{"id":195968,"date":"2026-08-30T06:59:15","date_gmt":"2026-08-30T06:59:15","guid":{"rendered":"https:\/\/www.europesays.com\/cz\/195968\/"},"modified":"2026-08-30T06:59:15","modified_gmt":"2026-08-30T06:59:15","slug":"zahadny-kupec-skupoval-tisice-starych-knih-airtag-odhalil-amazon-ktery-je-rozrezava-pro-hladovou-ai","status":"publish","type":"post","link":"https:\/\/www.europesays.com\/cz\/195968\/","title":{"rendered":"Z\u00e1hadn\u00fd kupec skupoval tis\u00edce star\u00fdch knih. AirTag odhalil Amazon, kter\u00fd je roz\u0159ez\u00e1v\u00e1 pro hladovou AI"},"content":{"rendered":"<p>Na neobvykl\u00e9 vyu\u017eit\u00ed star\u00fdch knih narazil prodejce p\u016fsob\u00edc\u00ed na platform\u011b Biblio, kdy\u017e obdr\u017eel podez\u0159ele velkou objedn\u00e1vku zhruba tis\u00edcovky titul\u016f od nezn\u00e1m\u00e9ho z\u00e1kazn\u00edka. Do jednoho v\u00fdtisku proto ukryl sledovac\u00ed za\u0159\u00edzen\u00ed Apple AirTag a\u00a0spole\u010dn\u011b s\u00a0novin\u00e1\u0159i americk\u00e9ho serveru 404 Media sledoval jeho cestu. Kniha se z\u00a0Kalifornie p\u0159es n\u011bkolik americk\u00fdch st\u00e1t\u016f nakonec dostala do skladu Amazonu na severov\u00fdchod\u011b Las Vegas.<\/p>\n<p>Pr\u00e1v\u011b tam se z<a href=\"https:\/\/www.e15.cz\/zahranicni\/zakaznici-chteji-jistotu-na-knihy-miri-nalepky-napsano-clovekem-garantuji-ze-je-nepsala-ai-1435112\" rel=\"nofollow noopener\" target=\"_blank\">\u00a0fyzick\u00fdch knih st\u00e1vaj\u00ed data<\/a>. Zam\u011bstnanci od\u0159\u00edznou h\u0159bet, jednotliv\u00e9 str\u00e1nky projdou skenerem a\u00a0jejich obsah se p\u0159evede do digit\u00e1ln\u00ed podoby. Data n\u00e1sledn\u011b slou\u017e\u00ed k<a href=\"https:\/\/www.e15.cz\/nazory-a-analyzy\/evropska-unie-chce-po-ai-aby-praskala-internet-se-okamzite-brani-1435140\" rel=\"nofollow noopener\" target=\"_blank\">\u00a0tr\u00e9nov\u00e1n\u00ed syst\u00e9m\u016f um\u011bl\u00e9 inteligence<\/a>. Amazon pro 404 Media uvedl pouze to, \u017ee knihy nakupuje b\u011b\u017en\u00fdmi komer\u010dn\u00edmi cestami, aby jejich prost\u0159ednictv\u00edm pom\u00e1hal vyv\u00edjet a\u00a0zlep\u0161ovat produkty a\u00a0slu\u017eby vyu\u017e\u00edvan\u00e9 z\u00e1kazn\u00edky.\u00a0<\/p>\n<p>Hodnotu nese text napsan\u00fd \u010dlov\u011bkem\u00a0<\/p>\n<p>D\u016fvodem, pro\u010d maj\u00ed pro technologick\u00e9 firmy hodnotu i\u00a0star\u00e9 a\u00a0d\u0159\u00edve obt\u00ed\u017en\u011b prodejn\u00e9 knihy, je p\u0159edev\u0161\u00edm jejich lidsk\u00fd p\u016fvod. V\u00fdzkumn\u00e1 organizace <a href=\"https:\/\/epoch.ai\/publications\/will-we-run-out-of-data-limits-of-llm-scaling-based-on-human-generated-data\" rel=\"nofollow noopener\" target=\"_blank\">Epoch AI<\/a> odhaduje mno\u017estv\u00ed kvalitn\u00edho a\u00a0ve\u0159ejn\u011b dostupn\u00e9ho textu vytvo\u0159en\u00e9ho lidmi p\u0159ibli\u017en\u011b na 300\u00a0bilion\u016f token\u016f. Podle jej\u00edch odhad\u016f by se tato z\u00e1soba mohla pro tr\u00e9nov\u00e1n\u00ed model\u016f vy\u010derpat mezi lety 2026 a\u00a02032, p\u0159i\u010dem\u017e st\u0159ed odhadu p\u0159ipad\u00e1 na rok 2028. Spoluzakladatel OpenAI Ilya Sutskever <a href=\"https:\/\/www.theverge.com\/2024\/12\/13\/24320811\/what-ilya-sutskever-sees-openai-model-data-training\" rel=\"nofollow noopener\" target=\"_blank\">u\u017e v\u00a0roce 2024 upozor\u0148oval<\/a>, \u017ee mno\u017estv\u00ed lidsk\u00fdch dat je omezen\u00e9 a\u00a0internet nelze jednodu\u0161e vytvo\u0159it znovu.<\/p>\n<p><a class=\"article-card__image\" data-ac-image=\"\" href=\"https:\/\/www.e15.cz\/zahranicni\/dostali-milionovou-dotaci-na-detskou-knihu-vytvorila-ji-ai-stezuji-si-umelci-1434473\" title=\"Dostali milionovou dotaci na d\u011btskou knihu. Vytvo\u0159ila ji AI, st\u011b\u017euj\u00ed si um\u011blci\" rel=\"nofollow noopener\" target=\"_blank\"><img decoding=\"async\" alt=\"Dostali milionovou dotaci na d\u011btskou knihu. Vytvo\u0159ila ji AI, st\u011b\u017euj\u00ed si um\u011blci\" title=\"Dostali milionovou dotaci na d\u011btskou knihu. Vytvo\u0159ila ji AI, st\u011b\u017euj\u00ed si um\u011blci\" class=\"\" data-magazine=\"\" src=\"https:\/\/www.europesays.com\/cz\/wp-content\/uploads\/2026\/08\/9917220.jpg\" loading=\"lazy\" style=\"aspect-ratio: 328 \/ 184\"\/><\/a><a class=\"article-card__title-link line-limit-3\" data-ac-title=\"\" href=\"https:\/\/www.e15.cz\/zahranicni\/dostali-milionovou-dotaci-na-detskou-knihu-vytvorila-ji-ai-stezuji-si-umelci-1434473\" title=\"Dostali milionovou dotaci na d\u011btskou knihu. Vytvo\u0159ila ji AI, st\u011b\u017euj\u00ed si um\u011blci\" rel=\"nofollow noopener\" target=\"_blank\">Dostali milionovou dotaci na d\u011btskou knihu. Vytvo\u0159ila ji AI, st\u011b\u017euj\u00ed si um\u011blci<\/a><\/p>\n<p>Situaci nav\u00edc komplikuje rostouc\u00ed mno\u017estv\u00ed obsahu vytvo\u0159en\u00e9ho samotnou um\u011blou inteligenc\u00ed. <a href=\"https:\/\/ahrefs.com\/blog\/ai-seo-statistics\/\" rel=\"nofollow noopener\" target=\"_blank\">Studie spole\u010dnosti Ahrefs<\/a> z\u00a0roku 2025 uv\u00e1d\u00ed, \u017ee n\u011bjakou formu AI textu obsahovalo 74\u00a0procent nov\u011b vznikl\u00fdch webov\u00fdch str\u00e1nek. Tr\u00e9nov\u00e1n\u00ed model\u016f na jejich vlastn\u00edch v\u00fdstupech p\u0159itom m\u016f\u017ee v\u00e9st k\u00a0postupn\u00e9mu hromad\u011bn\u00ed chyb a\u00a0degradaci v\u00fdsledk\u016f. Pro v\u00fdvoj\u00e1\u0159e proto z\u00edsk\u00e1vaj\u00ed na hodnot\u011b texty, u\u00a0nich\u017e lze s\u00a0vysokou pravd\u011bpodobnost\u00ed ur\u010dit, \u017ee vznikly je\u0161t\u011b p\u0159ed n\u00e1stupem generativn\u00ed AI. Typick\u00fdm p\u0159\u00edkladem jsou pr\u00e1v\u011b knihy vydan\u00e9 p\u0159ed rokem 2022.<\/p>\n<p>Samotn\u00e9 listy zrychluj\u00ed skenov\u00e1n\u00ed<\/p>\n<p>Digitalizace knih, p\u0159i n\u00ed\u017e dojde k\u00a0jejich zni\u010den\u00ed, je v\u00fdrazn\u011b rychlej\u0161\u00ed ne\u017e \u0161etrn\u00e9 skenov\u00e1n\u00ed. Po odstran\u011bn\u00ed h\u0159betu lze jednotliv\u00e9 listy vlo\u017eit do v\u00fdkonn\u00fdch skener\u016f s\u00a0automatick\u00fdm podava\u010dem. Nap\u0159\u00edklad <a href=\"https:\/\/digitization.archive.org\/digitization\/\" rel=\"nofollow noopener\" target=\"_blank\">Internet Archive<\/a> p\u0159i nedestruktivn\u00ed digitalizaci zvl\u00e1d\u00e1 p\u0159ibli\u017en\u011b 500 a\u017e 800\u00a0stran za hodinu a\u00a0str\u00e1nky mus\u00ed ru\u010dn\u011b obracet pracovn\u00edk. Pr\u016fmyslov\u00e9 skenov\u00e1n\u00ed samostatn\u00fdch list\u016f je podle \u017div\u011b.cz \u0159\u00e1dov\u011b rychlej\u0161\u00ed a\u00a0levn\u011bj\u0161\u00ed. N\u00e1sledn\u00e9 OCR dok\u00e1\u017ee z\u00a0kvalitn\u00edch sn\u00edmk\u016f rozpoznat 95 a\u017e 99\u00a0procent znak\u016f, st\u00e1le \u010dast\u011bji ale tuto f\u00e1zi zvl\u00e1daj\u00ed p\u0159\u00edmo multimod\u00e1ln\u00ed AI modely.<\/p>\n<p>Samotn\u00e9 knihy p\u0159itom objemem dat internet nenahrad\u00ed. Kdy\u017e harvardsk\u00e1 Institutional Data Initiative zve\u0159ejnila t\u00e9m\u011b\u0159 milion knih z\u00a0oblasti voln\u00fdch d\u011bl, jejich 394\u00a0milion\u016f stran p\u0159edstavovalo p\u0159ibli\u017en\u011b 242\u00a0miliard token\u016f. To odpov\u00edd\u00e1 zhruba jednomu procentu typick\u00e9ho sou\u010dasn\u00e9ho tr\u00e9novac\u00edho korpusu, zat\u00edmco ve\u0159ejn\u00fd webov\u00fd archiv Common Crawl obsahuje p\u0159es 100\u00a0bilion\u016f token\u016f. Hodnota star\u00fdch knih v\u0161ak spo\u010d\u00edv\u00e1 p\u0159edev\u0161\u00edm v\u00a0kvalit\u011b, jedine\u010dnosti a\u00a0v\u00a0jistot\u011b, \u017ee jejich obsah vytvo\u0159il \u010dlov\u011bk. Destruktivn\u00ed postup sou\u010dasn\u011b p\u0159in\u00e1\u0161\u00ed riziko, \u017ee mezi knihami skon\u010d\u00ed i\u00a0vz\u00e1cn\u00e9 nebo unik\u00e1tn\u00ed exempl\u00e1\u0159e, proto\u017ee p\u0159i n\u00e1kupu m\u016f\u017ee b\u00fdt rozhoduj\u00edc\u00ed <a href=\"https:\/\/www.e15.cz\/byznys\/ceskou-audioknihu-poprve-kompletne-namluvila-ai-novy-model-muze-zmenit-cely-trh-1432069\" rel=\"nofollow noopener\" target=\"_blank\">jedine\u010dnost textu pro tr\u00e9nink AI<\/a>, nikoliv historick\u00e1 \u010di bibliografick\u00e1 hodnota konkr\u00e9tn\u00edho v\u00fdtisku.<\/p>\n","protected":false},"excerpt":{"rendered":"Na neobvykl\u00e9 vyu\u017eit\u00ed star\u00fdch knih narazil prodejce p\u016fsob\u00edc\u00ed na platform\u011b Biblio, kdy\u017e obdr\u017eel podez\u0159ele velkou objedn\u00e1vku zhruba tis\u00edcovky&hellip;\n","protected":false},"author":2,"featured_media":195969,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":"","_share_on_mastodon":"0"},"categories":[12],"tags":[45169,891,127,67,68,825,14700,4949,845],"class_list":["post-195968","post","type-post","status-publish","format-standard","has-post-thumbnail","category-byznys","tag-404-media","tag-amazon","tag-apple","tag-business","tag-byznys","tag-internet","tag-kniha","tag-trenink","tag-umela-inteligence"],"share_on_mastodon":{"url":"https:\/\/pubeurope.com\/@cz\/117183169692553582","error":""},"_links":{"self":[{"href":"https:\/\/www.europesays.com\/cz\/wp-json\/wp\/v2\/posts\/195968","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.europesays.com\/cz\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.europesays.com\/cz\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/cz\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/cz\/wp-json\/wp\/v2\/comments?post=195968"}],"version-history":[{"count":0,"href":"https:\/\/www.europesays.com\/cz\/wp-json\/wp\/v2\/posts\/195968\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/cz\/wp-json\/wp\/v2\/media\/195969"}],"wp:attachment":[{"href":"https:\/\/www.europesays.com\/cz\/wp-json\/wp\/v2\/media?parent=195968"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.europesays.com\/cz\/wp-json\/wp\/v2\/categories?post=195968"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.europesays.com\/cz\/wp-json\/wp\/v2\/tags?post=195968"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}