how these points affected artistry. &quot;Program components&quot; (PCS) that should measure artistry are practically a joke now. You basically can multiply jump scores by some constant and get PCS. [1] All skating programs look the same now and are very predictable because of points maximization.<p>Back to AI. The rules for figure skating become more and more complex each year in order to duct-tape the constantly emerging issues. Many judges admit that complexity is so high that the human brain can&#x27;t even possibly apply them in real time and has to resort to cognitive shortcuts. There are some early flirtations with using AI for judging figure skating [2], but nothing serious yet.<p>As outlined in the &quot;Categorized Variants of Goodhart&#x27;s Law&quot; [3], one of the major reasons of this effect in sports is not just the fact that elements != greatness, but the fact that there is no shared definition of what is a True Goal of this sport. What exactly judging system try to measure? What would constitute the best skater?<p>Where there is no explicitly stated True Goal, any proxy variable used as metrics (like &quot;ability to perform complex elements&quot;) will be far from good and will be distorting the system even further.<p>Eventually, this boils down to the structure and decision-making in the international federation that governs this sport.<p>And these things cannot be solved by AI. In a way, using AI is a yak-shaving for ever-increasing rules of an inherently wrong judging system.<p>[1] <a href="http:&#x2F;&#x2F;libjournals.unca.edu&#x2F;OJS&#x2F;index.php&#x2F;mas&#x2F;article&#x2F;view&#x2F;23" rel="nofollow">http:&#x2F;&#x2F;libjournals.unca.edu&#x2F;OJS&#x2F;index.php&#x2F;mas&#x2F;article&#x2F;view&#x2F;2...</a><p>[2] <a href="https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=pq_mv1eyZyA" rel="nofollow">https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=pq_mv1eyZyA</a><p>[3] <a href="https:&#x2F;&#x2F;arxiv.org&#x2F;abs&#x2F;1803.04585" rel="nofollow">https:&#x2F;&#x2F;arxiv.org&#x2F;abs&#x2F;1803.04585</a>
评论 #39128441 未加载
评论 #39133809 未加载
评论 #39128400 未加载
评论 #39132060 未加载
评论 #39128374 未加载
评论 #39154723 未加载
评论 #39135962 未加载
kbos87超过 1 年前
I feel like the obvious effect of this is that gymnastics becomes all about testing against and optimizing for an algorithm. That sounds sterile and lacking any of the romance (best word I can think of) that goes into physical competition. Then I remind myself that optimizing for the opinions of judges is the status quo. Judges that are susceptible to wild bias in all sorts of undetectable ways. This will change the sport for sure, but I can&#x27;t say that I think it will be all bad.
ahaferburg超过 1 年前
How accurate is this? There&#x27;s a video embedded in the article.<p><a href="https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=CinAYBZYANg" rel="nofollow">https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=CinAYBZYANg</a><p>To me it looks very shaky, e. g. at 0:20-0:25. It seems like they&#x27;re not using an underlying physical model that takes momentum into account. Also it looks like the system is interpreting the movement of the ponytail as head movement.
jerf超过 1 年前
This is a specific issue of a general problem of AI&#x2F;computer judging, where humans think they&#x27;re judging on specific criteria, and write it into the computer, without ever taking a moment to ask whether that&#x27;s actually what they want.<p>I use speeding in vehicles a lot. The point of a 55MPH speed limit isn&#x27;t really that cars should be going exactly 55MPH at all times. It&#x27;s an approximation of a complex situation that could handle only one number. But it&#x27;s not actually a law of physics. Rigidly handing out tickets for people doing 56MPH is missing the point of speed limit. (As are many of the &quot;You&#x27;re doing X over&quot; flashing signs, generally put in places that are precisely <i>where the speed limit is too low</i> and what people are doing is actually <i>fine</i>.)<p>Now that we finally have the technology to determine whether a gymnast was or was not one millimeter over some line, the question should be raised, is that what we <i>actually</i> want to judge on? Because just because we humans <i>thought</i> we were judging on that, just because we <i>thought</i> that&#x27;s what we wanted, is it <i>really</i>?<p>It is this examination step that has been missing in every rush to insert computers into some judging system. We thoughtlessly reify accidents of history and what used to be easy to judge. At least in the case of gymnastics we&#x27;re not going to send anyone to jail because of such thoughtlessness, but the sport could be destroyed. In the end, it still needs views, however abstracted the sports may seem to be from that, and after the novelty wears off, seeing humans out there competing for the most robotic and soulless performances runs the risk of gradually, but steadily, destroying all support for the sport.<p>Now, personally, I&#x27;m not sure this wouldn&#x27;t be a net gain for humanity in this particular place. But it&#x27;s certainly not what the gymastics community as a whole wants and they really need to slow down and apply some thought to what they <i>really</i> want before they let someone reify that into code.<p>They won&#x27;t. Nobody else is either, for things far more important than this. But they <i>should</i>.
评论 #39131986 未加载
评论 #39131547 未加载
kouru225超过 1 年前
I honestly feel like what AI is really doing is just clarifying what we’re actually doing. This whole article reminds me of my reaction when I first saw chatgpt write an essay.<p>I immediately recognized that this essay would’ve been accepted by my teacher when I was a kid, and then remembered just how useless I thought essays were back when I had to write them. It feels like, with AI, I had proof that these essays were just “busy work” rather than anything else. As a student, I was just taking the prompt that my teacher gave me and expanding it into an essay, and there was no information in the essay that wasn’t already in the prompt itself. Why is it useful to take something that could be said in a single sentence and say it in 5 pages? I wasn’t encouraged to think about the book I was writing the essay about because my thoughts were seen as secondary to doing the technical work and simply copying the style and format of the essay, but now, with AI, that part had been automated and all that education was for nothing.<p>It seems like Gymnastics really is, at it’s heart, a very technical act. Your goal is to achieve perfect normal angles for some reason. But as robots begin to dispel the haze around technical work, we stop being so interested in technicality. You can see it as an attack if you’re really invested in that technical stuff, but really you gotta ask yourself why you were so interested in it in the first place.
评论 #39140341 未加载
maxglute超过 1 年前
Computer vision determining if a ball touches the line, or which body part crosses the line first seems fine. Judging artistic merits not so much. If this like 3d capturing the entire routine and analytically evaluating difficulting based on some biomechanical formula them maybe. Otherwise I&#x27;d rather biased judging be limited to people &#x2F; geopolitics than whichever vendor writes the code. Or artistic elements need to weigh down &#x2F; eliminated.
penjelly超过 1 年前
this is fine with me, but it shouldnt replace the existing panel, instead augment the panelists with it. It could analyze the rotations and timings much more closely and replay&#x2F;study it at a more technical level.
vander_elst超过 1 年前
Anyone aware of there are proposals to use computer vision for judging the technical part and the judges for the artistic part? On paper that should help fulfilling both schools of thought
rasz超过 1 年前
&gt;It became official in 2017, when the FIG formally announced its collaboration with Fujitsu<p>Does FIG have powers of prosecution like UK Postal Office? Asking for a friend.
TimSchumann超过 1 年前
<a href="https:&#x2F;&#x2F;archive.md&#x2F;sJrYj" rel="nofollow">https:&#x2F;&#x2F;archive.md&#x2F;sJrYj</a>
world2vec超过 1 年前
&gt;(...) says FIG’s Steve Butcher, who previously served as sport director for the organization and is now the president of the element-recognition working group for Fujitsu, the company that developed the AI.&quot;<p>Oh boy, considering the Horizon debacle going on in the UK, I can expect a similar story in Olympic gymnastics in the near future...
评论 #39128104 未加载
Kalanos超过 1 年前
great! now the gymnasts won&#x27;t have to sleep w the judges, and the judges won&#x27;t be subject to political pressure
nhinck3超过 1 年前
I&#x27;d be more than a little worried that AI might start picking up on the wrong signals (like uniform or skin colour) to assign points.
ungreased0675超过 1 年前
“Gymnastics is a subjective sport, like diving or dressage, where factors like nationality, body type, the location of a judge’s chair, and vague concepts like “artistry” and “performance” affect scores.”<p>Even though it’s a subjectively judged event, the scores somehow have three decimal places!
评论 #39128125 未加载
评论 #39128789 未加载
评论 #39128103 未加载
jpnc超过 1 年前
&gt;JSS is not currently used to make determinations about artistry, and it still isn’t considered ready to use on certain complex routine components, like beam connections and dance elements on floor; the tech just isn’t there yet<p>So at what point does a sensor of some kind that does some computation become &#x27;AI&#x27;? Is this just marketing buzzword bandwagoning?
评论 #39128292 未加载
评论 #39128150 未加载
评论 #39128106 未加载
aurareturn超过 1 年前
Where AI sports judging needs to happen fast is refereeing at the NBA and NFL level. Referees ruin games. Fans are turned off and always skeptical. If an AI can ref a game better, it&#x27;d be instant success for the leagues - not to mention getting rid of the unions who protect poorly performing referees.<p>MLB is weird because everything needed to ref a baseball game does not need any advanced AI. MLB simply refuses to let robots ref games.
评论 #39129323 未加载
评论 #39133939 未加载
评论 #39133446 未加载
评论 #39128763 未加载
评论 #39128990 未加载
评论 #39129149 未加载
fnordpiglet超过 1 年前
In discussing AI they lament bias. In discussing human judges they laud the fact nationality and body shape and the judges seat position plays a role. So, when a machine is biased it’s a problem but when nationalism, racism, etc is rooted in a human judge it’s a feature? Technology really shouldn’t be written about by journalism majors.