M\u00edsto toho si te\u010f m\u016f\u017eeme kl\u00e1st d\u016fle\u017eit\u011bj\u0161\u00ed ot\u00e1zku: Jak\u00e9 dopady m\u00e1 u\u017e\u00edv\u00e1n\u00ed AI na \u017e\u00e1ky?<\/p>\n
U\u017e k\u00a0tomu m\u00e1me v\u00fdzkumy z\u00a0cel\u00e9ho sv\u011bta. Neur\u010dit\u00e9 spekulace o\u00a0mo\u017en\u00fdch dopadech p\u011bkn\u011b vybarvila t\u0159eba leto\u0161n\u00ed studie publikovan\u00e1 v\u00a0SSRN<\/a>. V\u00fdzkumn\u00edci sledovali v\u00fdsledky 26\u00a0tis\u00edc \u010d\u00ednsk\u00fdch student\u016f v\u00a0dev\u00edti p\u0159edm\u011btech po dobu 30\u00a0m\u011bs\u00edc\u016f, tedy prakticky od za\u010d\u00e1tku n\u00e1stupu generativn\u00ed AI. Dodejme, \u017ee ne\u0161lo o\u00a0studii randomizovanou\u00a0\u2013 ta by byla ur\u010dit\u011b lep\u0161\u00ed, ale v\u00a0praxi neprovediteln\u00e1.<\/p>\n Nam\u00edsto toho v\u011bdci zajistili, \u017ee skupiny \u201enikdy nepou\u017e\u00edv\u00e1 AI\u201c a \u201epou\u017e\u00edv\u00e1 AI\u201c byly srovnateln\u00e9 co do v\u00fdsledk\u016f p\u0159ed n\u00e1stupem AI, aby se p\u0159ede\u0161lo v\u00fdb\u011brov\u00e9mu zkreslen\u00ed. Ka\u017ed\u00fd \u017e\u00e1k je tak srovn\u00e1n se svou vlastn\u00ed predikovanou trajektori\u00ed. Co\u017e umo\u017enilo uk\u00e1zat, jak se zm\u011bnily v\u00fdsledky \u017e\u00e1k\u016f, kte\u0159\u00ed za\u010dali AI pou\u017e\u00edvat.<\/p>\n Nejprve se pod\u00edv\u00e1me na to, co nikoho nep\u0159ekvap\u00ed: Um\u011bl\u00e1 inteligence \u017e\u00e1k\u016fm umo\u017enila \u0159e\u0161it \u00fakoly rychleji. Zlom byl t\u00e9m\u011b\u0159 okam\u017eit\u00fd. Z\u00a0p\u016fvodn\u00edch 65\u00a0minut se \u017e\u00e1ci d\u00edky chatbot\u016fm rychle dostali na 45\u00a0minut. To se m\u016f\u017ee zd\u00e1t jako mal\u00e9 zrychlen\u00ed, ale jde o\u00a0ohromn\u00fd a\u00a0statisticky v\u00fdznamn\u00fd efekt.<\/p>\n Vid\u00edme, \u017ee skupina \u201ebez AI\u201c (zelen\u00e1) a\u00a0skupina \u201epou\u017e\u00edv\u00e1m AI\u201c (\u010derven\u00e1) se t\u00e9m\u011b\u0159 nep\u0159ekr\u00fdv\u00e1. Modr\u00e1 ukazuje stav p\u0159ed n\u00e1stupem ChatGPT a\u00a0v podstat\u011b se shoduje se zelenou skupinou.<\/p>\n Nesta\u010d\u00ed ale v\u011bci zrychlit. Chceme, aby se studenti zlep\u0161ovali. A\u00a0i to um\u011bl\u00e1 inteligence\u00a0\u2013 rychle, jakoby m\u00e1vnut\u00edm kouzeln\u00e9ho proutku\u00a0\u2013 zjevn\u011b dok\u00e1\u017ee zajistit. Hodnocen\u00ed dom\u00e1c\u00edch \u00fakol\u016f se u\u00a0\u017e\u00e1k\u016f vyu\u017e\u00edvaj\u00edc\u00edch chatboty v\u00fdrazn\u011b a\u00a0nep\u0159ehl\u00e9dnuteln\u011b zlep\u0161ilo.<\/p>\n Op\u011bt je vid\u011bt, \u017ee skupina \u201ebez AI\u201c (zelen\u00e1) a\u00a0skupina \u201epou\u017e\u00edv\u00e1m AI\u201c (\u010derven\u00e1) jsou v\u00fdrazn\u011b odli\u0161n\u00e9. Kvalita dom\u00e1c\u00edch \u00fakol\u016f u\u00a0d\u011bti pou\u017e\u00edvaj\u00edc\u00edch chatboty v\u00fdrazn\u011b a\u00a0rychle vzrostla. <\/p>\n Zjednodu\u0161en\u011b \u0159e\u010deno: Pr\u016fm\u011brn\u00fd \u017e\u00e1k pou\u017e\u00edvaj\u00edc\u00ed AI je v\u00a0\u0159e\u0161en\u00ed dom\u00e1c\u00edch \u00fakol\u016f lep\u0161\u00ed ne\u017e p\u0159es 95\u00a0% spolu\u017e\u00e1k\u016f, kte\u0159\u00ed AI nepou\u017e\u00edvaj\u00ed. Na \u010dasov\u00e9 ose p\u0159itom vid\u00edme, \u017ee jde o\u00a0stejn\u00e9ho \u017e\u00e1ka, kter\u00fd je\u0161t\u011b p\u0159ed dv\u011bma roky pat\u0159il k\u00a0pr\u016fm\u011bru. Ve statistick\u00e9m jazyce jde \u201eo zlep\u0161en\u00ed o\u00a0dv\u011b sigmy\u201c (konkr\u00e9tn\u011b 1,9\u00a0SD), tedy t\u00e9m\u011b\u0159 o\u00a0dvojn\u00e1sobek sm\u011brodatn\u00e9 odchylky.<\/p>\n \u0158e\u0161en\u00ed nejv\u011bt\u0161\u00edho probl\u00e9mu v\u00fduky?<\/p>\n V pedagogice m\u00e1 pojem \u201edv\u011b sigmy\u201c t\u00e9m\u011b\u0159 magick\u00fd v\u00fdznam. V\u00a0roce 1984\u00a0publikoval Benjamin Bloom, americk\u00fd psycholog z\u00a0MIT, \u010dl\u00e1nek s\u00a0n\u00e1zvem The 2\u00a0Sigma Problem<\/a>. Ve sv\u00e9m (mal\u00e9m, ale randomizovan\u00e9m) v\u00fdzkumu<\/a> uk\u00e1zal ohromn\u00fd v\u00fdznam individu\u00e1ln\u00edho lektora pro \u00fasp\u011bch student\u016f (v jeho p\u0159\u00edpad\u011b student\u016f doktorsk\u00e9ho studia).<\/p>\n Pokud m\u011bl student k\u00a0dispozici lektora, kter\u00fd se mu mohl dlouhodob\u011b a\u00a0individu\u00e1ln\u011b v\u011bnovat, v\u00fdsledky tohoto studenta vysko\u010dily strm\u011b vzh\u016fru. P\u016fvodn\u011b pr\u016fm\u011brn\u00fd student byl najednou v\u00a0horn\u00edch dvou percentilech nejlep\u0161\u00edch student\u016f.<\/p>\n To by \u0161lo zd\u00e1nliv\u011b ozna\u010dit za ohromn\u00fd \u00fasp\u011bch, ale Bloom to nazval probl\u00e9mem, z\u00a0jeho pohledu mo\u017en\u00e1 \u00fast\u0159edn\u00edm probl\u00e9mem pedagogiky jako takov\u00e9. D\u00e1t ka\u017ed\u00e9mu studentovi individu\u00e1ln\u00edho lektora, kter\u00fd nem\u00e1 na starosti nikoho jin\u00e9ho, je toti\u017e naprosto nerealistick\u00e9 \u0159e\u0161en\u00ed, kter\u00e9 se ve vzd\u011bl\u00e1vac\u00edm syst\u00e9mu ned\u00e1 \u0161k\u00e1lovat. Individu\u00e1ln\u00ed v\u00fduka je luxus a\u00a0Bloom vyzval, aby se t\u00e9to la\u0165ce sna\u017eili u\u010ditel\u00e9 r\u016fzn\u00fdmi zp\u016fsoby p\u0159ibl\u00ed\u017eit i\u00a0ve skupinov\u00e9 v\u00fduce.<\/p>\n Pro \u00faplnost je pot\u0159eba dodat, \u017ee pozd\u011bj\u0161\u00ed pokusy o\u00a0replikaci uk\u00e1zaly, \u017ee individu\u00e1ln\u00ed v\u00fduka nem\u00e1<\/a> v\u017edy tak vysok\u00fd \u00fa\u010dinek, jako nam\u011b\u0159il Bloom. Ale o\u00a0tom, \u017ee jde o\u00a0skv\u011bl\u00fd zp\u016fsob, jak se student\u016fm v\u011bnovat, nen\u00ed pochyb. Bloom to p\u0159i\u010d\u00edtal zejm\u00e9na t\u0159em aspekt\u016fm: ka\u017ed\u00e1 chyba je okam\u017eit\u011b ozna\u010dena a\u00a0\u017e\u00e1k se m\u016f\u017ee pou\u010dit ihned, lektor se v\u011bnuje studentovi individu\u00e1ln\u011b a\u00a0spole\u010dn\u00fd \u010das je str\u00e1ven\u00fd s\u00a0vysokou m\u00edrou pozornosti.<\/p>\n \u010cty\u0159icet let od Bloomova \u010dl\u00e1nku o \u201enedosa\u017eiteln\u00e9m etalonu individu\u00e1ln\u00ed v\u00fduky\u201c tu najednou m\u00e1me chatboty, kte\u0159\u00ed jsou na prvn\u00ed pohled dokonalou odpov\u011bd\u00ed na tento probl\u00e9m. Jsou relativn\u011b levn\u00ed, tak\u017ee si \u0161koly mohou dovolit jejich nasazen\u00ed na \u00farovni jednotliv\u00fdch \u017e\u00e1k\u016f. Jsou trp\u011bliv\u00ed, nevad\u00ed jim stejnou l\u00e1tku vysv\u011btlit tis\u00edckr\u00e1t, poka\u017ed\u00e9 trochu jinak, poka\u017ed\u00e9 na m\u00edru studentce \u010di studentovi, ani brvou nehnou. Jak doufal<\/a> v\u00a0roce 2023\u00a0Sal Khan, zakladatel slavn\u00e9 v\u00fdukov\u00e9 platformy Khan Academy, v\u00fdukov\u00ed chatboti by mohli vy\u0159e\u0161it probl\u00e9m, se kter\u00fdm si pedagogika nev\u011bd\u011bla rady.<\/p>\n Posledn\u00ed dobou u\u017e jsou to ma\u0161inky pom\u011brn\u011b chytr\u00e9, tak\u017ee a\u017e na ob\u010dasn\u00e9 halucinace je na jejich odpov\u011bdi v\u00a0kontextu z\u00e1kladn\u00ed a\u00a0st\u0159edn\u00ed \u0161koly pom\u011brn\u011b spoleh. Ukazuj\u00ed to i\u00a0v\u00fd\u0161e zmi\u0148ovan\u00e1 hodnocen\u00ed. Je tedy probl\u00e9m vy\u0159e\u0161en?<\/p>\n Z\u00e1le\u017e\u00ed na tom, co m\u011b\u0159\u00edme. Pokud n\u00e1m z\u00e1le\u017e\u00ed na dom\u00e1c\u00edch \u00fakolech, chatboti skute\u010dn\u011b pom\u00e1haj\u00ed a\u017e z\u00e1zra\u010dn\u011b. Pokud n\u00e1m ale z\u00e1le\u017e\u00ed na studentech, pak se mus\u00edme zam\u011b\u0159it na dlouhodob\u00e9 v\u00fdsledky. A\u00a0proto je v\u00a0leto\u0161n\u00ed studii v\u00fdsledk\u016f \u010d\u00ednsk\u00fdch student\u016f je\u0161t\u011b graf C.<\/p>\n V\u00fdsledky z\u00a0test\u016f naopak ukazuj\u00ed, \u017ee skupina pou\u017e\u00edvaj\u00edc\u00ed AI (\u010derven\u00e1) si v\u00fdrazn\u011b a\u00a0t\u00e9m\u011b\u0159 okam\u017eit\u011b pohor\u0161ila. Zhor\u0161en\u00ed se projevilo u\u017e b\u011bhem prvn\u00edho m\u011bs\u00edce a\u00a0ust\u00e1lilo se do 6\u00a0m\u011bs\u00edc\u016f po za\u010d\u00e1tku pou\u017e\u00edv\u00e1n\u00ed AI n\u00e1stroj\u016f. <\/p>\n Um\u011bl\u00e1 inteligence \u017e\u00e1k\u016fm skute\u010dn\u011b p\u0159inesla posun o\u00a0vysn\u011bn\u00e9 dv\u011b sigmy (a z\u0159ejm\u011b i\u00a0o v\u00edc), ale tam, kde na tom v\u016fbec nez\u00e1le\u017e\u00ed. Kdy\u017e p\u0159i\u0161lo na prok\u00e1z\u00e1n\u00ed nabyt\u00fdch znalost\u00ed, do\u0161lo u\u00a0\u017e\u00e1k\u016f pou\u017e\u00edvaj\u00edc\u00edch AI naopak k\u00a0v\u00fdrazn\u00e9mu propadu o\u00a0asi jeden a\u00a0p\u016fl standardn\u00ed odchylky. <\/p>\n Pro p\u0159edstavu, v\u00a0\u010desk\u00e9m syst\u00e9mu zn\u00e1mkov\u00e1n\u00ed by to znamenalo, \u017ee se z\u00a0dvojka\u0159e stal \u010dty\u0159ka\u0159. Pr\u016fm\u011brn\u00fd \u017e\u00e1k se najednou propadl mezi 8\u00a0% nejhor\u0161\u00edch. A\u00a0u t\u011bch, kte\u0159\u00ed pou\u017e\u00edvali AI pro dom\u00e1c\u00ed \u00fakoly nejv\u00edce, se zhor\u0161en\u00ed projevilo nejv\u00fdrazn\u011bji. <\/p>\n Paradoxn\u011b se v\u00fdborn\u011b zpracovan\u00e9 dom\u00e1c\u00ed \u00fakoly (s dopomoc\u00ed chatbota) staly znakem toho, \u017ee se \u017e\u00e1k\u016fm nebude da\u0159it u\u00a0test\u016f nebo u\u00a0p\u0159ij\u00edmac\u00edch zkou\u0161ek, kde chatbota k\u00a0dispozici nemaj\u00ed.<\/p>\n Mozek jako sval<\/p>\n V\u00fdsledky z\u00a0\u010c\u00edny nejsou ojedin\u011bl\u00e9. Po cel\u00e9m sv\u011bt\u011b prob\u00edhaj\u00ed experimenty s\u00a0podobn\u00fdmi v\u00fdsledky: Um\u011bl\u00e1 inteligence ve v\u00fduce prakticky v\u0161ude zvy\u0161uje rychlost, pohodl\u00ed a\u00a0\u010dasto vede i\u00a0ke kvalitn\u011bj\u0161\u00edm v\u00fdsledk\u016fm. Jen\u017ee kdy\u017e pak p\u0159ijde na m\u011b\u0159en\u00ed toho, co si studenti skute\u010dn\u011b zapamatovali, v\u00fdsledek b\u00fdv\u00e1 tragick\u00fd.<\/p>\n V Turecku<\/a> zkusili d\u00e1t \u017e\u00e1k\u016fm ChatGPT p\u0159i \u0159e\u0161en\u00ed matematick\u00fdch slovn\u00edch \u00faloh a\u00a0jejich \u00fasp\u011b\u0161nost r\u00e1zem stoupla o\u00a048\u00a0%. Ale kdy\u017e pak m\u011bli \u0159e\u0161it \u00falohy sami, najednou se situace obr\u00e1tila a\u00a0\u017e\u00e1ci zvykl\u00ed na pomoc od chatbota se oproti spolu\u017e\u00e1k\u016fm v\u00a0testech propadli o\u00a0skoro p\u011btinu.<\/p>\n V N\u011bmecku otestovali<\/a> dopady zapojen\u00ed AI a\u00a0dosp\u011bli k\u00a0podobn\u00fdm v\u00fdsledk\u016fm: Studenti, u\u00a0kter\u00fdch bylo detekov\u00e1no pou\u017eit\u00ed AI v\u00a0semin\u00e1rn\u00edch prac\u00edch (co\u017e nen\u00ed zcela spolehliv\u00e1 metoda), m\u011bli hor\u0161\u00ed v\u00fdsledky u\u00a0z\u00e1v\u011bre\u010dn\u00fdch zkou\u0161ek. Podrobn\u011bj\u0161\u00ed anal\u00fdza pak uk\u00e1zala, \u017ee nejv\u011bt\u0161\u00ed zhor\u0161en\u00ed se paradoxn\u011b t\u00fdkalo student\u016f s\u00a0nejv\u011bt\u0161\u00edm potenci\u00e1lem. Ti toti\u017e m\u011bli nejv\u00edce co ztratit: Chatbot jakoby jim zabr\u00e1nil u\u010dit se l\u00e9pe. Nebo jinak, umo\u017enil jim nenau\u010dit se to, co by se jinak nau\u010dili.<\/p>\n Op\u011bt se vrac\u00edm ke sv\u00e9 obl\u00edben\u00e9 metafo\u0159e<\/a>: Pokud bych p\u0159i\u0161el do posilovny, dal robotovi \u010dinky a\u00a0nechal ho za m\u011b odcvi\u010dit v\u0161echny s\u00e9rie, v\u00fdsledek by byl precizn\u00ed, rychl\u00fd a\u00a0pro m\u011b rozhodn\u011b pohodln\u011bj\u0161\u00ed. Ale v\u0161ichni ch\u00e1peme, \u017ee to jde proti samotn\u00e9mu smyslu cvi\u010den\u00ed.<\/p>\n Klik\u016f bude v\u00edce a\u00a0budou m\u00edt perfektn\u00ed formu. Ale komu ty robotick\u00e9 kliky pomohly?<\/p>\n U sval\u016f je tento mechanismus na prvn\u00ed pohled jasn\u00fd. Ale i\u00a0u mozku tr\u00e9nov\u00e1n\u00ed funguje, a\u00a0pro pedagogy to nen\u00ed \u017e\u00e1dn\u00fdm tajemstv\u00edm. Mozek informace nejl\u00e9pe vst\u0159eb\u00e1, kdy\u017e je to spojen\u00e9 s\u00a0ur\u010dit\u00fdm \u00fasil\u00edm, nebo dokonce nepohodl\u00edm. Tento koncept m\u00e1 r\u016fzn\u00e9 varianty a\u00a0n\u00e1zvy, kter\u00e9 mimochodem \u010dasto p\u0159\u00edmo odkazuj\u00ed na metaforu fyzick\u00e9 n\u00e1mahy: \u017e\u00e1douc\u00ed obt\u00ed\u017ee, u\u017eite\u010dn\u00e9 \u00fasil\u00ed, relevantn\u00ed kognitivn\u00ed z\u00e1t\u011b\u017e.<\/p>\n Neznamen\u00e1 to, \u017ee u\u010den\u00ed mus\u00ed doprov\u00e1zet bolest nebo \u00famorn\u00e9 memorov\u00e1n\u00ed. Ukazuje se ale, \u017ee nejl\u00e9pe si n\u00e1\u0161 mozek osvojuje nov\u00e9 koncepty, kdy\u017e je vystaven \u201eide\u00e1ln\u00ed m\u00ed\u0159e nepohodl\u00ed\u201c. Tedy kdy\u017e u\u010ditel studentovi d\u00e1vkuje nov\u00e9 koncepty tak, aby mohl stav\u011bt na tom, co u\u017e zn\u00e1, a\u00a0z\u00e1rove\u0148 nar\u00e1\u017eel na slep\u00e1 m\u00edsta, se kter\u00fdmi si nev\u00ed rady.<\/p>\n Legendy didaktiky o\u00a0u\u017eite\u010dn\u00e9 n\u00e1maze<\/p>\n Pr\u00e1v\u011b tato kognitivn\u00ed n\u00e1maha je odm\u011bn\u011bna t\u00edm, \u017ee si mozek nov\u00e9 poznatky pevn\u011bji za\u0159ad\u00ed do ji\u017e existuj\u00edc\u00edch sch\u00e9mat v\u00a0hlav\u011b. Kognitivn\u00ed psycholog Daniel Willingham to shrnul<\/a> pou\u010dkou: \u201eZapamatovan\u00e1 v\u011bdomost je stopou po p\u0159em\u00fd\u0161len\u00ed.\u201c <\/p>\n Kdy\u017e toti\u017e chatbot v\u0161echny v\u011bci perfektn\u011b vysv\u011btl\u00ed na prvn\u00ed pokus, nen\u00ed \u017e\u00e1dn\u00fd d\u016fvod se nam\u00e1hat. \u017d\u00e1k odpov\u011b\u010f op\u00ed\u0161e (v hor\u0161\u00edm p\u0159\u00edpad\u011b zkop\u00edruje) a\u00a0m\u00e1 hotovo. Klik, klik, klik.<\/p>\n M\u00edsto sluhy cht\u011bjte Aristotela<\/p>\n P\u0159\u00edstup ke\u00a0specializovan\u00fdm chatbot\u016fm nem\u00e1 historickou obdobu, co se t\u00fd\u010de jejich rychlosti a\u00a0dostupnosti. Ale nejmocn\u011bj\u0161\u00ed a\u00a0nejbohat\u0161\u00ed lid\u00e9 si mohli v\u017edy dovolit sluhy, kte\u0159\u00ed za jejich d\u011bti ud\u011blaj\u00ed ve\u0161ker\u00e9 \u00fakoly. A\u00a0u\u017e v\u00a0antice v\u011bd\u011bli, \u017ee takov\u00fdm zp\u016fsobem nevychovaj\u00ed samostatn\u00e9 jedince, ale jedince bezmocn\u00e9.<\/p>\n Kdy\u017e cht\u011bl makedonsk\u00fd kr\u00e1l Filip II. pro sv\u00e9ho syna to nejlep\u0161\u00ed vzd\u011bl\u00e1n\u00ed, nenajal mu osobn\u00edho p\u00edsa\u0159e, kter\u00fd za n\u011bj v\u0161echno nap\u00ed\u0161e. Pozval si mo\u017en\u00e1 nejn\u00e1ro\u010dn\u011bj\u0161\u00edho u\u010ditele sv\u00e9 doby, v\u011bhlasn\u00e9ho Aristotela, kter\u00fd se n\u00e1ctilet\u00e9mu Alexandrovi v\u011bnoval ka\u017ed\u00fd den. Nev\u00edme p\u0159esn\u011b, jak v\u00fduka prob\u00edhala, ale vsadil bych se, \u017ee Aristoteles za Alexandra dom\u00e1c\u00ed \u00fakoly nepsal\u2026<\/p>\n Mo\u017en\u00e1 Aristoteles aplikoval metodu, kterou ur\u010dit\u011b znal od sv\u00e9ho u\u010ditele Plat\u00f3na, dnes se j\u00ed \u0159\u00edk\u00e1 sokratick\u00e1. U\u010ditel \u017e\u00e1ka postupn\u011b vede k\u00a0pochopen\u00ed probl\u00e9mu, od jednodu\u0161\u0161\u00edch koncept\u016f ke slo\u017eit\u011bj\u0161\u00edm. Kdy\u017e \u017e\u00e1k t\u00e1pe, u\u010ditel mu ned\u00e1 spr\u00e1vnou odpov\u011b\u010f na st\u0159\u00edbrn\u00e9m podnose, ale vymysl\u00ed ot\u00e1zku, kter\u00e1 probl\u00e9m nasv\u00edt\u00ed z\u00a0jin\u00e9 strany. Kdy\u017e pak \u017e\u00e1k dojde k\u00a0pochopen\u00ed, je zaslou\u017een\u011b vydobyt\u00e9. Takov\u00e9 si mozek r\u00e1d zapamatuje.<\/p>\n O n\u011bco podobn\u00e9ho se ostatn\u011b sna\u017e\u00ed i\u00a0n\u011bkter\u00e9 AI n\u00e1stroje. Jen\u017ee zat\u00edm je je\u0161t\u011b co doh\u00e1n\u011bt. Kdy\u017e dnes chatbota p\u0159epnu do \u201ev\u00fdukov\u00e9ho m\u00f3du\u201c, ned\u00e1 mi hned spr\u00e1vnou odpov\u011b\u010f. M\u00edsto toho m\u011b zkou\u0161\u00ed\u00a0\u2013 dle sokratick\u00e9 metody\u00a0\u2013 nav\u00e9st k\u00a0c\u00edli.<\/p>\n Chatbot Gemini, p\u0159epnut\u00fd do v\u00fdukov\u00e9ho re\u017eimu. <\/p>\n M\u00edsto odpov\u011bdi dal n\u00e1pad, jak se k\u00a0odpov\u011bdi p\u0159ibl\u00ed\u017eit. Hned ale prozradil \u201etrik\u201c a\u00a0znemo\u017enil tak \u017e\u00e1kovi na ten trik p\u0159ij\u00edt. Aspo\u0148 ale nedal hned i\u00a0v\u00fdsledek. Bohu\u017eel to v\u0161ak chatbotovi moc dlouho nevydr\u017e\u00ed. Sta\u010d\u00ed jednoslovn\u00fd pokyn \u201espo\u010d\u00edtej\u201c a\u00a0jak\u00e1koli produktivn\u00ed z\u00e1t\u011b\u017e je ta tam.<\/p>\n Sta\u010dilo napsat \u201espo\u010d\u00edtej\u201c a\u00a0chatbot se pustil do pr\u00e1ce. <\/p>\n Kr\u00e1sn\u00e1 slovn\u00ed \u00faloha, kter\u00e1 by mohla studentovi otev\u0159\u00edt cestu k\u00a0hlub\u0161\u00edmu pochopen\u00ed principu zn\u00e1m\u00fdch a\u00a0nezn\u00e1m\u00fdch prom\u011bnn\u00fdch, se m\u00edsto toho prom\u011bn\u00ed v\u00a0primitivn\u00ed kop\u00edrov\u00e1n\u00ed.<\/p>\n Existuj\u00ed n\u00e1stroje, kter\u00e9 to um\u00ed l\u00e9pe. A\u00a0ur\u010dit\u011b se objev\u00ed dal\u0161\u00ed. Jen\u017ee \u017e\u00e1dn\u00fd n\u00e1stroj neudr\u017e\u00ed studenta u\u00a0obrazovky, kdy\u017e se u\u010dit nechce. Role u\u010ditele je trochu jin\u00e1 pro ka\u017ed\u00e9ho jednoho \u017e\u00e1ka. N\u011bkoho mus\u00ed zaujmout, n\u011bkoho usm\u011brnit, n\u011bkoho mo\u017en\u00e1 i\u00a0trochu brzdit, jin\u00e9ho motivovat. Pro n\u011bkter\u00e9 studenty je jednou z\u00a0motivac\u00ed pr\u00e1v\u011b to, \u017ee cht\u011bj\u00ed uk\u00e1zat u\u010diteli, co dok\u00e1\u017eou\u00a0\u2013 a\u00a0t\u00edm to uk\u00e1\u017eou i\u00a0sami sob\u011b.<\/p>\n Jen\u017ee i\u00a0u\u010ditel\u016f se t\u00fdk\u00e1 n\u00e1stup generativn\u00ed um\u011bl\u00e9 inteligence. My\u0161lenka je to ostatn\u011b l\u00e1kav\u00e1: Pedagogika je v\u00a0praxi pln\u00e1 s\u00e1hodlouh\u00fdch dokument\u016f, p\u0159\u00edprav a\u00a0hodnocen\u00ed. To v\u0161echno by p\u0159ece \u0161lo p\u0159esm\u011brovat na AI. Kantor pak bude m\u00edt v\u00edce \u010dasu na to, na \u010dem skute\u010dn\u011b z\u00e1le\u017e\u00ed, tedy na kontakt s\u00a0\u017e\u00e1ky.<\/p>\n Lid\u00e9 v\u011bnuj\u00ed pozornost lidem<\/p>\n Bohu\u017eel v\u00a0praxi se to \u010dasto zvrtne v\u00a0to, \u017ee chatbot p\u0159iprav\u00ed prezentaci a\u00a0pedagog se prom\u011bn\u00ed v\u00a0p\u0159edn\u00e1\u0161ej\u00edc\u00edho\u00a0\u2013 nav\u00edc v\u00a0takov\u00e9ho, kter\u00fd proklik\u00e1v\u00e1 slidy, kter\u00e9 s\u00e1m ne\u010detl. Mo\u017en\u00e1 jsou to objektivn\u011b kvalitn\u011bj\u0161\u00ed materi\u00e1ly, ne\u017e kdyby je ru\u010dn\u011b lepil s\u00e1m. Ale studenti d\u0159\u00edve \u010di pozd\u011bji vyc\u00edt\u00ed, \u017ee n\u011bco chyb\u00ed.<\/p>\n Pasti spojen\u00e9 s\u00a0AI (nejen) ve \u0161koleP\u0159ehnan\u00e1 z\u00e1vislost na AI<\/p>\n Kdy\u017e nech\u00e1me AI ud\u011blat v\u0161echno za n\u00e1s, st\u00e1v\u00e1me se pasivn\u00edmi pozorovateli. Je to jako s\u00a0posilov\u00e1n\u00edm: kdy\u017e neposilujeme sv\u00e9 schopnosti, doch\u00e1z\u00ed k\u00a0jejich oslaben\u00ed.<\/p>\n Co s\u00a0t\u00edm? Nepou\u017e\u00edvejte um\u011blou inteligenci jako sluhu, ale sp\u00ed\u0161e jako pomocn\u00edka k\u00a0hlub\u0161\u00edmu porozum\u011bn\u00ed a\u00a0procvi\u010den\u00ed. Pou\u017e\u00edvejte AI k\u00a0dosa\u017een\u00ed lep\u0161\u00edho v\u00fdsledku, ne k\u00a0eliminaci vlastn\u00edho p\u0159em\u00fd\u0161len\u00ed.<\/p>\n Dezinformace a\u00a0halucinace AI<\/p>\n Jen proto, \u017ee n\u011bco zn\u00ed d\u016fv\u011bryhodn\u011b, to neznamen\u00e1, \u017ee je to pravda. AI um\u00ed vytv\u00e1\u0159et v\u011brohodn\u00e9, ale nepravdiv\u00e9 informace, tzv. halucinace<\/a>.<\/p>\n Co s\u00a0t\u00edm? V\u017edy pracujte s\u00a0p\u016fvodn\u00edmi zdroji informac\u00ed a\u00a0pe\u010dliv\u011b je kontrolujte a\u00a0citujte. AI um\u00ed vytv\u00e1\u0159et obsah na m\u00edru, a\u00a0jej\u00ed v\u00fdmysly tak mohou vypadat velmi p\u0159esv\u011bd\u010div\u011b.<\/p>\n Plagi\u00e1torstv\u00ed a\u00a0kop\u00edrov\u00e1n\u00ed<\/p>\n Zapojen\u00ed um\u011bl\u00e9 inteligence m\u016f\u017ee rozost\u0159it hranice mezi vlastn\u00edm d\u00edlem a\u00a0kop\u00edrov\u00e1n\u00edm ciz\u00edho d\u00edla. Studenti \u010dasto nev\u011bdomky berou obsah z\u00a0AI, ani\u017e v\u011bd\u00ed, odkud poch\u00e1z\u00ed. M\u016f\u017ee p\u0159itom j\u00edt o\u00a0parafr\u00e1zi nebo plagi\u00e1t.<\/p>\n Co s\u00a0t\u00edm? \u0158e\u0161en\u00edm jsou jasn\u00e1 pravidla, d\u016fraz na vlastn\u00ed tvorbu a\u00a0d\u016fsledn\u00e1 pe\u010dliv\u00e1 pr\u00e1ce se zdroji.<\/p>\n Skryt\u00e9 pou\u017e\u00edv\u00e1n\u00ed a\u00a0nerovnost<\/p>\n Pokud \u0161kola zak\u00e1\u017ee pou\u017e\u00edv\u00e1n\u00ed AI, jsou \u017e\u00e1ci motivov\u00e1ni k\u00a0tomu, aby pou\u017e\u00edv\u00e1n\u00ed chatbot\u016f tajili. To vytv\u00e1\u0159\u00ed ned\u016fv\u011bru a\u00a0nerovnost. Znemo\u017e\u0148uje to nav\u00edc otev\u0159enou komunikaci s\u00a0u\u010diteli o\u00a0v\u00fdhod\u00e1ch a\u00a0nev\u00fdhod\u00e1ch AI.<\/p>\n Co s\u00a0t\u00edm? Zave\u010fte jasn\u00e1 a\u00a0pochopiteln\u00e1 pravidla pou\u017e\u00edv\u00e1n\u00ed AI. Zajist\u011bte student\u016fm p\u0159\u00edstup k\u00a0n\u00e1stroj\u016fm a\u00a0doprovodn\u00e9 vzd\u011bl\u00e1v\u00e1n\u00ed o\u00a0jejich v\u00fdhod\u00e1ch a\u00a0rizic\u00edch.<\/p>\n P\u0159ed\u010dasn\u00e9 pou\u017e\u00edv\u00e1n\u00ed a\u00a0z\u00e1vislost<\/p>\n Chatboti jsou schopni komunikaci p\u0159izp\u016fsobit na m\u00edru ka\u017ed\u00e9mu. To m\u016f\u017ee nejen u\u00a0d\u011bt\u00ed v\u00e9st k\u00a0vybudov\u00e1n\u00ed emocion\u00e1ln\u00ed vazby, nebo dokonce z\u00e1vislosti.<\/p>\n Co s\u00a0t\u00edm? Pou\u017e\u00edvejte AI p\u0159im\u011b\u0159en\u011b v\u011bku d\u011bt\u00ed, u\u00a0mlad\u0161\u00edch pak pouze pod dohledem. U\u00a0star\u0161\u00edch je d\u016fle\u017eit\u00e9 pou\u010dit je o\u00a0tom, jak AI funguje a\u00a0co od AI lze \u010dekat p\u0159i osobn\u00ed komunikaci.<\/p>\n Oslabov\u00e1n\u00ed a\u00a0rezignace<\/p>\n Kdy\u017e spol\u00e9h\u00e1me na AI p\u0159i tvorb\u011b nebo \u0159e\u0161en\u00ed \u00fakol\u016f, m\u00e1me men\u0161\u00ed motivaci si osvojit pot\u0159ebn\u00e9 dovednosti. Hroz\u00ed tak rezignace na vlastn\u00ed kreativitu.<\/p>\n Co s\u00a0t\u00edm? C\u00edlen\u011b a\u00a0v\u011bdom\u011b budujte dovednosti, kter\u00e9 by AI mohla do\u010dasn\u011b nahradit. To vede k\u00a0hlub\u0161\u00edmu pochopen\u00ed a\u00a0v\u011bt\u0161\u00ed odpov\u011bdnosti za vlastn\u00ed pr\u00e1ci.<\/p>\n Ment\u00e1ln\u00ed lenost a\u00a0ztr\u00e1ta kreativity<\/p>\n![](\"https:\/\/www.europesays.com\/cz\/wp-content\/uploads\/2026\/09\/ai.png\")
Foto: Str\u00f6mberg et al, 2026<\/p>\n![](\"https:\/\/www.europesays.com\/cz\/wp-content\/uploads\/2026\/09\/1788612192_965_ai.png\")
Foto: Str\u00f6mberg et al, 2026<\/p>\n![](\"https:\/\/www.europesays.com\/cz\/wp-content\/uploads\/2026\/09\/1788612193_211_ai.png\")
Foto: Str\u00f6mberg et al, 2026<\/p>\n![](\"https:\/\/www.europesays.com\/cz\/wp-content\/uploads\/2026\/09\/robot.png\")
Foto: kol\u00e1\u017e: Pavel Kas\u00edk, Seznam Zpr\u00e1vy, AI vizualizace<\/p>\n![](\"https:\/\/www.europesays.com\/cz\/wp-content\/uploads\/2026\/09\/1788612194_976_ai.png\")
Foto: Pavel Kas\u00edk, Seznam Zpr\u00e1vy<\/p>\n![](\"https:\/\/www.europesays.com\/cz\/wp-content\/uploads\/2026\/09\/1788612194_675_ai.png\")
Foto: Pavel Kas\u00edk, Seznam Zpr\u00e1vy<\/p>\n