{"id":29097,"date":"2023-10-18T12:40:09","date_gmt":"2023-10-18T10:40:09","guid":{"rendered":"https:\/\/kinit.sk\/how-can-ai-destroy-civilization-and-what-can-we-do-to-prevent-it\/"},"modified":"2023-10-18T12:48:02","modified_gmt":"2023-10-18T10:48:02","slug":"ako-moze-umela-inteligencia-znicit-civilizaciu-a-co-mozeme-urobit-aby-sme-tomu-predisli","status":"publish","type":"post","link":"https:\/\/kinit.sk\/sk\/ako-moze-umela-inteligencia-znicit-civilizaciu-a-co-mozeme-urobit-aby-sme-tomu-predisli\/","title":{"rendered":"Ako m\u00f4\u017ee umel\u00e1 inteligencia zni\u010di\u0165 civiliz\u00e1ciu a \u010do m\u00f4\u017eeme urobi\u0165, aby sme tomu predi\u0161li?"},"content":{"rendered":"
\n

V\u00fdchodoeur\u00f3pska letn\u00e1 \u0161kola strojov\u00e9ho u\u010denia (EEML)<\/em><\/a> je prest\u00ed\u017ena udalos\u0165, ktor\u00e1 sp\u00e1ja mlad\u00fdch v\u00fdskumn\u00edkov a odborn\u00edkov v oblasti umelej inteligencie zo strednej a v\u00fdchodnej Eur\u00f3py. T\u00e1to letn\u00e1 \u0161kola je jedine\u010dnou pr\u00edle\u017eitos\u0165ou u\u010di\u0165 sa od svetov\u00fdch expertov z r\u00f4znych discipl\u00edn strojov\u00e9ho u\u010denia. Tento rok sa \u0161kola konala v Ko\u0161iciach a <\/em>z\u00fa\u010dastnili sa aj na\u0161i doktorandi.<\/em><\/a> V tomto \u010dl\u00e1nku sa dozviete viac o rizik\u00e1ch umelej inteligencie, ktor\u00e9 predstavil priekopn\u00edk hlbok\u00e9ho u\u010denia Yoshua Bengio, a ktor\u00e9 n\u00e1sledne zhrnuli na\u0161i doktorandi <\/em>Peter Pavl\u00edk<\/em><\/a> a <\/em>Matej \u010cief<\/em><\/a>.<\/em><\/p>\n<\/div>\n\n

\n

Ke\u010f\u017ee je Yoshua Bengio<\/a> pova\u017eovan\u00fd za jednu z najd\u00f4le\u017eitej\u0161\u00edch post\u00e1v v oblasti umelej inteligencie (AI) a hlbok\u00e9ho u\u010denia, mnoh\u00ed \u00fa\u010dastn\u00edci letnej \u0161koly o\u010dak\u00e1vali, \u017ee jeho predn\u00e1\u0161ka bude v stredu vrcholom programu. Mnoh\u00ed v\u0161ak boli prekvapen\u00ed t\u00e9mou, o ktorej chcel hovori\u0165.<\/p>\n<\/div>\n\n

\n

Bengiova predn\u00e1\u0161ka za\u010dala v ponurom duchu: „V\u00e4\u010d\u0161inu mojej kari\u00e9ry som pracoval na tom ako zlep\u0161i\u0165 svet pomocou umelej inteligencie. Posledn\u00e9 mesiace v\u0161ak rozm\u00fd\u0161\u013eam nad t\u00fdm, \u010do v\u0161etko sa m\u00f4\u017ee pokazi\u0165, ak sa stane pr\u00edli\u0161 pokro\u010dilou.“<\/em> Je jasn\u00e9, \u010do sp\u00f4sobilo t\u00fato zmenu myslenia. Hoci v\u00fdzvy smerova\u0165 zdroje na bezpe\u010dnos\u0165 umelej inteligencie tu s\u00fa u\u017e desa\u0165ro\u010dia, jej rizik\u00e1 nikdy neboli tak\u00e9 hmatate\u013en\u00e9, ako s\u00fa teraz.<\/p>\n<\/div>\n\n

\n

V septembri 2022 bol cel\u00fd svet v \u0161oku, ke\u010f OpenAI uviedla svoj model DALL-E 2<\/a>, ktor\u00fd umo\u017enil ka\u017ed\u00e9mu pomocou umelej inteligencie\u00a0 generova\u0165 umenie len jednoduch\u00fdm pokynom. Model priniesol tvorcom mnoho pr\u00edle\u017eitost\u00ed, ale z\u00edskal aj kritiku od mnoh\u00fdch umelcov. Dovtedy pred verejnos\u0165ou skryt\u00e9 schopnosti generat\u00edvnych modelov umelej inteligencie sa stali v\u0161eobecne zn\u00e1mymi.<\/p>\n<\/div>\n\n

\n

Av\u0161ak e\u0161te v\u00e4\u010d\u0161ie odhalenie pri\u0161lo kr\u00e1tko na to. V novembri 2022 bol predstaven\u00fd ve\u013ek\u00fd jazykov\u00fd model ChatGPT<\/a>. Tento chatbot zalo\u017een\u00fd na umelej inteligencii dok\u00e1zal presved\u010divo napodob\u0148ova\u0165 \u013eudsk\u00e9 konverz\u00e1cie, mal rozsiahlu znalos\u0165 mnoh\u00fdch t\u00e9m, vedel p\u00edsa\u0165 programy, hudbu, b\u00e1sne a, do istej miery, dokonca s\u00e1m generova\u0165 nov\u00e9 n\u00e1pady.\u00a0 Model, ktor\u00fd bol schopn\u00fd vies\u0165 autentick\u00e9 konverz\u00e1cie a rie\u0161i\u0165 \u00falohy zmenil poh\u013ead mnoh\u00fdch \u013eud\u00ed na bud\u00facnos\u0165 umelej inteligencie. Dokonca sa zdalo, \u017ee n\u00e1s priv\u00e1dza bli\u017e\u0161ie k m\u00fdtickej AGI<\/a> – v\u0161eobecnej umelej inteligencii.<\/p>\n<\/div>\n\n

\n

Pozn\u00e1me ve\u013ea sci-fi pr\u00edbehov, ktor\u00e9 popisuj\u00fa nad\u013eudsk\u00fa AI, ktor\u00e1 sa vzb\u00farila a zni\u010dila na\u0161u civiliz\u00e1ciu. Tento probl\u00e9m v\u0161ak mus\u00edme posudzova\u0165 z vedeck\u00e9ho h\u013eadiska. Je potrebn\u00e9 zhodnoti\u0165 v\u0161etky mo\u017en\u00e9 rizik\u00e1, ktor\u00e9 so sebou t\u00e1to technol\u00f3gia prin\u00e1\u0161a a vyvin\u00fa\u0165 tak\u00e9 pravidl\u00e1 bezpe\u010dnosti AI, ktor\u00e9 tieto rizik\u00e1 minimalizuj\u00fa.<\/strong> Nehovor\u00edme tu o nere\u00e1lnom sci-fi scen\u00e1ri, \u013eudsk\u00fd mozog je tie\u017e iba biologick\u00fd stroj a pokia\u013e vieme, ni\u010d n\u00e1m nebr\u00e1ni v bud\u00facnosti toto jeho spr\u00e1vanie replikova\u0165.<\/p>\n<\/div>\n\n

\n

Existuj\u00fa n\u00e1zory, ktor\u00e9 tvrdia, \u017ee fantaz\u00edrovanie o bud\u00facnosti umelej inteligencie je bezpredmetn\u00e9 a rizik\u00e1 budeme rie\u0161i\u0165, a\u017e ke\u010f nastan\u00fa. „\u013dudstvo v\u017edy n\u00e1jde rie\u0161enie.“<\/em> Nemali by sme v\u0161ak pri tak\u00fdchto obrovsk\u00fdch rizik\u00e1ch rad\u0161ej urobi\u0165 v\u0161etko, \u010do je v na\u0161ich sil\u00e1ch, aby sme sa vyhli katastrofick\u00fdm d\u00f4sledkom?<\/p>\n<\/div>\n\n

\n

Pr\u00edchod umelej inteligencie m\u00f4\u017ee ma\u0165 mnoho negat\u00edvnych n\u00e1sledkov. S prv\u00fdm rizikom sa u\u017e teraz stret\u00e1vame – riziko zneu\u017eitia<\/strong>. O tomto si povieme ve\u013emi stru\u010dne. Z h\u013eadiska bezpe\u010dnosti AI n\u00e1s ove\u013ea viac zauj\u00edma strata kontroly nad syst\u00e9mami z d\u00f4vodu nespr\u00e1vneho nastavenia cie\u013eov a hodn\u00f4t.<\/strong><\/p>\n<\/div>\n\n

\n

Mus\u00edme presne \u0161pecifikova\u0165, \u010do v tomto kontexte mysl\u00edme pod pojmom AI. Existuje mnoho \u0161pecializovan\u00fdch modelov AI, ktor\u00e9 s\u00fa tr\u00e9novan\u00e9 na rie\u0161enie konkr\u00e9tnej \u00falohy, napr\u00edklad na asistenciu pri predpovedi po\u010dasia alebo pri lek\u00e1rskej diagnostike. V skuto\u010dnosti nie je \u017eiadne riziko, \u017ee by sme nad tak\u00fdmto modelom stratili kontrolu a sp\u00f4sobil \u0161kodu. Bez schopnosti pochopenia \u013eudskej spolo\u010dnosti a mo\u017enosti p\u00f4sobi\u0165 na re\u00e1lny svet je tak\u00fdto model len n\u00e1strojom a nepredstavuje pre \u013eudstvo \u017eiadne riziko. Rizik\u00e1 prich\u00e1dzaj\u00fa s generalistick\u00fdmi modelmi, ktor\u00e9 s\u00fa tr\u00e9novan\u00e9 na porozumenie \u013eudskej re\u010di. Tieto modely m\u00f4\u017eu samostatne plni\u0165 ciele a optimalizova\u0165 \u00falohy.  <\/strong><\/p>\n<\/div>\n\n

\n

Tak\u00e9to generalistick\u00e9 modely s\u00edce v s\u00fa\u010dasnosti neexistuj\u00fa, no ve\u013ek\u00e9 jazykov\u00e9 modely sa im u\u017e dos\u0165 pribli\u017euj\u00fa. <\/strong>Napr\u00edklad, v septembri 2023 dostal Google Bard funkcionalitu prip\u00e1ja\u0165 sa k in\u00fdm aplik\u00e1ci\u00e1m a slu\u017eb\u00e1m Google, aby mohol vykon\u00e1va\u0165 na\u0161e po\u017eiadavky. V tomto bode u\u017e  nie je tak \u0165a\u017ek\u00e9 si predstavi\u0165 model, ktor\u00fd bude posiela\u0165 \u013eu\u010fom e-maily z d\u00f4vodu dosahovania vlastn\u00fdch cie\u013eov a ovplyv\u0148ova\u0165 tak re\u00e1lny svet.<\/p>\n<\/div>\n\n

\n

Pre\u010do by n\u00e1s to malo zauj\u00edma\u0165?<\/h3>\n<\/div>\n\n
\n

Rizik\u00e1 spojen\u00e9 s umelou inteligenciou, predov\u0161etk\u00fdm mo\u017enos\u0165 straty kontroly kv\u00f4li nespr\u00e1vnemu nastaveniu cie\u013eov alebo hodn\u00f4t, by mali by\u0165 ve\u013ekou obavou pre spolo\u010dnos\u0165. Mus\u00edme pochopi\u0165, pre\u010do s\u00fa tieto rizik\u00e1 v\u00fdznamn\u00e9, nako\u013eko to ovplyn\u0148uje n\u00e1\u0161 postoj k v\u00fdvoju a pou\u017e\u00edvaniu umelej inteligencie.<\/p>\n<\/div>\n\n

\n

Po prv\u00e9, tieto rizik\u00e1 s\u00fa d\u00f4le\u017eit\u00e9, preto\u017ee by mohli sp\u00f4sobi\u0165 v\u00fdrazn\u00e9 probl\u00e9my v spolo\u010dnosti<\/strong>. S rozvojom umelej inteligencie sa m\u00f4\u017ee meni\u0165 sp\u00f4sob fungovania podnikov, ekonomiky a aj dostupnos\u0165 pracovn\u00fdch miest. Tieto zmeny m\u00f4\u017eu by\u0165 prospe\u0161n\u00e9 – napr\u00edklad t\u00fdm, \u017ee prispej\u00fa k zv\u00fd\u0161eniu efekt\u00edvnosti a kreativity.  Nesmieme v\u0161ak zab\u00fada\u0165 na to, \u017ee m\u00f4\u017eu vies\u0165 k strate pracovn\u00fdch miest a nerovnomern\u00e9mu rozde\u013eovaniu bohatstva. Na tak\u00e9to zmeny sa mus\u00edme pripravi\u0165, aby sme zabezpe\u010dili, \u017ee bud\u00fa spravodliv\u00e9.<\/strong>  <\/p>\n<\/div>\n\n

\n

Rizik\u00e1 sa v\u0161ak net\u00fdkaj\u00fa iba pe\u0148az\u00ed a pracovn\u00fdch miest. Niektor\u00e9 syst\u00e9my AI (napr. tie, ktor\u00e9 dok\u00e1\u017eu porozumie\u0165 a komunikova\u0165 ako \u013eudia) s\u00fa schopn\u00e9 vykon\u00e1va\u0165 samostatn\u00e9 rozhodnutia. Vyvol\u00e1va to d\u00f4le\u017eit\u00e9 ot\u00e1zky, \u010do je spr\u00e1vne a nespr\u00e1vne. Bez poriadnych pravidiel m\u00f4\u017eu by\u0165 syst\u00e9my AI zneu\u017eit\u00e9 na sledovanie \u013eud\u00ed, m\u00f4\u017eu s niek\u00fdm nespravodlivo zaobch\u00e1dza\u0165 alebo dokonca niekomu ubl\u00ed\u017ei\u0165.<\/strong> Predstavte si, \u017ee by syst\u00e9m AI mohol manipulova\u0165 s \u013eu\u010fmi prostredn\u00edctvom zasielania podvodn\u00fdch alebo inak \u0161kodi\u0165 spolo\u010dnosti bez toho, aby si uvedomoval, \u017ee sp\u00f4sobuje \u0161kodu.  <\/p>\n<\/div>\n\n

\n

To, \u017ee tieto rizik\u00e1 neberieme v\u00e1\u017ene, m\u00f4\u017ee znamena\u0165, \u017ee nikto nechce prevzia\u0165 zodpovednos\u0165 v pr\u00edpade, \u017ee sa nie\u010do pokaz\u00ed. Ak syst\u00e9my AI s vlastn\u00fdmi rozhodovac\u00edmi pr\u00e1vomocami sp\u00f4sobia \u0161kodu, m\u00f4\u017ee by\u0165 ve\u013emi \u0165a\u017ek\u00e9 zisti\u0165, kto je za to zodpovedn\u00fd. <\/strong>Vo v\u00fdsledku by to mohlo sp\u00f4sobi\u0165 stratu d\u00f4very v AI spomali\u0165 jej bezpe\u010dn\u00e9 vyu\u017e\u00edvanie.<\/p>\n<\/div>\n\n

\n

V kr\u00e1tkosti povedan\u00e9, rizik\u00e1 spojen\u00e9 s AI – predov\u0161etk\u00fdm tie, ktor\u00e9 s\u00favisia so stratou kontroly a s odli\u0161n\u00fdmi hodnotami a cie\u013emi, ako maj\u00fa \u013eudia –  nie s\u00fa len v teoretickej rovine. Tieto rizik\u00e1 m\u00f4\u017eu zmeni\u0165 na\u0161u spolo\u010dnos\u0165, na\u0161e ka\u017edodenn\u00e9 \u017eivoty a v\u0161etko, \u010do je pre n\u00e1s d\u00f4le\u017eit\u00e9. Preto ich nesmieme bra\u0165 na \u013eahk\u00fa v\u00e1hu a mus\u00edme by\u0165 vo v\u00fdvoji AI zodpovedn\u00ed a stanovi\u0165 pravidl\u00e1.<\/strong> Nie je mo\u017en\u00e9 tieto rizik\u00e1 ignorova\u0165, preto\u017ee ne\u010dinnos\u0165 by mohla ma\u0165 ve\u013emi v\u00e1\u017ene n\u00e1sledky. <\/p>\n<\/div>\n\n

\n

Hodnotenie pravdepodobnosti riz\u00edk generat\u00edvnej AI<\/h3>\n<\/div>\n\n
\n

Predt\u00fdm, ako sa hlb\u0161ie pozrieme na d\u00f4vody, pre\u010do by sme mali bra\u0165 v\u00e1\u017ene rizik\u00e1 spojen\u00e9 s generat\u00edvnou AI, zastavme sa na chv\u00ed\u013eu a zamyslime sa nad t\u00fdm, ak\u00e1 je pravdepodobnos\u0165, \u017ee sa tieto rizik\u00e1 stan\u00fa realitou. T\u00e1to ot\u00e1zka je legit\u00edmna, ke\u010f\u017ee \u010dasto rob\u00edme rozhodnutia na z\u00e1klade pravdepodobnost\u00ed.<\/p>\n<\/div>\n\n

\n

Vo svete umelej inteligencie je \u0165a\u017ek\u00e9 predpoveda\u0165 presn\u00fa bud\u00facnos\u0165. Nem\u00f4\u017eeme s absol\u00fatnou istotou poveda\u0165, \u017ee generat\u00edvna AI sa zvrtne a sp\u00f4sob\u00ed \u0161kodu. Jej potenci\u00e1lne d\u00f4sledky s\u00fa v\u0161ak tak v\u00e1\u017ene, \u017ee aj mal\u00e1 pravdepodobnos\u0165 katastrof\u00e1lnej udalosti by mala by\u0165 pre n\u00e1s varovan\u00edm. Nemali by sme rizik\u00e1 ignorova\u0165 iba preto, \u017ee nie s\u00fa \u00faplne ist\u00e9. Namiesto toho mus\u00edme zv\u00e1\u017ei\u0165 potenci\u00e1lne n\u00e1sledky s ich pravdepodobnos\u0165ou.<\/p>\n<\/div>\n\n

\n

V\u00fdskumn\u00edci a odborn\u00edci z oblasti bezpe\u010dnosti umelej inteligencie si uvedomuj\u00fa n\u00e1ro\u010dnos\u0165 priradenia presn\u00fdch pravdepodobnost\u00ed tak\u00fdmto scen\u00e1rom. Potenci\u00e1lne rizik\u00e1 spojen\u00e9 s AGI s\u00fa \u010dasto ozna\u010dovan\u00e9 ako „unknown unknowns“. To znamen\u00e1, \u017ee existuj\u00fa rizik\u00e1, o ktor\u00fdch mo\u017eno e\u0161te nevieme. Z t\u00fdchto d\u00f4vodov sa diskusie o rizik\u00e1ch a bezpe\u010dnosti umelej inteligencie \u010dasto s\u00fastre\u010fuj\u00fa na potenci\u00e1lnu z\u00e1va\u017enos\u0165 d\u00f4sledkov, nie na presn\u00e9 pravdepodobnosti.<\/strong><\/p>\n<\/div>\n\n

\n

Investovanie do v\u00fdskumu bezpe\u010dnosti AI: Rozumn\u00e1 vo\u013eba<\/h3>\n<\/div>\n\n
\n

Teraz sa pozrime na d\u00f4vody, pre\u010do m\u00e1 zmysel investova\u0165 do v\u00fdskumu bezpe\u010dnosti AI. A to aj napriek tomu, \u017ee nevieme presne odmera\u0165 pravdepodobnos\u0165, \u017ee sa stane nie\u010do zl\u00e9. Predstavte si, \u017ee hr\u00e1te hru, v ktorej m\u00f4\u017eete bu\u010f z\u00edska\u0165 ve\u013ek\u00fa odmenu, alebo \u010deli\u0165 mal\u00e9mu ale potenci\u00e1lne \u017eivot ohrozuj\u00facemu riziku. Hrali by ste t\u00fato hru, ak by ste nevedeli pravdepodobnos\u0165? V\u00e4\u010d\u0161ina z n\u00e1s nie.<\/p>\n<\/div>\n\n

\n

Tu vstupuje do hry koncept pripravenosti a v\u00fdskumu bezpe\u010dnosti AI. Investovan\u00edm do v\u00fdskumu bezpe\u010dnosti si prakticky kupujeme poistenie proti rizik\u00e1m spojen\u00fdm s generat\u00edvnou AI. <\/strong>Zais\u0165ujeme sa, podobne ako pri k\u00fape poistenia pre na\u0161e domy alebo aut\u00e1. Aj ke\u010f je pravdepodobnos\u0165 katastrofy n\u00edzka, potenci\u00e1lne d\u00f4sledky s\u00fa tak\u00e9 z\u00e1va\u017en\u00e9, \u017ee je rozumn\u00e9 investova\u0165 do bezpe\u010dnostn\u00fdch opatren\u00ed na minimalizovanie t\u00fdchto riz\u00edk.<\/p>\n<\/div>\n\n

\n

Rusk\u00e1 ruleta<\/h3>\n<\/div>\n\n
\n

Pozrime sa na pr\u00edklad ruskej rulety, nebezpe\u010dnej hry s nabitou zbra\u0148ou. Nikto so zdrav\u00fdm rozumom by t\u00fato hru nehral, aj keby veril, \u017ee existuje iba mal\u00e1 pravdepodobnos\u0165, \u017ee zbra\u0148 v\u00fdstrel\u00ed. D\u00f4sledky s\u00fa jednoducho pr\u00edli\u0161 v\u00e1\u017ene. Podobne ako rusk\u00fa ruletu, by sme nemali hra\u0165 ani hru s rizikami spojen\u00fdmi s generat\u00edvnou AI. Namiesto toho by sme mali uprednostni\u0165 bezpe\u010dnos\u0165 a v\u00fdskum, aby sme zabezpe\u010dili, \u017ee aj ke\u010f nepozn\u00e1me presn\u00e9 pravdepodobnosti riz\u00edk, nehrozia n\u00e1m katastrofick\u00e9 d\u00f4sledky.<\/strong><\/p>\n<\/div>\n\n

\n

Na z\u00e1ver m\u00f4\u017eeme zhrn\u00fa\u0165, \u017ee hoci nem\u00f4\u017eeme presne predpoveda\u0165 bud\u00facnos\u0165 generat\u00edvnej AI, mus\u00edme s \u0148ou spojen\u00e9 rizik\u00e1 bra\u0165 v\u00e1\u017ene. Investovanie do v\u00fdskumu bezpe\u010dnosti AI je ako k\u00fapa poistenia proti potenci\u00e1lnemu nebezpe\u010denstvu. Ide o rozumn\u00fa vo\u013ebu, ke\u010f\u017ee ne\u010dinnos\u0165 m\u00f4\u017ee ma\u0165 v\u00e1\u017ene d\u00f4sledky.  <\/p>\n<\/div>\n\n

\n

Rizik\u00e1 umelej inteligencie pre spolo\u010dnos\u0165, demokraciu a \u013eudstvo<\/h3>\n<\/div>\n\n
\n

Prv\u00fdm z riz\u00edk, ktor\u00e9 m\u00f4\u017eu zapr\u00ed\u010dini\u0165 negat\u00edvne d\u00f4sledky umelej inteligencie, je mo\u017enos\u0165 zneu\u017eitia generat\u00edvnych modelov na jednoduch\u00e9 vytv\u00e1ranie falo\u0161n\u00fdch spr\u00e1v a propagandy. Modely po\u010d\u00edta\u010dov\u00e9ho videnia u\u017e s\u00fa vyu\u017e\u00edvan\u00e9 na tvorbu tzv. \u201cDeep fakes\u201d. Deep fakes s\u00fa vide\u00e1 \u013eud\u00ed, ktor\u00ed robia alebo hovoria nie\u010do, \u010do po\u0161kodzuje ich reput\u00e1ciu. Modely spracovania prirodzen\u00e9ho jazyka m\u00f4\u017eu by\u0165 pou\u017eit\u00e9 na r\u00fdchle p\u00edsanie dezinforma\u010dn\u00fdch \u010dl\u00e1nkov alebo zamorenie online diskusi\u00ed vygenerovan\u00edm tis\u00edcok pr\u00edspevkov, ktor\u00e9 pretl\u00e1\u010daj\u00fa ur\u010dit\u00fa agendu.<\/p>\n<\/div>\n\n

\n

Hoci tieto rizik\u00e1 samy o sebe nepredstavuj\u00fa existen\u010dn\u00fa hrozbu pre \u013eudstvo, v\u00fdsledkom by mohla by\u0165 v\u00fdrazn\u00e1 strata d\u00f4very vo\u010di inform\u00e1ci\u00e1m na internete – a d\u00f4sledky toho sa nedaj\u00fa predv\u00edda\u0165.<\/strong> Podobne ako odpor\u00fa\u010dacie algoritmy soci\u00e1lnych siet\u00ed, ktor\u00e9 zv\u00fd\u0161ili polariz\u00e1ciu v spolo\u010dnosti, s\u00fa\u010dasn\u00e1 revol\u00facia v oblasti umelej inteligencie nepochybne ovplyvn\u00ed spolo\u010dnos\u0165 a demokraciu.<\/p>\n<\/div>\n\n

\n
\"\"<\/figure>\n<\/div>\n\n
\n

\u010eal\u0161ie riziko, ktor\u00e9 mo\u017eno ozna\u010di\u0165 ako hlavn\u00fd cie\u013e celej oblasti AI bezpe\u010dnosti, je nes\u00falad hodn\u00f4t<\/strong>. Znamen\u00e1 to, \u017ee ciele pou\u017e\u00edvate\u013eov nie s\u00fa v s\u00falade s cie\u013emi umelej inteligencie. Bengio naz\u00fdva s\u00falad hodn\u00f4t ako zmluvu medzi pou\u017e\u00edvate\u013eom a modelom. \u013dudia nie\u010do chc\u00fa a AI by to mala poskytn\u00fa\u0165. \u00daspe\u0161n\u00e9 vytvorenie tejto zmluvy je v\u0161ak ve\u013emi \u0165a\u017ek\u00e9.<\/p>\n<\/div>\n\n

\n

Ak rozvinieme t\u00fato metaforu, m\u00f4\u017eeme t\u00fato zmluvu prirovna\u0165 k dohode medzi dvoma \u013eu\u010fmi alebo firmami. Ak nie\u010do chcete, \u0161pecifikujete \u010do, kedy, za ak\u00fa cenu, at\u010f. Nemus\u00edte \u0161pecifikova\u0165, \u017ee dod\u00e1vate\u013e nesmie niekoho zavra\u017edi\u0165, aby v\u00e1m to poskytol. Toto vypl\u00fdva z faktu, \u017ee \u017eijeme v spolo\u010dnosti, v ktorej je to zak\u00e1zan\u00e9, a kde existuje r\u00e1mec z\u00e1konov a mor\u00e1lnych pravidiel, ktor\u00fd sa nesmie poru\u0161i\u0165 ako s\u00fa\u010das\u0165 zmluvy.<\/p>\n<\/div>\n\n

\n

Av\u0161ak v pr\u00edpade zml\u00fav medzi \u010dlovekom a AI mus\u00ed by\u0165 ka\u017ed\u00e1 implicitn\u00e1 \u010das\u0165 be\u017enej zmluvy explicitne \u0161pecifikovan\u00e1. V opa\u010dnom pr\u00edpade, v oblasti AGI to m\u00f4\u017ee to vies\u0165 ku katastrof\u00e1lnym d\u00f4sledkom. \u0160v\u00e9dsky filozof Nick Bosom v roku 2003 predstavil sl\u00e1vny my\u0161lienkov\u00fd experiment o umelej inteligencii, ktor\u00e1 maximalizuje v\u00fdrobu papierov\u00fdch spiniek.<\/p>\n<\/div>\n\n

\n

Predstavte si, \u017ee m\u00e1me umel\u00fa inteligenciu, ktorej jedin\u00fdm cie\u013eom je vyrobi\u0165 \u010do najviac papierov\u00fdch spiniek. AI si r\u00fdchlo uvedom\u00ed, \u017ee by bolo ove\u013ea lep\u0161ie, keby tam neboli \u017eiadni \u013eudia, ke\u010f\u017ee \u013eudia by sa mohli rozhodn\u00fa\u0165 ju vypn\u00fa\u0165. Ak by tak \u013eudia urobili, bolo by menej papierov\u00fdch spiniek. \u013dudsk\u00e9 tel\u00e1 tie\u017e obsahuj\u00fa ve\u013ea at\u00f3mov, ktor\u00e9 by mohli by\u0165 premenen\u00e9 na papierov\u00e9 spinky. Bud\u00facnos\u0165, ktor\u00fa by sa AI sna\u017eila dosiahnu\u0165, by bola tak\u00e1, v ktorej by bolo ve\u013ea papierov\u00fdch spiniek, ale \u017eiadni \u013eudia.<\/em><\/p>\n<\/div>\n\n

\n

Bostrom, Nick. „Ethical issues in advanced artificial intelligence.“ Science fiction and philosophy: from time travel to superintelligence (2003): 277-284.<\/em><\/p>\n<\/div>\n\n

\n

Tento my\u0161lienkov\u00fd experiment ukazuje, \u017ee aj pokro\u010dl\u00e1 AGI, ktor\u00e1 pln\u00ed zdanlivo ne\u0161kodn\u00fd cie\u013e, predstavuje z d\u00f4vodu nes\u00faladu hodn\u00f4t existen\u010dn\u00e9 riziko pre \u013eudstvo.<\/p>\n<\/div>\n\n

\n

Na prv\u00fd poh\u013ead to m\u00f4\u017ee vyzera\u0165 ako prehnan\u00fd scen\u00e1r zo sci-fi filmu. A aj je. Probl\u00e9m je v\u0161ak st\u00e1le re\u00e1lny. Ako sa vyjadril britsk\u00fd po\u010d\u00edta\u010dov\u00fd vedec, Stuart Russel:<\/p>\n<\/div>\n\n

\n

„Syst\u00e9m, ktor\u00fd optimalizuje funkciu n premenn\u00fdch, pri\u010dom cie\u013e z\u00e1vis\u00ed na podmno\u017eine k-<\/em>premenn\u00fdch, kde k<n, \u010dasto nastav\u00ed zvy\u0161n\u00e9 neobmedzen\u00e9 premenn\u00e9 na extr\u00e9mne hodnoty. Ak je jedna z t\u00fdchto neobmedzen\u00fdch premenn\u00fdch v skuto\u010dnosti nie\u010do, na \u010dom n\u00e1m z\u00e1le\u017e\u00ed, rie\u0161enie, ktor\u00e9 syst\u00e9m na\u0161iel, m\u00f4\u017ee by\u0165 ve\u013emi ne\u017eiaduce.“<\/em><\/p>\n<\/div>\n\n

\n

V laickej re\u010di, pokra\u010duj\u00fac v anal\u00f3gii so zmluvou: s t\u00fdm, \u010do explicitne neuvedieme do zmluvy AGI (nezahrnieme ako funkciu do modelu), sa m\u00f4\u017eeme rozl\u00fa\u010di\u0165, preto\u017ee to bude odoptimalizovan\u00e9 pre\u010d. „Ide v podstate o star\u00fd pr\u00edbeh o d\u017einovi v lampe, \u010darodejn\u00edkovom u\u010d\u0148ovi, alebo o Kr\u00e1\u013eovi Midasovi: dostanete presne to, \u010do si \u017eiadate, nie to, \u010do chcete,“<\/em> pokra\u010duje Russel. Zos\u00faladenie \u00fa\u010delovej funkcie modelu s \u013eudsk\u00fdmi hodnotami, ktor\u00e9 je takmer nemo\u017en\u00e9 definova\u0165, je monument\u00e1lnou \u00falohou.<\/strong> A to sme e\u0161te ani nezoh\u013eadnili problematiku pudu sebaz\u00e1chovy ak\u00e9hoko\u013evek schopn\u00e9ho inteligentn\u00e9ho syst\u00e9mu.<\/p>\n<\/div>\n\n

\n

Ako to vyjadril Rob Miles v \u201c\u00davod do Bezpe\u010dnosti AI<\/a>\u201d, V\u0161eobecn\u00e1 Umel\u00e1 Inteligencia je nebezpe\u010dn\u00e1 u\u017e vo svojej podstate. Nehovor\u00edme tu o nejakej umelej inteligencii so zl\u00fdmi \u00famyslami alebo o novom, n\u00e1hodne vytvorenom samo-replikuj\u00facom sa druhu. \u0160tandardn\u00e9 spr\u00e1vanie AGI doslova ohrozuje existenciu \u013eudskej civiliz\u00e1cie.<\/p>\n<\/div>\n\n

\n

Ale prejdime od AGI sp\u00e4\u0165 do s\u00fa\u010dasnosti. U\u017e teraz existuj\u00fa desiatky pr\u00edkladov modelov AI, ktor\u00e9 kv\u00f4li nes\u00faladu produkuj\u00fa nepredv\u00eddate\u013en\u00e9 v\u00fdsledky.<\/strong><\/p>\n<\/div>\n\n

\n

Pr\u00edkladom na nes\u00falad hodn\u00f4t v AI n\u00e1stroji je nes\u00falad v n\u00e1stroji na dop\u013a\u0148anie k\u00f3du, ktor\u00fd pou\u017e\u00edva mno\u017estvo \u013eud\u00ed – Github Copilot. Konkr\u00e9tne sa nach\u00e1dza v modeli Codex, na ktorom je Github Copilot zalo\u017een\u00fd. Tento n\u00e1stroj funguje tak, \u017ee nap\u00ed\u0161ete popis \u00fa\u010delu k\u00f3du a AI ho dokon\u010d\u00ed k\u00f3dom, ktor\u00fd rob\u00ed to, \u010do chcete. Samozrejme, cie\u013eom pou\u017e\u00edvate\u013ea je z\u00edska\u0165 kvalitn\u00fd k\u00f3d. Av\u0161ak neexistuje explicitn\u00e1 funkcia \u201ckvality\u201d, ktor\u00fa model optimalizuje. Codex je model predikcie \u010fal\u0161ieho tokenu, jednoducho tr\u00e9novan\u00fd na ve\u013ekom mno\u017estve verejn\u00fdch k\u00f3dov, aby doplnil text k\u00f3du \u010do najpodobnej\u0161ie jeho tr\u00e9novaciemu rozdeleniu.<\/p>\n<\/div>\n\n

\n

Toto vytv\u00e1ra medzeru medzi p\u00f4vodn\u00fdm z\u00e1merom a skuto\u010dne optimalizovan\u00fdm cie\u013eom. M\u00f4\u017ee nasta\u0165 situ\u00e1cia, kedy model vie nap\u00edsa\u0165 dobr\u00fd k\u00f3d, ale rozhodne sa tak neurobi\u0165. Experiment\u00e1lne to bolo preuk\u00e1zan\u00e9 v \u010dl\u00e1nku, ktor\u00fd predstavil Codex tak, \u017ee poskytoval r\u00f4zne zlo\u017eit\u00fdm modelom vstupy obsahuj\u00face jemn\u00e9 chyby v predch\u00e1dzaj\u00facom k\u00f3de. \u010c\u00edm je model komplexnej\u0161\u00ed, t\u00fdm lep\u0161ie dok\u00e1\u017ee napodobni\u0165 rozdelenie k\u00f3du a tak produkova\u0165 k\u00f3d s viacer\u00fdmi chybami ako k\u00f3d bez ch\u00fdb v predch\u00e1dzaj\u00facom k\u00f3de.<\/p>\n<\/div>\n\n

\n
\"\"
Chen, Mark, Jerry Tworek, Heewoo Jun, Qiming Yuan, Henrique Ponde de Oliveira Pinto, Jared Kaplan, Harri Edwards et al. „Evaluating large language models trained on code.“ arXiv preprint arXiv:2107.03374<\/em> (2021).<\/figcaption><\/figure>\n<\/div>\n\n
\n

V tomto pr\u00edpade bol nes\u00falad hodn\u00f4t modelu pova\u017eovan\u00fd za dostato\u010dne dobr\u00fd na skuto\u010dn\u00e9 v\u00fdrobn\u00e9 pou\u017eitie. Hoci nes\u00falad Codexu nie je nebezpe\u010dn\u00fd, ur\u010dite stanovuje precedens a ukazuje, ako v bud\u00facnosti m\u00f4\u017ee nasta\u0165 nes\u00falad v syst\u00e9me, v ktorom m\u00f4\u017ee sp\u00f4sobi\u0165 skuto\u010dn\u00e9 \u0161kody.<\/p>\n<\/div>\n\n

\n

Nakoniec sa zamyslime nad kognit\u00edvnou silou. Kognit\u00edvna sila generat\u00edvnej umelej inteligencie je koncept, ktor\u00fd sa venuje pokro\u010dil\u00fdm schopnostiam syst\u00e9mov umelej inteligencie. Ide o druh sily, ktor\u00fd nenech\u00e1va v\u00fdskumn\u00edkov, etikov a politikov pokojne spa\u0165.<\/strong> Kognit\u00edvna sila sa rozklad\u00e1 na dve k\u013e\u00fa\u010dov\u00e9 dimenzie: v\u0161eobecnos\u0165<\/em> a schopnosti agenta.<\/em><\/p>\n<\/div>\n\n

\n

V\u0161eobecnos\u0165 v umelej inteligencii sa podob\u00e1 viac\u00fa\u010delov\u00e9mu n\u00e1stroju umelej inteligencie. Je to kapacita AI syst\u00e9mov vykon\u00e1va\u0165 \u0161irok\u00fa \u0161k\u00e1lu \u00faloh a plynulo sa prisp\u00f4sobova\u0165 r\u00f4znym oblastiam, podobne ako to robia \u013eudia. Zatia\u013e \u010do \u0161pecializovan\u00e9 AI syst\u00e9my exceluj\u00fa v konkr\u00e9tnych \u00faloh\u00e1ch, v\u0161eobecn\u00e1 AI, \u010dasto ozna\u010dovan\u00e1 ako Umel\u00e1 V\u0161eobecn\u00e1 Inteligencia (AGI), je v\u00fdborn\u00fd multitasker.<\/strong><\/p>\n<\/div>\n\n

\n

Predstavte si AI, ktor\u00e1 dok\u00e1\u017ee nielen ovl\u00e1da\u0165 spracovanie prirodzen\u00e9ho jazyka, ale aj bezprobl\u00e9movo prejs\u0165 na \u00falohy ako rozpozn\u00e1vanie obrazov, hranie hier alebo dokonca diagn\u00f3za chor\u00f4b. \u010co je e\u0161te fascinuj\u00facej\u0161ie je jej schopnos\u0165 u\u010di\u0165 sa z jednej \u00falohy a aplikova\u0165 t\u00fato znalos\u0165 tak, aby len s minim\u00e1lnym \u010fal\u0161\u00edm tr\u00e9novan\u00edm vynikla v inej \u00falohe. T\u00fato super schopnos\u0165 AI naz\u00fdvame Transitional learning<\/em> (prenosov\u00e9 u\u010denie).<\/p>\n<\/div>\n\n

\n

M\u00f4\u017ee sa zda\u0165, \u017ee Transitional learning<\/em> (prenosov\u00e9 u\u010denie) je dokonal\u00fd n\u00e1stroj na efekt\u00edvne vyu\u017eitie schopnost\u00ed umelej inteligencie, ale prich\u00e1dza s vlastn\u00fdm s\u00faborom v\u00fdziev. Pozrime sa, napr\u00edklad, na svet spracovania prirodzen\u00e9ho jazyka. Ke\u010f AI syst\u00e9my vyu\u017e\u00edvaj\u00fa prenosov\u00e9 u\u010denie, m\u00f4\u017eu ne\u00famyselne generova\u0165 toxick\u00fd alebo \u0161kodliv\u00fd jazyk, \u010do vyvol\u00e1va zna\u010dn\u00e9 etick\u00e9 obavy.<\/p>\n<\/div>\n\n

\n

V oblasti rozpozn\u00e1vania obrazov m\u00f4\u017ee prenosov\u00e9 u\u010denie umelej inteligencie vyvola\u0165 ot\u00e1zky t\u00fdkaj\u00face sa ochrany s\u00fakromia. Ak nebud\u00fa AI modely riadne kontrolovan\u00e9, m\u00f4\u017eu na obr\u00e1zkoch rozpozna\u0165 citliv\u00e9 inform\u00e1cie a potenci\u00e1lne poru\u0161i\u0165 pr\u00e1va na ochranu s\u00fakromia.<\/p>\n<\/div>\n\n

\n

V medic\u00ednskom zobrazovan\u00ed, kde m\u00f4\u017ee AI excelova\u0165 v diagnostikovan\u00ed chor\u00f4b, m\u00f4\u017ee nespr\u00e1vne pou\u017eitie prenosov\u00e9ho u\u010denia vies\u0165 k nespr\u00e1vnym diagn\u00f3zam. AI model vy\u0161kolen\u00fd na jednej demografickej skupine nemus\u00ed dosahova\u0165 tak\u00fa presnos\u0165 pri aplik\u00e1cii na in\u00fa popul\u00e1ciu alebo regi\u00f3n.<\/p>\n<\/div>\n\n

\n

Schopnosti agenta, na druhej strane, povy\u0161uj\u00fa AI na \u00faplne nov\u00fa \u00farove\u0148 aut\u00f3nomie. Je to miesto, kde m\u00f4\u017eu AI syst\u00e9my robi\u0165 samostatn\u00e9 rozhodnutia a vykon\u00e1va\u0165 akcie, niekedy dokonca aj bez explicitn\u00fdch pokynov od \u010dloveka. T\u00fato auton\u00f3miu mo\u017eno dosiahnu\u0165 prostredn\u00edctvom procesu naz\u00fdvan\u00e9ho \u201creinforcment learning\u201d, kde sa AI u\u010d\u00ed optimalizova\u0165 svoje \u010dinnosti na z\u00e1klade sp\u00e4tnej v\u00e4zby z prostredia.<\/p>\n<\/div>\n\n

\n

Hoci m\u00f4\u017eu by\u0165 samostatn\u00e9 AI syst\u00e9my neuverite\u013ene u\u017eito\u010dn\u00e9, rovnako m\u00f4\u017eu by\u0165 pri nes\u00falade cie\u013eov aj nebezpe\u010dn\u00e9. Ak ciele AI syst\u00e9mu nezodpovedaj\u00fa \u013eudsk\u00fdm hodnot\u00e1m alebo z\u00e1merom, AI m\u00f4\u017ee robi\u0165 rozhodnutia a \u010dinnosti, ktor\u00e9 s\u00fa v rozpore s na\u0161imi \u017eelaniami.<\/strong><\/p>\n<\/div>\n\n

\n

Predstavte si napr\u00edklad AI model, ktor\u00fd spravuje auton\u00f3mne vozidlo. Ak nie je jeho cie\u013e plne zos\u00faladen\u00fd s bezpe\u010dnos\u0165ou \u013eud\u00ed a etick\u00fdmi z\u00e1sadami, m\u00f4\u017ee robi\u0165 rozhodnutia, ktor\u00e9 d\u00e1vaj\u00fa prednos\u0165 svojim cie\u013eom pred blahom \u013eud\u00ed. Toto m\u00f4\u017ee ma\u0165 potenci\u00e1lne katastrof\u00e1lne n\u00e1sledky.<\/p>\n<\/div>\n\n

\n

Pochopenie kognit\u00edvnej sily generat\u00edvnej umelej inteligencie, ktor\u00e1 zah\u0155\u0148a v\u0161eobecnos\u0165 a schopnosti agenta, odha\u013euje nielen pr\u00ednosy, ale aj rizik\u00e1 pokro\u010dil\u00fdch AI syst\u00e9mov. Hoci tieto schopnosti maj\u00fa obrovsk\u00fd potenci\u00e1l na rie\u0161enie komplexn\u00fdch probl\u00e9mov a zlep\u0161ovanie n\u00e1\u0161ho \u017eivota, nes\u00fa so sebou aj v\u00fdznamn\u00fa zodpovednos\u0165. Je mimoriadne d\u00f4le\u017eit\u00e9, aby boli AI syst\u00e9my v s\u00falade s \u013eudsk\u00fdmi hodnotami a pou\u017e\u00edvali sa zodpovedne – najm\u00e4, ke\u010f prech\u00e1dzame nezn\u00e1mymi \u00fazemiami kognit\u00edvnych schopnost\u00ed umelej inteligencie.<\/strong><\/p>\n<\/div>\n\n

\n

Z\u00e1ver<\/h3>\n<\/div>\n\n
\n

Na na\u0161ej ceste svetom generat\u00edvnej umelej inteligencie sme videli, \u017ee to, \u010do bolo kedysi len skvel\u00fdm sci-fi n\u00e1metom, je teraz nie\u010do ve\u013emi re\u00e1lne a d\u00f4le\u017eit\u00e9. Rizik\u00e1 spojen\u00e9 s generat\u00edvnou umelou inteligenciou nie s\u00fa len vymyslen\u00e9 pr\u00edbehy; s\u00fa to skuto\u010dn\u00e9 probl\u00e9my, s ktor\u00fdmi sa mus\u00edme \u010do najsk\u00f4r vysporiada\u0165.<\/strong><\/p>\n<\/div>\n\n

\n

Na z\u00e1ver si je d\u00f4le\u017eit\u00e9 uvedomi\u0165, \u017ee vstupujeme do novej \u00e9ry s umelou inteligenciou –  a to je ako vst\u00fapi\u0165 do ve\u013ekej nezn\u00e1mej oblasti. AI m\u00e1 ohromn\u00fd potenci\u00e1l, ale prich\u00e1dza aj s ve\u013ekou zodpovednos\u0165ou. Ak ju pou\u017eijeme nespr\u00e1vne alebo sa na \u0148u nedostato\u010dne priprav\u00edme, m\u00f4\u017ee to sp\u00f4sobi\u0165 probl\u00e9my.<\/p>\n<\/div>\n\n

\n

Nechceme by\u0165 nadmerne negat\u00edvni, chceme v\u0161ak zd\u00f4razni\u0165, \u017ee mus\u00edme by\u0165 pripraven\u00ed a pou\u017e\u00edva\u0165 AI rozumne. Nach\u00e1dzame sa na kri\u017eovatke a rozhodnutia, ktor\u00e9 dnes prijmeme, bud\u00fa formova\u0165 na\u0161u bud\u00facnos\u0165. Mus\u00edme spolupracova\u0165, aby sme zabezpe\u010dili bezpe\u010dn\u00e9 pou\u017e\u00edvanie AI, ktor\u00e1 bude ma\u0165 pre v\u0161etk\u00fdch pozit\u00edvny pr\u00ednos.<\/strong> Nebude to \u013eahk\u00e9, ale rozhodne to stoj\u00ed za to.<\/p>\n<\/div>\n\n

\n

P.S.: Ak ste sa dostali a\u017e sem a chcete sa zahra\u0165, vysk\u00fa\u0161ajte si hru Universal Paperclips<\/a> zalo\u017een\u00fa na my\u0161lienkovom experimente o maximaliz\u00e1cii produkcie papierov\u00fdch spiniek.<\/p>\n<\/div>","protected":false},"excerpt":{"rendered":"

V\u00fdchodoeur\u00f3pska letn\u00e1 \u0161kola strojov\u00e9ho u\u010denia (EEML) je prest\u00ed\u017ena udalos\u0165, ktor\u00e1 sp\u00e1ja mlad\u00fdch v\u00fdskumn\u00edkov a odborn\u00edkov v oblasti umelej inteligencie zo strednej a v\u00fdchodnej Eur\u00f3py. T\u00e1to letn\u00e1 \u0161kola je jedine\u010dnou pr\u00edle\u017eitos\u0165ou…<\/p>\n","protected":false},"author":26,"featured_media":29094,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"footnotes":""},"categories":[88,142],"tags":[111,420],"class_list":["post-29097","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-pop-science-sk","category-2023-sk","tag-disinformation-sk","tag-ai-ethics-sk"],"acf":[],"yoast_head":"\nAko m\u00f4\u017ee umel\u00e1 inteligencia zni\u010di\u0165 civiliz\u00e1ciu a \u010do m\u00f4\u017eeme urobi\u0165, aby sme tomu predi\u0161li? - KInIT<\/title>\n<meta name=\"description\" content=\"V\u00fdchodoeur\u00f3pska letn\u00e1 \u0161kola strojov\u00e9ho u\u010denia (EEML) je prest\u00ed\u017ena udalos\u0165, ktor\u00e1 sp\u00e1ja mlad\u00fdch v\u00fdskumn\u00edkov a odborn\u00edkov v oblasti umelej inteligencie zo strednej a v\u00fdchodnej Eur\u00f3py. T\u00e1to letn\u00e1 \u0161kola je jedine\u010dnou pr\u00edle\u017eitos\u0165ou u\u010di\u0165 sa od svetov\u00fdch expertov z r\u00f4znych discipl\u00edn strojov\u00e9ho u\u010denia. Tento rok sa \u0161kola konala v Ko\u0161iciach a z\u00fa\u010dastnili sa aj na\u0161i doktorandi. V tomto \u010dl\u00e1nku sa dozviete viac o rizik\u00e1ch umelej inteligencie, ktor\u00e9 predstavil priekopn\u00edk hlbok\u00e9ho u\u010denia Yoshua Bengio, a ktor\u00e9 n\u00e1sledne zhrnuli na\u0161i doktorandi Peter Pavl\u00edk a Matej \u010cief\" \/>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/kinit.sk\/sk\/ako-moze-umela-inteligencia-znicit-civilizaciu-a-co-mozeme-urobit-aby-sme-tomu-predisli\/\" \/>\n<meta property=\"og:locale\" content=\"sk_SK\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"Ako m\u00f4\u017ee umel\u00e1 inteligencia zni\u010di\u0165 civiliz\u00e1ciu a \u010do m\u00f4\u017eeme urobi\u0165, aby sme tomu predi\u0161li? - KInIT\" \/>\n<meta property=\"og:description\" content=\"V\u00fdchodoeur\u00f3pska letn\u00e1 \u0161kola strojov\u00e9ho u\u010denia (EEML) je prest\u00ed\u017ena udalos\u0165, ktor\u00e1 sp\u00e1ja mlad\u00fdch v\u00fdskumn\u00edkov a odborn\u00edkov v oblasti umelej inteligencie zo strednej a v\u00fdchodnej Eur\u00f3py. T\u00e1to letn\u00e1 \u0161kola je jedine\u010dnou pr\u00edle\u017eitos\u0165ou u\u010di\u0165 sa od svetov\u00fdch expertov z r\u00f4znych discipl\u00edn strojov\u00e9ho u\u010denia. Tento rok sa \u0161kola konala v Ko\u0161iciach a z\u00fa\u010dastnili sa aj na\u0161i doktorandi. V tomto \u010dl\u00e1nku sa dozviete viac o rizik\u00e1ch umelej inteligencie, ktor\u00e9 predstavil priekopn\u00edk hlbok\u00e9ho u\u010denia Yoshua Bengio, a ktor\u00e9 n\u00e1sledne zhrnuli na\u0161i doktorandi Peter Pavl\u00edk a Matej \u010cief\" \/>\n<meta property=\"og:url\" content=\"https:\/\/kinit.sk\/sk\/ako-moze-umela-inteligencia-znicit-civilizaciu-a-co-mozeme-urobit-aby-sme-tomu-predisli\/\" \/>\n<meta property=\"og:site_name\" content=\"KInIT\" \/>\n<meta property=\"article:published_time\" content=\"2023-10-18T10:40:09+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2023-10-18T10:48:02+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/kinit.sk\/wp-content\/uploads\/2023\/10\/202310_web_news_EEML_articles_1_feature.png.png\" \/>\n\t<meta property=\"og:image:width\" content=\"1201\" \/>\n\t<meta property=\"og:image:height\" content=\"629\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/png\" \/>\n<meta name=\"author\" content=\"Marianna Palkova\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:creator\" content=\"@kinit\" \/>\n<meta name=\"twitter:site\" content=\"@kinit\" \/>\n<meta name=\"twitter:label1\" content=\"Autor\" \/>\n\t<meta name=\"twitter:data1\" content=\"Marianna Palkova\" \/>\n\t<meta name=\"twitter:label2\" content=\"Predpokladan\u00fd \u010das \u010d\u00edtania\" \/>\n\t<meta name=\"twitter:data2\" content=\"13 min\u00fat\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\\\/\\\/schema.org\",\"@graph\":[{\"@type\":\"Article\",\"@id\":\"https:\\\/\\\/kinit.sk\\\/sk\\\/ako-moze-umela-inteligencia-znicit-civilizaciu-a-co-mozeme-urobit-aby-sme-tomu-predisli\\\/#article\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/kinit.sk\\\/sk\\\/ako-moze-umela-inteligencia-znicit-civilizaciu-a-co-mozeme-urobit-aby-sme-tomu-predisli\\\/\"},\"author\":{\"name\":\"Marianna Palkova\",\"@id\":\"https:\\\/\\\/kinit.sk\\\/#\\\/schema\\\/person\\\/8b175aaaf3267b5bbbbb97e4a6db8cea\"},\"headline\":\"Ako m\u00f4\u017ee umel\u00e1 inteligencia zni\u010di\u0165 civiliz\u00e1ciu a \u010do m\u00f4\u017eeme urobi\u0165, aby sme tomu predi\u0161li?\",\"datePublished\":\"2023-10-18T10:40:09+00:00\",\"dateModified\":\"2023-10-18T10:48:02+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\\\/\\\/kinit.sk\\\/sk\\\/ako-moze-umela-inteligencia-znicit-civilizaciu-a-co-mozeme-urobit-aby-sme-tomu-predisli\\\/\"},\"wordCount\":3873,\"image\":{\"@id\":\"https:\\\/\\\/kinit.sk\\\/sk\\\/ako-moze-umela-inteligencia-znicit-civilizaciu-a-co-mozeme-urobit-aby-sme-tomu-predisli\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/kinit.sk\\\/wp-content\\\/uploads\\\/2023\\\/10\\\/202310_web_news_EEML_articles_1_feature.png.png\",\"keywords\":[\"disinformation\",\"AI ethics\"],\"articleSection\":[\"Pop science\",\"2023\"],\"inLanguage\":\"sk-SK\"},{\"@type\":\"WebPage\",\"@id\":\"https:\\\/\\\/kinit.sk\\\/sk\\\/ako-moze-umela-inteligencia-znicit-civilizaciu-a-co-mozeme-urobit-aby-sme-tomu-predisli\\\/\",\"url\":\"https:\\\/\\\/kinit.sk\\\/sk\\\/ako-moze-umela-inteligencia-znicit-civilizaciu-a-co-mozeme-urobit-aby-sme-tomu-predisli\\\/\",\"name\":\"Ako m\u00f4\u017ee umel\u00e1 inteligencia zni\u010di\u0165 civiliz\u00e1ciu a \u010do m\u00f4\u017eeme urobi\u0165, aby sme tomu predi\u0161li? - KInIT\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/kinit.sk\\\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\\\/\\\/kinit.sk\\\/sk\\\/ako-moze-umela-inteligencia-znicit-civilizaciu-a-co-mozeme-urobit-aby-sme-tomu-predisli\\\/#primaryimage\"},\"image\":{\"@id\":\"https:\\\/\\\/kinit.sk\\\/sk\\\/ako-moze-umela-inteligencia-znicit-civilizaciu-a-co-mozeme-urobit-aby-sme-tomu-predisli\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/kinit.sk\\\/wp-content\\\/uploads\\\/2023\\\/10\\\/202310_web_news_EEML_articles_1_feature.png.png\",\"datePublished\":\"2023-10-18T10:40:09+00:00\",\"dateModified\":\"2023-10-18T10:48:02+00:00\",\"author\":{\"@id\":\"https:\\\/\\\/kinit.sk\\\/#\\\/schema\\\/person\\\/8b175aaaf3267b5bbbbb97e4a6db8cea\"},\"description\":\"V\u00fdchodoeur\u00f3pska letn\u00e1 \u0161kola strojov\u00e9ho u\u010denia (EEML) je prest\u00ed\u017ena udalos\u0165, ktor\u00e1 sp\u00e1ja mlad\u00fdch v\u00fdskumn\u00edkov a odborn\u00edkov v oblasti umelej inteligencie zo strednej a v\u00fdchodnej Eur\u00f3py. T\u00e1to letn\u00e1 \u0161kola je jedine\u010dnou pr\u00edle\u017eitos\u0165ou u\u010di\u0165 sa od svetov\u00fdch expertov z r\u00f4znych discipl\u00edn strojov\u00e9ho u\u010denia. Tento rok sa \u0161kola konala v Ko\u0161iciach a z\u00fa\u010dastnili sa aj na\u0161i doktorandi. V tomto \u010dl\u00e1nku sa dozviete viac o rizik\u00e1ch umelej inteligencie, ktor\u00e9 predstavil priekopn\u00edk hlbok\u00e9ho u\u010denia Yoshua Bengio, a ktor\u00e9 n\u00e1sledne zhrnuli na\u0161i doktorandi Peter Pavl\u00edk a Matej \u010cief\",\"breadcrumb\":{\"@id\":\"https:\\\/\\\/kinit.sk\\\/sk\\\/ako-moze-umela-inteligencia-znicit-civilizaciu-a-co-mozeme-urobit-aby-sme-tomu-predisli\\\/#breadcrumb\"},\"inLanguage\":\"sk-SK\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\\\/\\\/kinit.sk\\\/sk\\\/ako-moze-umela-inteligencia-znicit-civilizaciu-a-co-mozeme-urobit-aby-sme-tomu-predisli\\\/\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"sk-SK\",\"@id\":\"https:\\\/\\\/kinit.sk\\\/sk\\\/ako-moze-umela-inteligencia-znicit-civilizaciu-a-co-mozeme-urobit-aby-sme-tomu-predisli\\\/#primaryimage\",\"url\":\"https:\\\/\\\/kinit.sk\\\/wp-content\\\/uploads\\\/2023\\\/10\\\/202310_web_news_EEML_articles_1_feature.png.png\",\"contentUrl\":\"https:\\\/\\\/kinit.sk\\\/wp-content\\\/uploads\\\/2023\\\/10\\\/202310_web_news_EEML_articles_1_feature.png.png\",\"width\":1201,\"height\":629},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\\\/\\\/kinit.sk\\\/sk\\\/ako-moze-umela-inteligencia-znicit-civilizaciu-a-co-mozeme-urobit-aby-sme-tomu-predisli\\\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\\\/\\\/kinit.sk\\\/sk\\\/uvod\\\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"Pop science\",\"item\":\"https:\\\/\\\/kinit.sk\\\/sk\\\/category\\\/pop-science-sk\\\/\"},{\"@type\":\"ListItem\",\"position\":3,\"name\":\"Ako m\u00f4\u017ee umel\u00e1 inteligencia zni\u010di\u0165 civiliz\u00e1ciu a \u010do m\u00f4\u017eeme urobi\u0165, aby sme tomu predi\u0161li?\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\\\/\\\/kinit.sk\\\/#website\",\"url\":\"https:\\\/\\\/kinit.sk\\\/\",\"name\":\"KInIT\",\"description\":\"Vyu\u017e\u00edvame v\u00fdskum pre \u013eud\u00ed a priemysel\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\\\/\\\/kinit.sk\\\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"sk-SK\"},{\"@type\":\"Person\",\"@id\":\"https:\\\/\\\/kinit.sk\\\/#\\\/schema\\\/person\\\/8b175aaaf3267b5bbbbb97e4a6db8cea\",\"name\":\"Marianna Palkova\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"Ako m\u00f4\u017ee umel\u00e1 inteligencia zni\u010di\u0165 civiliz\u00e1ciu a \u010do m\u00f4\u017eeme urobi\u0165, aby sme tomu predi\u0161li? - KInIT","description":"V\u00fdchodoeur\u00f3pska letn\u00e1 \u0161kola strojov\u00e9ho u\u010denia (EEML) je prest\u00ed\u017ena udalos\u0165, ktor\u00e1 sp\u00e1ja mlad\u00fdch v\u00fdskumn\u00edkov a odborn\u00edkov v oblasti umelej inteligencie zo strednej a v\u00fdchodnej Eur\u00f3py. T\u00e1to letn\u00e1 \u0161kola je jedine\u010dnou pr\u00edle\u017eitos\u0165ou u\u010di\u0165 sa od svetov\u00fdch expertov z r\u00f4znych discipl\u00edn strojov\u00e9ho u\u010denia. Tento rok sa \u0161kola konala v Ko\u0161iciach a z\u00fa\u010dastnili sa aj na\u0161i doktorandi. V tomto \u010dl\u00e1nku sa dozviete viac o rizik\u00e1ch umelej inteligencie, ktor\u00e9 predstavil priekopn\u00edk hlbok\u00e9ho u\u010denia Yoshua Bengio, a ktor\u00e9 n\u00e1sledne zhrnuli na\u0161i doktorandi Peter Pavl\u00edk a Matej \u010cief","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/kinit.sk\/sk\/ako-moze-umela-inteligencia-znicit-civilizaciu-a-co-mozeme-urobit-aby-sme-tomu-predisli\/","og_locale":"sk_SK","og_type":"article","og_title":"Ako m\u00f4\u017ee umel\u00e1 inteligencia zni\u010di\u0165 civiliz\u00e1ciu a \u010do m\u00f4\u017eeme urobi\u0165, aby sme tomu predi\u0161li? - KInIT","og_description":"V\u00fdchodoeur\u00f3pska letn\u00e1 \u0161kola strojov\u00e9ho u\u010denia (EEML) je prest\u00ed\u017ena udalos\u0165, ktor\u00e1 sp\u00e1ja mlad\u00fdch v\u00fdskumn\u00edkov a odborn\u00edkov v oblasti umelej inteligencie zo strednej a v\u00fdchodnej Eur\u00f3py. T\u00e1to letn\u00e1 \u0161kola je jedine\u010dnou pr\u00edle\u017eitos\u0165ou u\u010di\u0165 sa od svetov\u00fdch expertov z r\u00f4znych discipl\u00edn strojov\u00e9ho u\u010denia. Tento rok sa \u0161kola konala v Ko\u0161iciach a z\u00fa\u010dastnili sa aj na\u0161i doktorandi. V tomto \u010dl\u00e1nku sa dozviete viac o rizik\u00e1ch umelej inteligencie, ktor\u00e9 predstavil priekopn\u00edk hlbok\u00e9ho u\u010denia Yoshua Bengio, a ktor\u00e9 n\u00e1sledne zhrnuli na\u0161i doktorandi Peter Pavl\u00edk a Matej \u010cief","og_url":"https:\/\/kinit.sk\/sk\/ako-moze-umela-inteligencia-znicit-civilizaciu-a-co-mozeme-urobit-aby-sme-tomu-predisli\/","og_site_name":"KInIT","article_published_time":"2023-10-18T10:40:09+00:00","article_modified_time":"2023-10-18T10:48:02+00:00","og_image":[{"width":1201,"height":629,"url":"https:\/\/kinit.sk\/wp-content\/uploads\/2023\/10\/202310_web_news_EEML_articles_1_feature.png.png","type":"image\/png"}],"author":"Marianna Palkova","twitter_card":"summary_large_image","twitter_creator":"@kinit","twitter_site":"@kinit","twitter_misc":{"Autor":"Marianna Palkova","Predpokladan\u00fd \u010das \u010d\u00edtania":"13 min\u00fat"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/kinit.sk\/sk\/ako-moze-umela-inteligencia-znicit-civilizaciu-a-co-mozeme-urobit-aby-sme-tomu-predisli\/#article","isPartOf":{"@id":"https:\/\/kinit.sk\/sk\/ako-moze-umela-inteligencia-znicit-civilizaciu-a-co-mozeme-urobit-aby-sme-tomu-predisli\/"},"author":{"name":"Marianna Palkova","@id":"https:\/\/kinit.sk\/#\/schema\/person\/8b175aaaf3267b5bbbbb97e4a6db8cea"},"headline":"Ako m\u00f4\u017ee umel\u00e1 inteligencia zni\u010di\u0165 civiliz\u00e1ciu a \u010do m\u00f4\u017eeme urobi\u0165, aby sme tomu predi\u0161li?","datePublished":"2023-10-18T10:40:09+00:00","dateModified":"2023-10-18T10:48:02+00:00","mainEntityOfPage":{"@id":"https:\/\/kinit.sk\/sk\/ako-moze-umela-inteligencia-znicit-civilizaciu-a-co-mozeme-urobit-aby-sme-tomu-predisli\/"},"wordCount":3873,"image":{"@id":"https:\/\/kinit.sk\/sk\/ako-moze-umela-inteligencia-znicit-civilizaciu-a-co-mozeme-urobit-aby-sme-tomu-predisli\/#primaryimage"},"thumbnailUrl":"https:\/\/kinit.sk\/wp-content\/uploads\/2023\/10\/202310_web_news_EEML_articles_1_feature.png.png","keywords":["disinformation","AI ethics"],"articleSection":["Pop science","2023"],"inLanguage":"sk-SK"},{"@type":"WebPage","@id":"https:\/\/kinit.sk\/sk\/ako-moze-umela-inteligencia-znicit-civilizaciu-a-co-mozeme-urobit-aby-sme-tomu-predisli\/","url":"https:\/\/kinit.sk\/sk\/ako-moze-umela-inteligencia-znicit-civilizaciu-a-co-mozeme-urobit-aby-sme-tomu-predisli\/","name":"Ako m\u00f4\u017ee umel\u00e1 inteligencia zni\u010di\u0165 civiliz\u00e1ciu a \u010do m\u00f4\u017eeme urobi\u0165, aby sme tomu predi\u0161li? - KInIT","isPartOf":{"@id":"https:\/\/kinit.sk\/#website"},"primaryImageOfPage":{"@id":"https:\/\/kinit.sk\/sk\/ako-moze-umela-inteligencia-znicit-civilizaciu-a-co-mozeme-urobit-aby-sme-tomu-predisli\/#primaryimage"},"image":{"@id":"https:\/\/kinit.sk\/sk\/ako-moze-umela-inteligencia-znicit-civilizaciu-a-co-mozeme-urobit-aby-sme-tomu-predisli\/#primaryimage"},"thumbnailUrl":"https:\/\/kinit.sk\/wp-content\/uploads\/2023\/10\/202310_web_news_EEML_articles_1_feature.png.png","datePublished":"2023-10-18T10:40:09+00:00","dateModified":"2023-10-18T10:48:02+00:00","author":{"@id":"https:\/\/kinit.sk\/#\/schema\/person\/8b175aaaf3267b5bbbbb97e4a6db8cea"},"description":"V\u00fdchodoeur\u00f3pska letn\u00e1 \u0161kola strojov\u00e9ho u\u010denia (EEML) je prest\u00ed\u017ena udalos\u0165, ktor\u00e1 sp\u00e1ja mlad\u00fdch v\u00fdskumn\u00edkov a odborn\u00edkov v oblasti umelej inteligencie zo strednej a v\u00fdchodnej Eur\u00f3py. T\u00e1to letn\u00e1 \u0161kola je jedine\u010dnou pr\u00edle\u017eitos\u0165ou u\u010di\u0165 sa od svetov\u00fdch expertov z r\u00f4znych discipl\u00edn strojov\u00e9ho u\u010denia. Tento rok sa \u0161kola konala v Ko\u0161iciach a z\u00fa\u010dastnili sa aj na\u0161i doktorandi. V tomto \u010dl\u00e1nku sa dozviete viac o rizik\u00e1ch umelej inteligencie, ktor\u00e9 predstavil priekopn\u00edk hlbok\u00e9ho u\u010denia Yoshua Bengio, a ktor\u00e9 n\u00e1sledne zhrnuli na\u0161i doktorandi Peter Pavl\u00edk a Matej \u010cief","breadcrumb":{"@id":"https:\/\/kinit.sk\/sk\/ako-moze-umela-inteligencia-znicit-civilizaciu-a-co-mozeme-urobit-aby-sme-tomu-predisli\/#breadcrumb"},"inLanguage":"sk-SK","potentialAction":[{"@type":"ReadAction","target":["https:\/\/kinit.sk\/sk\/ako-moze-umela-inteligencia-znicit-civilizaciu-a-co-mozeme-urobit-aby-sme-tomu-predisli\/"]}]},{"@type":"ImageObject","inLanguage":"sk-SK","@id":"https:\/\/kinit.sk\/sk\/ako-moze-umela-inteligencia-znicit-civilizaciu-a-co-mozeme-urobit-aby-sme-tomu-predisli\/#primaryimage","url":"https:\/\/kinit.sk\/wp-content\/uploads\/2023\/10\/202310_web_news_EEML_articles_1_feature.png.png","contentUrl":"https:\/\/kinit.sk\/wp-content\/uploads\/2023\/10\/202310_web_news_EEML_articles_1_feature.png.png","width":1201,"height":629},{"@type":"BreadcrumbList","@id":"https:\/\/kinit.sk\/sk\/ako-moze-umela-inteligencia-znicit-civilizaciu-a-co-mozeme-urobit-aby-sme-tomu-predisli\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/kinit.sk\/sk\/uvod\/"},{"@type":"ListItem","position":2,"name":"Pop science","item":"https:\/\/kinit.sk\/sk\/category\/pop-science-sk\/"},{"@type":"ListItem","position":3,"name":"Ako m\u00f4\u017ee umel\u00e1 inteligencia zni\u010di\u0165 civiliz\u00e1ciu a \u010do m\u00f4\u017eeme urobi\u0165, aby sme tomu predi\u0161li?"}]},{"@type":"WebSite","@id":"https:\/\/kinit.sk\/#website","url":"https:\/\/kinit.sk\/","name":"KInIT","description":"Vyu\u017e\u00edvame v\u00fdskum pre \u013eud\u00ed a priemysel","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/kinit.sk\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"sk-SK"},{"@type":"Person","@id":"https:\/\/kinit.sk\/#\/schema\/person\/8b175aaaf3267b5bbbbb97e4a6db8cea","name":"Marianna Palkova"}]}},"_links":{"self":[{"href":"https:\/\/kinit.sk\/sk\/wp-json\/wp\/v2\/posts\/29097","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/kinit.sk\/sk\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/kinit.sk\/sk\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/kinit.sk\/sk\/wp-json\/wp\/v2\/users\/26"}],"replies":[{"embeddable":true,"href":"https:\/\/kinit.sk\/sk\/wp-json\/wp\/v2\/comments?post=29097"}],"version-history":[{"count":3,"href":"https:\/\/kinit.sk\/sk\/wp-json\/wp\/v2\/posts\/29097\/revisions"}],"predecessor-version":[{"id":29101,"href":"https:\/\/kinit.sk\/sk\/wp-json\/wp\/v2\/posts\/29097\/revisions\/29101"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/kinit.sk\/sk\/wp-json\/wp\/v2\/media\/29094"}],"wp:attachment":[{"href":"https:\/\/kinit.sk\/sk\/wp-json\/wp\/v2\/media?parent=29097"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/kinit.sk\/sk\/wp-json\/wp\/v2\/categories?post=29097"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/kinit.sk\/sk\/wp-json\/wp\/v2\/tags?post=29097"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}