{"id":17625,"date":"2025-09-25T09:30:46","date_gmt":"2025-09-25T07:30:46","guid":{"rendered":"https:\/\/www.slovensko.ai\/?p=17625"},"modified":"2025-09-25T08:04:20","modified_gmt":"2025-09-25T06:04:20","slug":"sef-anthropic-varuje-pred-ai-katastrofou","status":"publish","type":"post","link":"https:\/\/www.slovensko.ai\/en\/sef-anthropic-varuje-pred-ai-katastrofou\/","title":{"rendered":"\u0160\u00e9f Anthropic varuje pred AI katastrofou"},"content":{"rendered":"
<\/span> 477<\/span> Viden\u00ed<\/span><\/div>

Pod\u013ea Amodeiho je riziko v\u00e1\u017ene a nesmie sa podceni\u0165.<\/strong> Na summite Axios AI+ DC<\/strong> v septembri 2025 upozornil CEO spolo\u010dnosti Anthropic<\/strong>, \u017ee pod\u013ea jeho osobn\u00e9ho odhadu existuje pribli\u017ene 25\u202f% \u0161anca<\/strong>, \u017ee v\u00fdvoj umelej inteligencie povedie k v\u00e1\u017enym alebo a\u017e existenci\u00e1lnym d\u00f4sledkom<\/strong> pre \u013eudstvo. Pod katastrof\u00e1lnym scen\u00e1rom m\u00e1 na mysli situ\u00e1cie, v ktor\u00fdch AI nadobudne schopnosti mimo \u013eudskej kontroly.<\/p>\n

Amodei z\u00e1rove\u0148 zd\u00f4raznil, \u017ee riziko nie je istota<\/strong>, ale dostato\u010dne v\u00e1\u017ena pravdepodobnos\u0165<\/strong>, aby si vy\u017eadovala maxim\u00e1lnu pozornos\u0165 vedcov, firiem aj vl\u00e1d.<\/p>\n

\u010co znamen\u00e1 25\u202f% pravdepodobnos\u0165 v praxi?<\/strong><\/h3>\n

Je to sign\u00e1l na okam\u017eit\u00fa akciu, nie d\u00f4vod na paniku.<\/strong> Odhad 25\u202f% predstavuje v\u00e1\u017ene riziko<\/strong>, ktor\u00e9 by sa v in\u00fdch oblastiach (napr. zdravotn\u00edctvo alebo bezpe\u010dnos\u0165 jadrov\u00fdch elektr\u00e1rn\u00ed) pova\u017eovalo za neprijate\u013en\u00e9 bez pr\u00edsnej kontroly<\/strong>. Amodei otvorene prizn\u00e1va, \u017ee tak\u00e1to pravdepodobnos\u0165 nie je vedecky podlo\u017een\u00e1 \u0161tatistika, ale subjekt\u00edvny odborn\u00fd n\u00e1zor<\/strong> na z\u00e1klade doteraj\u0161\u00edch trendov a vlastn\u00e9ho v\u00fdskumu.<\/p>\n

Pre porovnanie:<\/p>\n

\n
\n\n\n\n\n\n\n\n
Riziko<\/th>\nPribli\u017en\u00e1 pravdepodobnos\u0165<\/th>\n<\/tr>\n<\/thead>\n
P\u00e1d ve\u013ek\u00e9ho meteoru<\/td>\n<0.01\u202f%<\/td>\n<\/tr>\n
Pand\u00e9mia podobn\u00e1 COVID-19<\/td>\n1\u20132\u202f% ro\u010dne<\/td>\n<\/tr>\n
Katastrofick\u00fd v\u00fdvoj AI<\/td>\n25\u202f% (pod\u013ea Amodeiho)<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n<\/div>\n

Ako reaguje odborn\u00e1 verejnos\u0165?<\/strong><\/h3>\n

Odborn\u00edci s\u00fa rozdelen\u00ed niektor\u00ed podporuj\u00fa varovanie, in\u00ed ho spochyb\u0148uj\u00fa.<\/strong> Viacer\u00ed v\u00fdskumn\u00edci v oblasti AI bezpe\u010dnosti s\u00fahlasia, \u017ee potenci\u00e1lne rizik\u00e1 existuj\u00fa a treba ich bra\u0165 v\u00e1\u017ene. Z\u00e1stupcovia firiem ako OpenAI<\/strong>, DeepMind<\/strong> alebo MIRI<\/strong> dlhodobo upozor\u0148uj\u00fa na potrebu regul\u00e1cie a testovania tzv. frontier modelov<\/strong> (najpokro\u010dilej\u0161\u00edch LLM a AGI syst\u00e9mov).<\/p>\n

Na druhej strane, niektor\u00ed akademici tvrdia, \u017ee podobn\u00e9 odhady s\u00fa prehnan\u00e9 alebo \u0161pekulat\u00edvne<\/strong>, a upozor\u0148uj\u00fa sk\u00f4r na aktu\u00e1lne probl\u00e9my AI<\/strong>, ako je dezinform\u00e1cia, diskrimin\u00e1cia \u010di zneu\u017eitie technol\u00f3gi\u00ed v politike.<\/p>\n

\u010co rob\u00ed Anthropic pre bezpe\u010dnos\u0165 AI?<\/strong><\/h3>\n

Spolo\u010dnos\u0165 sa s\u00fastred\u00ed na tzv. kontextov\u00e9 vyrovn\u00e1vanie riz\u00edk (Constitutional AI).<\/strong> Anthropic je zn\u00e1ma najm\u00e4 v\u00fdvojom jazykov\u00e9ho modelu Claude<\/strong>, ktor\u00e9ho tr\u00e9ning sa riadi vopred definovan\u00fdmi etick\u00fdmi princ\u00edpmi. Pr\u00edstup firmy kladie d\u00f4raz na transparentnos\u0165 spr\u00e1vania modelu, interpretovate\u013enos\u0165 v\u00fdstupov<\/strong> a pr\u00edsne testovanie v r\u00f4znych scen\u00e1roch.<\/p>\n

Okrem toho Anthropic spolupracuje s vl\u00e1dnymi a v\u00fdskumn\u00fdmi in\u0161tit\u00faciami na v\u00fdvoji \u0161tandardov pre AI bezpe\u010dnos\u0165<\/strong>, vr\u00e1tane syst\u00e9mu AI evaluation benchmarks<\/strong>.<\/p>\n

Ak\u00e9 s\u00fa odpor\u00fa\u010dania Amodeiho a \u010fal\u0161\u00edch expertov?<\/strong><\/h3>\n

Pre zn\u00ed\u017eenie riz\u00edk AI navrhuj\u00fa:<\/strong><\/p>\n

    \n
  1. \n

    Pr\u00edsne testovanie AI syst\u00e9mov pred nasaden\u00edm do praxe<\/strong><\/p>\n<\/li>\n

  2. \n

    Zriadenie nez\u00e1visl\u00fdch regula\u010dn\u00fdch agent\u00far<\/strong><\/p>\n<\/li>\n

  3. \n

    Medzin\u00e1rodn\u00fa koordin\u00e1ciu pravidiel pre v\u00fdvoj frontier AI<\/strong><\/p>\n<\/li>\n

  4. \n

    Financovanie v\u00fdskumu bezpe\u010dnosti a etiky AI<\/strong><\/p>\n<\/li>\n

  5. \n

    Transparentnos\u0165 zo strany v\u00fdvoj\u00e1rov a firiem<\/strong><\/p>\n<\/li>\n<\/ol>\n

    Tento pr\u00edstup by mal minimalizova\u0165 \u0161ance na ne\u00famyseln\u00e9 spr\u00e1vanie AI syst\u00e9mov a zabr\u00e1ni\u0165 mo\u017en\u00e9mu nekontrolovan\u00e9mu rozvoju superinteligencie<\/strong>.<\/p>","protected":false},"excerpt":{"rendered":"

    477 Viden\u00edPod\u013ea Amodeiho je riziko v\u00e1\u017ene a nesmie sa podceni\u0165. Na summite Axios AI+ DC v septembri 2025 upozornil CEO spolo\u010dnosti Anthropic, \u017ee pod\u013ea jeho osobn\u00e9ho odhadu existuje pribli\u017ene 25\u202f% \u0161anca, \u017ee v\u00fdvoj umelej inteligencie povedie k v\u00e1\u017enym alebo a\u017e existenci\u00e1lnym d\u00f4sledkom pre \u013eudstvo. Pod katastrof\u00e1lnym scen\u00e1rom m\u00e1 na mysli situ\u00e1cie, v ktor\u00fdch AI nadobudne […]<\/p>\n","protected":false},"author":5,"featured_media":17626,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"footnotes":""},"categories":[51],"tags":[377,79,53],"class_list":["post-17625","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-blog","tag-ai-novinky","tag-slovensko-ai","tag-umela-inteligencia"],"acf":[],"_links":{"self":[{"href":"https:\/\/www.slovensko.ai\/en\/wp-json\/wp\/v2\/posts\/17625"}],"collection":[{"href":"https:\/\/www.slovensko.ai\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.slovensko.ai\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.slovensko.ai\/en\/wp-json\/wp\/v2\/users\/5"}],"replies":[{"embeddable":true,"href":"https:\/\/www.slovensko.ai\/en\/wp-json\/wp\/v2\/comments?post=17625"}],"version-history":[{"count":1,"href":"https:\/\/www.slovensko.ai\/en\/wp-json\/wp\/v2\/posts\/17625\/revisions"}],"predecessor-version":[{"id":17627,"href":"https:\/\/www.slovensko.ai\/en\/wp-json\/wp\/v2\/posts\/17625\/revisions\/17627"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.slovensko.ai\/en\/wp-json\/wp\/v2\/media\/17626"}],"wp:attachment":[{"href":"https:\/\/www.slovensko.ai\/en\/wp-json\/wp\/v2\/media?parent=17625"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.slovensko.ai\/en\/wp-json\/wp\/v2\/categories?post=17625"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.slovensko.ai\/en\/wp-json\/wp\/v2\/tags?post=17625"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}