Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 



Forgot your password?
typodupeerror
×
AI Education Technology

These Students Figured Out Their Tests Were Graded by AI -- and the Easy Way To Cheat (theverge.com) 216

Monica Chin, reporting for The Verge: On Monday, Dana Simmons came downstairs to find her 12-year-old son, Lazare, in tears. He'd completed the first assignment for his seventh-grade history class on Edgenuity, an online platform for virtual learning. He'd received a 50 out of 100. That wasn't on a practice test -- it was his real grade. "He was like, I'm gonna have to get a 100 on all the rest of this to make up for this," said Simmons in a phone interview with The Verge. "He was totally dejected." At first, Simmons tried to console her son. "I was like well, you know, some teachers grade really harshly at the beginning," said Simmons, who is a history professor herself. Then, Lazare clarified that he'd received his grade less than a second after submitting his answers. A teacher couldn't have read his response in that time, Simmons knew -- her son was being graded by an algorithm. Simmons watched Lazare complete more assignments. She looked at the correct answers, which Edgenuity revealed at the end. She surmised that Edgenuity's AI was scanning for specific keywords that it expected to see in students' answers. And she decided to game it.

Now, for every short-answer question, Lazare writes two long sentences followed by a disjointed list of keywords -- anything that seems relevant to the question. "The questions are things like... 'What was the advantage of Constantinople's location for the power of the Byzantine empire,'" Simmons says. "So you go through, okay, what are the possible keywords that are associated with this? Wealth, caravan, ship, India, China, Middle East, he just threw all of those words in." "I wanted to game it because I felt like it was an easy way to get a good grade," Lazare told The Verge. He usually digs the keywords out of the article or video the question is based on. Apparently, that "word salad" is enough to get a perfect grade on any short-answer question in an Edgenuity test. Edgenuity didn't respond to repeated requests for comment.

This discussion has been archived. No new comments can be posted.

These Students Figured Out Their Tests Were Graded by AI -- and the Easy Way To Cheat

Comments Filter:
  • by gwills ( 3593013 ) on Thursday September 03, 2020 @12:35PM (#60469844)
    I'm very sympathetic to teachers in general, but you're supposed to be teaching children how to WRITE and make sense to a person - not to be evaluated by a VERY NARROW algorithm. Fire the teacher for negligence. This is the teachers version of cheating.
    • by Anonymous Coward on Thursday September 03, 2020 @12:40PM (#60469866)

      fire the school board administrators who decided this was the solution and probably shoved it down the teachers throats with little to no warning

    • And the kid deserves extra credit for figuring out that most of K-12 is just regurgitated fact seduction that is supposed to teach you how to study rather than actually teach you any information.

      And given today's "distance learning" a quick google in a different window on a different screen can easily get you a list of keywords associated with a topic.

    • by Archtech ( 159117 ) on Thursday September 03, 2020 @12:48PM (#60469910)

      It wasn't any specific teacher's decision to use Edgenuity. That kind of decision is taken at the level of school management at least - if not a school district or higher. Using "AI" to mark tests is exactly the sort of ignorant stupidity one would expect of politicians and managers, not teachers.

      From the article:

      "As COVID-19 has driven schools around the US to move teaching to online or hybrid models, many are outsourcing some instruction and grading to virtual education platforms".

      There have been excellent computer-based education suites ever since the 1950s - PLATO was a brilliant early example. I was greatly impressed by it when I saw it during a course with Control Data in 1971. https://en.wikipedia.org/wiki/... [wikipedia.org]

      But they have distinct limitations. The PLATO examples I saw were to do with maths and science; so most questions had definite right and wrong answers. The beautiful part of how PLATO worked was that trained teachers wrote the scripts, and included diagnosis of any mistakes. If a student got the solution of an equation wrong, he would be taken back to the relevant section and taken through it again more slowly, with more steps and examples to make sure everything was understood.

      Another brilliant feature was that every student worked at her own pace.

      • Good online learning and setting up a good course takes time and effort. Creating good exams - even in a traditional classroom and using pen and paper - takes time and effort. And generally, people are lazy.

        Not sure how it works at the K-12 level, but at the college/university level the publishing companies, etc. approach instructors about using their texts, etc. and say "Oh, and we have online resources with all the tests, etc. ready to go". I've not seen any where keyword abuse like this works, but fac

      • by Drethon ( 1445051 ) on Thursday September 03, 2020 @01:52PM (#60470132)

        It wasn't any specific teacher's decision to use Edgenuity. That kind of decision is taken at the level of school management at least - if not a school district or higher. Using "AI" to mark tests is exactly the sort of ignorant stupidity one would expect of politicians and managers, not teachers.

        From the article:

        "As COVID-19 has driven schools around the US to move teaching to online or hybrid models, many are outsourcing some instruction and grading to virtual education platforms".

        There have been excellent computer-based education suites ever since the 1950s - PLATO was a brilliant early example. I was greatly impressed by it when I saw it during a course with Control Data in 1971. https://en.wikipedia.org/wiki/... [wikipedia.org]

        But they have distinct limitations. The PLATO examples I saw were to do with maths and science; so most questions had definite right and wrong answers. The beautiful part of how PLATO worked was that trained teachers wrote the scripts, and included diagnosis of any mistakes. If a student got the solution of an equation wrong, he would be taken back to the relevant section and taken through it again more slowly, with more steps and examples to make sure everything was understood.

        Another brilliant feature was that every student worked at her own pace.

        I taught a college class that used automatically generated tests from a set of pre-existing, multiple choice, questions. I tried to scan the generated questions before each test, but I didn't always catch bad ones. I would go through the test results each time and if any question had too many students getting it wrong, I'd read over the question and if the wrong answer most people picked looked like something I might have picked, the grade was adjusted to give a pass to the students picking the bad answer. This isn't even factoring in an AI, it just doesn't work when tests are created without careful thought going into them.

        • by MobyDisk ( 75490 )

          I would go through the test results each time and if any question had too many students getting it wrong, I'd read over the question and if the wrong answer most people picked looked like something I might have picked, the grade was adjusted to give a pass to the students picking the bad answer.

          What you calculated is a piece of Item Response Theory (IRT) called the Discriminator. It is one of the ways the testing industry tests their tests. The testing company has people of varying but known skill levels take the test, then they data mine the results. For each question, calculate the Discriminator, which is a correlation between correct answers and candidate skill levels, ranging from -1 to 1. 1.0 = only the smart people got it right: good question, very difficult. 0.0 = nobody got it right, q

    • If the management decides to replace half of their workforce with an AI, why should the remaining workforce pay the price when the AI fails? Why shouldn't management be invited to take a long walk off a short plank?

      • by 1s44c ( 552956 )

        We always see people being called morons on slashdot because they manage systems that are badly configured. Nevermind that these people probably have hundreds of other systems to manage, are working 14 hour days, and have a shoestring budget and an obstructive PHB.

        It's just slashdot reasoning, blame the human nearest to the system.

    • I think its far too generous to call this an algorithm let alone an AI. There is little to no logic here. They are just setting some key words as the correct answer and searching the response for them. That's just a CTRL+F.

    • Umm, I know a teacher. Believe me, they have zero, nada, nothing to do with decisions like this. This stuff is in the purview of the superintendent who makes the big bucks. Fire your school board.
    • While I don’t think it is the teacher’s fault in this case, I did have AP clasees in school that the instructor checked off key words from the response. It is a common way to objectively score an otherwise subjective response. There was a subjective component as well though... but that was maybe 20% of the score.

      Point being, it is a stupid AI for focusing just on keywords and not linguistics. Or, a stupid AI for scoring keywords higher than linguistic integrity.

  • Artificial Idiot (Score:4, Insightful)

    by flyingfsck ( 986395 ) on Thursday September 03, 2020 @12:38PM (#60469854)
    Anything where an Artificial Idiot is involved, such as 'Resume Rewriting' and 'Search Engine Optimization', can be spoofed this way.
    • by 91degrees ( 207121 ) on Thursday September 03, 2020 @12:44PM (#60469886) Journal
      "AI" as a term is overused. This is clearly a pretty naive algorithm. There's nothing intelligent about it.
      • "AI" as a term is overused. This is clearly a pretty naive algorithm. There's nothing intelligent about it.

        But I answered the test by copy/paste the dictionary and ended up with an 174% grade. Are you telling me that the teacher is sometimes WRONG?

        AI is magic crap. Occasionally you actually know how it works, more occasionally you only THINK you do. And even MORE occasionally: as long as the test results seem correct, well that's good enough.

      • Perhaps, one might say: it's artificial. Like artificial flavouring, not the real thing, likely cheaper, not as healthy or even unhealthy, faked. Fits the bill if you ask me. The only problem is, artificial intelligence is talked about (and invested in) like it's the holy grail. Whereas the term could (rightly?) be taken as derogative.

        This message was brought to you through artificial stupidity (digital systems that have no pretence of intelligence) after having been cooked up by natural intelligence. (Pe

      • "AI" as a term is overused.

        No your understanding of it is just lacking. "AI" has a specific meaning when applied to computer algorithms and it covers a wide variety of algorithms from the woefully naive to the type that can outperform human experts.

    • by methano ( 519830 ) on Thursday September 03, 2020 @12:51PM (#60469918)
      My son told me that his resume .pdf has all the needed keywords for any job in a small font in white text printed at the end. In other words, visible to only the all knowing AI scanner. I don't know if it ever helped. He's got a pretty good job now because he's smart and he knew somebody who knew somebody. The latter probably got him the job and the former helps him keep it.

      Knowing how to beat the AI is probably the best job skill you can have these days.
      • There was a time when this worked, but the scripts that decide who gets considered for hiring have since been upgraded to detect this and disqualify people who do it.

    • by MrL0G1C ( 867445 ) on Thursday September 03, 2020 @12:53PM (#60469932) Journal

      Don't sound like AI at all to me, just a simple keyword match - very lazy system.

      • If it was trained with a data set, it is AI. Your Bayesian spam filter is AI.
        • by MrL0G1C ( 867445 )

          I don't agree that an algorithm that counts words is A.I.. And I think it's more likely that this company called it A.I. because they thought it'd sell better and be perceived as more valuable than a system where they give students a grade based upon them picking the right words.

          • by cb88 ( 1410145 )
            Just because you understand how it works doesn't make it any more or less AI than anything else.... AI is in essence finding a system that replaces intelligent management of a function with artificial management of that function, sometimes it is bad enough that you can easily figure out what it is. Even so called advanced AI often has such failings that are well understood... for instance there are cases were Nvidia's DLSS falls flat on it's face even in it's latest iteration... artifacts are still possible
  • Totally justified. (Score:5, Insightful)

    by Brain-Fu ( 1274756 ) on Thursday September 03, 2020 @12:40PM (#60469862) Homepage Journal

    If the student's grade is determined by a computer that doesn't actually understand the answer, then it can't possibly be an accurate representation of the student's level of understanding of the subject. This is inappropriate and unfair to the student. So, tossing in word salad is the best way for the student to defend himself from this unfairness.

    Well, that is the second best way. The best way would be to switch to an education provider that actually does their job.

    • Tossing in word salad is something most of us do regularly with search engines. It's quite inappropriate for teaching.

      But then history is perhaps the worst conceivable subject in which to bring in a computer. I studied history through school and university and even with very good teachers and small class sizes, I found it very hard to get a balanced, comprehensive understanding of anything. The world's best historians have often persuaded most people of a particular thesis, largely through sheer rhetorical

      • And what about one-letter misspellings like "Ceasar" - would they score zero?

        the sad answer here is "yes".
        I knew someone who was trying to get, in the US, some college courses done online and the grading system for math was an atrocity. Basically, it dismissed a right answer if it had the wrong amount of decimal places, or was rounded differently to what the AI expected, without having the decency of doing multiple choice if the answer was *that* fixed format, which let said friend to actively second-g

        • by DarkOx ( 621550 )

          That system and it sounds like the one in the TFS are not AI at all! When did searching for strings in text become AI?

          Is every perl monkey that has tossed together a few regex's suddenly an AI engineer now?

    • It's teaching students early that real understanding doesn't matter in life. This prepares them for resume writing and upper management.

      If they can learn the system, isn't that the better lesson?

    • by ranton ( 36917 )

      I hardly consider this cheating. It is more like jury rigging a broken grading system. It would likely be hard to come up with the correct keywords to add without understanding the subject matter being graded, so it isn't like the student wouldn't be learning. Just as long as the parent isn't doing it for the child each time.

    • I would sue.
  • I don't understand how these "AI"-companies get funded. Is this all part of a larger money laundry scam?

    • I don't understand how these "AI"-companies get funded. Is this all part of a larger money laundry scam?

      There is no need for malice or money laundering. The situation includes two of the most common factors: lazy people and money for automated services.

      I've worked as adjunct faculty at university with intro CS courses. While many people would teach, a few instructors were lazy, stating we didn't need to actually review what the students submitted, just take the scores generated by the script and copy them to the grades; tests submitted electronically they would hit the 'normalize scores' button in Canvas an

    • Hey it only needs to work a few times. How much do you think the district paid for this software? I'm guessing an average years salary for a teacher. Sell shitty software that works at first glance to a few schools and you're set.

  • by SuperKendall ( 25149 ) on Thursday September 03, 2020 @12:42PM (#60469874)

    'What was the advantage of Constantinople's location for the power of the Byzantine empire,'" Simmons says. "So you go through, okay, what are the possible keywords that are associated with this? Wealth, caravan, ship, India, China, Middle East,...

    If you know enough to put together the right list of keywords, for that kind of question isn't that pretty much good enough? It's not just word salad, it's a concise way of listing important factors that require some understanding of the subject.

    • If you know enough to put together the right list of keywords....isn't that pretty much good enough?

      No, because this kid was not properly gaming the system but was trying to make a semi-decent try to get around a broken grading system. If you really wanted to set out to game the system just paste a large enough dictionary word list at it and all your answers to anything will be graded as perfect without you even needing to read the question: that's how utterly broken this "grading" system is.

      • by Shotgun ( 30919 )

        So, what you're saying is that the answer to life, the universe and everything was NOT 42? Instead, it was Webster's Dictionary?

    • by Headw1nd ( 829599 ) on Thursday September 03, 2020 @12:58PM (#60469958)

      Not really, because you have no idea how the keywords are related. Were they trading with the Middle East or did they conquer them? Was trade with china revolutionary or expected? Was India a trading partner of part of the empire? Was the location good because it was near China or because it was far away?

      The easy way to show this is take the list of keywords and a person who doesn't know the actual answer (or better yet, the question) and tell them to write out several facts, then see how well they did.

    • by dfghjk ( 711126 )

      So "some understanding of the subject" is "pretty much good enough"? Good enough to warrant a perfect score?

      The "right list of keywords" is likely to be "put together"-able using an algorithm. Would you say then that knowing to use software to get perfect scores on tests is "pretty much good enough"?

      I know your standard doesn't really require knowing anything, SuperKendall, but others want better for our children...and future productive members of society.

      • by DarkOx ( 621550 )

        WebDriver the entire lesson, capturing screenshots on the way, feed images to tesseract generating some sorta reasonable text, create Winword COM object, open the text file, fire off auto-summary, past the result into the quiz.

        If its like those dumb compliance training things they make us grown ups do at work, you can probably automate the entire thing. (but I would never do that of course, because that would be wrong ;-) )

    • I disagree. [lmgtfy.com]
    • Totally missed opportunity to use for an answer -
      So, take me back to Constantinople
      No, you can't go back to Constantinople
      Been a long time gone, Constantinople
      Why did Constantinople get the works?
      That's nobody's business but the Turks

    • Presumably based on the description there are no âoeincorrectâ keywords and so just spamming every answer with a singular list of keywords would get you a passing grade.

      And even if China was a keyword, knowing a keyword doesnâ(TM)t prove knowledge sufficient to answer a question. âoeConstantinople is well positioned on the Silk Road where it can launch nukes at Chinaâ is clearly a wrong answer to âoewhy did the Silk Road go through Constantinople?â

    • 'What was the advantage of Constantinople's location for the power of the Byzantine empire,'" Simmons says. "So you go through, okay, what are the possible keywords that are associated with this? Wealth, caravan, ship, India, China, Middle East,...

      If you know enough to put together the right list of keywords, for that kind of question isn't that pretty much good enough? It's not just word salad, it's a concise way of listing important factors that require some understanding of the subject.

      (Teacher, asking the question IRL) 'What was the advantage of Constantinople's location for the power of the Byzantine empire?"

      (Your "legit" Student) "Caravan. India. China. Ship."

      ("Teacher") "Good enough."

      We should stop complaining about how stupid people are. It's obvious as to why when you set the "high" bar on the fucking floor.

  • by PCM2 ( 4486 ) on Thursday September 03, 2020 @12:43PM (#60469878) Homepage

    I once heard of a guy who gamed the dumb algorithms that some recruiters use to screen job applicants in much the same way. He knew some software was going to process his CV before it ever made it to a human. So he just submitted his resume as the normal PDF or Word doc, but at the bottom of the document, he inserted a huge word salad of every possible keyword he thought would be relevant to the position, rendered in 2-point type in white lettering on a white background. Any human who actually opened the PDF would never see what he was doing ... but the stupid machine would, and it would increase his chances of getting the interview (which, just like a batter making it to first base in baseball, might be all that really matters).

    Is this cheating? Is it unethical? I don't know. If a dog figures out how to get treats from you, is it a bad dog?

    • by fulldecent ( 598482 ) on Thursday September 03, 2020 @12:48PM (#60469906) Homepage

      That's me. And here is the article, including the full keyword list (at bottom) https://fulldecent.blogspot.co... [blogspot.com]

    • by dfghjk ( 711126 )

      Whether it is cheating or unethical depends on circumstance. In this particular resume example it is neither cheating nor unethical, it is just disappointing because it's a workaround for something unethical. That doesn't mean doing that is never cheating or unethical.

      Our society is increasingly about gaming everything. People should view that generally as "cheating and unethical".

    • by Anubis IV ( 1279820 ) on Thursday September 03, 2020 @02:36PM (#60470320)

      I remember an intro-level Physics class at my university, in which they had each student take a calculus quiz to establish that the student possessed the prerequisite level of calculus knowledge. It was an online pass/fail quiz with 10 multiple choice questions, but a 100% was the only passing grade, and you could take it as many times as you wanted.

      The questions weren't particularly difficult, but each one took a bit of time to work through. I spent 30 minutes solving the 10 questions, submitted my answers, and got 9 out of 10...I had made a trivial mistake that caused me to miss one of them, and it just so happened that they had a choice that matched that particular mistake. I refreshed the page, got a number of new questions mixed in with a few of the ones I had already done, spent another 20-25 minutes solving them, submitted my answers, and got 9 out of 10 again. A different trivial mistake had cost me an answer.

      Rather than solving them again for a third time, I submitted random choices for the new questions while choosing the ones I knew to be correct for the old questions. Submit. 4 out of 10. Write down the answers I got right. Repeat. Lots of new questions. Fill in. Submit. 5 out of 10. Repeat. More new questions. Repeat. More new questions. Repeat. Slowly the scores went up as there were fewer questions I hadn't yet seen and determined the answer to. 8 out of 10. 9 out of 10. Finally, a perfect score after a large number of additional attempts, though it only cost me maybe another 30 minutes, so not much more than one more attempt would have taken.

      The next morning I headed into class. The professor proclaimed that everyone in the 100-person class had passed and that he was very proud of the hard work we had all put in, especially the one student who took the quiz 53 times.

      *cough*

      I regret nothing.

  • That's not AI (Score:4, Insightful)

    by lsllll ( 830002 ) on Thursday September 03, 2020 @12:43PM (#60469884)
    This is just some program looking for keywords in answers. Any half-decent AI would do much better than this.
  • Good education (Score:4, Insightful)

    by Dog-Cow ( 21281 ) on Thursday September 03, 2020 @12:44PM (#60469890)

    I wonder who the genius was that decided the best lesson that could be taught is how to cheat. When the mother, who is a teacher herself, tells her son to cheat, you know it's completely broken.

  • This is not AI (Score:3, Insightful)

    by n0n0n4t0r ( 7204206 ) on Thursday September 03, 2020 @12:45PM (#60469894)
    This is an algorithm, and a stupid one not AI.
    • by flex941 ( 521675 )
      Every algo is AI nowadays it seems, like every server not on your premises became cloud.
    • AI is an algorithm. Even simple word selection can be trained through an AI algorithm. Stop passing judgement on an output when you have no idea what happens underneath. Just because an algorithm is clever or complex doesn't mean it's results aren't stupid. Just because its results are stupid doesn't mean it is or isn't AI based.

  • PERFECT! (Score:5, Insightful)

    by darkain ( 749283 ) on Thursday September 03, 2020 @12:49PM (#60469916) Homepage

    To be fair, this was probably the absolute BEST lesson the kids could get on what REAL LIFE is actually like in the present day world.

    Guess how recruiters evaluate thousands of resumes? YUP! Exact same bullshit, using a computer that filters on keywords.

    These kids now have an invaluable employable skill thanks to these bullshit tests.

    I give the kids an "A" for being such forward thinkers!

    • by dfghjk ( 711126 )

      Imagine a future society when this was "the absolute BEST lesson" that kids learned. I don't think your opinion has much "actual" merit.

      • It is indeed the best, just not what others would prefer the "best" be. Evolution and human society are competitive, be it Zog braining Og with a club to take his mate or modern wars for resources.
        Evil wins more often than not and when it doesn't we confuse that with karma instead of (insert real causes).

      • Fake it until you make it. How much of your education do you actually use to do your daily job? Chances are you're using experience more than anything.

    • Itâ(TM)s a great lesson to learn, but this was a history class, not a resume-writing class.

      • by Shotgun ( 30919 )

        It was a life skills class.

        Don't look at just how to solve the problem presented to you. Look at what is driving the request and what supports it.

        I am respected at my work because I do this. I get a request like "We need a system to do X", and (when appropriate) my first response is, "Is that really what we want, because the root of the problem is something else. What happens if we tackle that directly?"

  • Apparently, that "word salad" is enough to get a perfect grade on any short-answer question in an Edgenuity test.

    Or to be a mid-level manager or resume writer.

  • by zuckie13 ( 1334005 ) on Thursday September 03, 2020 @12:53PM (#60469936)
    Remember when teachers were better than this - when they wanted to do the work themselves. Back when teaching was treated as a to profession, not a fallback when you failed to do something else. I remember graduating HS in 1999, and many of the good teachers were retiring at that time. By the time my brother went through three years later, he had a pile of them that were just going through the motions - and we're talking in honors level classes here. Guess this is what happens when we pay teachers peanuts. Try convincing someone (that had to get a college degree) to take a job that pays 40k when the have offers from companies to pay twice that.
    • This has nothing to do with teachers. This has everything to do with various states appointing school boards whose job it is to ensure that the absolute minimum money is spent on school so they can help pay down debt or fund other projects.

      The teachers don't opt into this kind of thing, it's mandated at the board level.

      So stop throwing rocks at the teachers, start throwing rocks at your governors and mayors. Think about bond issues before you check 'yes' or 'no'.

    • by djp2204 ( 713741 )
      Sounds like we need to abolish the public school system and give everyone vouchers. Some will go to private schools, some will form pods and hire teachers that are very well paid. Why waste all this money on buildings, busses, books, etc?
  • As an IT guy at a school, I really enjoy stories like this. Mainly because the people making decisions don't know anything about IT, and worse, don't want to know.

    This isn't cheating, and I will stand up for any student who does this. Teachers, stop having machines do your job, and do your job. Kids aren't machines.

    • It isn't teachers who aren't doing their job. It's administrators who want to save money.

      Do you really think teachers have a choice about stuff like this?

  • Not Cheating (Score:5, Insightful)

    by gurps_npc ( 621217 ) on Thursday September 03, 2020 @01:03PM (#60469980) Homepage

    The school created a stupid way to test, then gave students really poor instructions on how to get the 'right' answers for a stupid test.

    The students discovered the REAL way to get the 'right answers', and provided it.

    They demonstrated a) an understanding of what the test was really asking, b) knowledge of the proper 'answer', c) problem solving in figuring out how to provide the 'proper' answer.

    They should be applauded for their intelligence, not denigrated for figuring out the right way to get a good score on a badly designed test.

    • It sounds like the the free text equivalent of a multiple choice test that you can easily ace because checking all answers guarantees you checked the right answers. Anyway, the poor students who learn this way will hit a wall when they actually need to reason about things because the answer space becomes too big to memorize. You know the type: They complain that there is no way anyone can learn that much by heart. They're baffled by their peers who still "know" the answers. The problem with this test isn't
      • You are correct that the teachers have failed to teach and have tested merely for paying attention rather than understanding.

  • If a normal teacher had words they loved to hear and you kept lacing your sentences with them which caused you to get unnaturally high grades, we might say the teacher is incompetent but not that mock-fawning students were cheating.

  • Homeschool?

    The teacher may not be as educated, but what they miss in book learning, they make up for in attention because the teacher to student ratio is really good.

    Not for everybody, but if the kids are going to be at home anyway, why not?

    • by Shotgun ( 30919 )

      How educated do you have to be to teach a middle schooler?

      Seriously, how does that require a four year college education?

      • It doesn't... My wife taught our two children from birth through high school and she's a highs cool graduate with 2 years of community college. My kids both graduated from a well known university with STEM degrees with reasonable GPA's.

        Homeschooling is more about commitment and caring about your kid's education than it is about the teacher's mastery of the subjects being taught. Once you get the kids past the reading, writing and basic math part in a homeschool, it gradually moves to a 'self taught" mod

      • You're right. Employ high school students instead.

  • "Virtue" = human self-sacrifice for others to their own disadvantage. Society is bucket of crabs. It is not sociopathic to game an evil society run by evil, stupid and clueless people when you cannot possibly change anything but can be punished for simple honesty.

    I have more respect for those who lie, cheat and game the enemy system than for chumps who believe it can be fixed. AI will not at first (if ever) be used for "good", just "convenience" of those in power. It is personal and social duty to learn ho

  • HR departments routinely have an algorithm scan resumes. Just throw some "word salad" in 1pt white font at the end of your resume, and you pass the test.
  • The real intelligence of teachers can be gamed in the same way. Each one is different, and a student can get much better grades by just paying attention to what the teacher wants and doing it.

  • This is not AI but any definition. If it was, it would be able to understand the meaning of the written text. Looking for keywords = a text search algorithm.
  • I had a very good history teacher in high school, who one time dinged me for not mentioning a certain name in an essay/answer on a test. I had the name written next to the question, but somehow neglected to remember to incorporate it into the essay. Had I been graded by a robot I would have gotten a better grade!

    Really though I don't think computers should be grading written assignments. If the teachers don't have time to grade an essay, then reduce the essay answers to more multiple choice or something.

  • There is no difference between this and learning what buzzwords your human instructor wants to see and then focusing your answers around them. If that's "cheating", then welcome to the twentieth century where anyone with half a brain did the exact same thing.
  • by VeryFluffyBunny ( 5037285 ) on Thursday September 03, 2020 @03:42PM (#60470560)

    If it can be graded by a computer, it can be gamed by a computer. Why take the trouble of writing sentences & thinking up keywords when a computer can do that for you?: https://babel-generator.heroku... [herokuapp.com]

    Even the most advanced, bleeding edge AI can't understand what human language means. It can't distinguish between a meaningful sentence &, "Colourless green ideas sleep furiously." It also can't tell you the correct answer to Winograd schemas, which are trivial for humans to answer.

    If we want our next generations to learn useful, coherent, cohesive stuff so that they can keep the world running when we're too old, stop trying to teach or test them with AI.

What hath Bob wrought?

Working...