{"id":5425,"date":"2026-06-03T14:14:43","date_gmt":"2026-06-03T12:14:43","guid":{"rendered":"https:\/\/certainqms.com\/?p=5425"},"modified":"2026-06-04T05:26:49","modified_gmt":"2026-06-04T03:26:49","slug":"ai-risiko","status":"publish","type":"post","link":"https:\/\/certainqms.com\/no\/ressurser\/ai-risiko\/","title":{"rendered":"Hallusinasjoner, bias og\u00a0black\u00a0boxes:\u00a0AI-risikoene ledere m\u00e5 forst\u00e5\u00a0"},"content":{"rendered":"

[et_pb_section fb_built=»1″ _builder_version=»4.27.5″ _module_preset=»default» background_color=»RGBA(255,255,255,0)» use_background_color_gradient=»off» custom_padding=»0px||0px||true|false» global_colors_info=»{}» theme_builder_area=»et_body_layout»][et_pb_row _builder_version=»4.27.5″ _module_preset=»default» width=»100%» global_colors_info=»{}» theme_builder_area=»et_body_layout»][et_pb_column type=»4_4″ _builder_version=»4.27.5″ _module_preset=»default» global_colors_info=»{}» theme_builder_area=»et_body_layout»][et_pb_image src=»https:\/\/certainqms.com\/wp-content\/uploads\/2026\/06\/AI-risiko-1.webp» alt=»AI risiko» title_text=»AI risiko» _builder_version=»4.27.6″ _module_preset=»default» hover_enabled=»0″ border_radii=»on|28px|28px|28px|28px» global_colors_info=»{}» theme_builder_area=»et_body_layout» sticky_enabled=»0″][\/et_pb_image][\/et_pb_column][\/et_pb_row][\/et_pb_section][et_pb_section fb_built=»1″ module_class=»fullwidth-bg» _builder_version=»4.27.5″ _module_preset=»default» background_color=»#FFFFFF» use_background_color_gradient=»off» custom_margin=»||||true|false» custom_padding=»20px||20px||true|false» custom_css_free_form=».fullwidth-bg {|| width: 100vw !important;|| position: relative;|| left: 50%;|| right: 50%;|| margin-left: -50vw;|| margin-right: -50vw;|| background-color: #ffeedd !important;|| padding-top: 40px; || padding-bottom: 40px; ||}||||.fullwidth-bg .et_pb_row {|| max-width: 870px; || margin: 0 auto; || padding-left: 0 !important;|| padding-right: 0 !important;||}||||@media (max-width: 768px) {|| .fullwidth-bg .et_pb_row {|| max-width: 500px !important;|| }||}||||@media (max-width: 480px) {|| .fullwidth-bg .et_pb_row {|| max-width: 250px !important;|| }||}» global_colors_info=»{}» theme_builder_area=»et_body_layout»][et_pb_row _builder_version=»4.27.5″ _module_preset=»default» custom_padding=»20px||20px||true|true» global_colors_info=»{}» theme_builder_area=»et_body_layout»][et_pb_column type=»4_4″ _builder_version=»4.27.5″ _module_preset=»default» global_colors_info=»{}» theme_builder_area=»et_body_layout»][et_pb_text _builder_version=»4.27.5″ _module_preset=»default» text_font=»|600|||||||» text_text_color=»#060C31″ text_font_size=»18px» custom_margin=»-10px||||false|false» custom_padding=»0px||||false|false» global_colors_info=»{%22gcid-7a729979-e95d-4a37-b4c2-da38eea9cb79%22:%91%22header_text_color%22,%22header_text_color%22%93}» theme_builder_area=»et_body_layout»]<\/p>\n

Kunstig intelligens er p\u00e5 vei inn i stadig flere virksomhetsprosesser. Ansatte bruker\u00a0AI\u00a0til \u00e5 skrive, analysere, oppsummere, s\u00f8ke etter informasjon og generere beslutningsgrunnlag. Mange virksomheter eksperimenterer ogs\u00e5 med\u00a0AI-l\u00f8sninger som integreres med interne systemer, dokumenter og arbeidsflyter.<\/span>\u00a0<\/span><\/p>\n

For mange ledere oppleves\u00a0AI\u00a0f\u00f8rst og fremst som en ny type programvare. Men det er nettopp her mye av utfordringen ligger.\u00a0AI\u00a0fungerer fundamentalt annerledes enn de fleste systemene virksomheter er vant til \u00e5 styre og kontrollere.<\/span>\u00a0<\/span><\/p>\n

Tradisjonell programvare f\u00f8lger regler.\u00a0AI\u00a0arbeider med sannsynligheter. Tradisjonelle systemer er bygget for \u00e5 v\u00e6re forutsigbare.\u00a0AI\u00a0er bygget for \u00e5 h\u00e5ndtere usikkerhet. Derfor oppst\u00e5r det ogs\u00e5 risikoer som mange ledere ikke tidligere har m\u00e5ttet forholde seg til.<\/span>\u00a0<\/span><\/p>\n

\n

[\/et_pb_text][\/et_pb_column][\/et_pb_row][\/et_pb_section][et_pb_section fb_built=»1″ module_class=»fullwidth-bg» _builder_version=»4.27.6″ _module_preset=»default» background_color=»#FFF3EB» use_background_color_gradient=»off» custom_margin=»||||true|false» custom_padding=»20px||20px||true|false» hover_enabled=»0″ custom_css_free_form=».fullwidth-bg {|| width: 100vw !important;|| position: relative;|| left: 50%;|| right: 50%;|| margin-left: -50vw;|| margin-right: -50vw;|| background-color: #ffeedd !important;|| padding-top: 40px; || padding-bottom: 40px; ||}||||.fullwidth-bg .et_pb_row {|| max-width: 870px; || margin: 0 auto; || padding-left: 0 !important;|| padding-right: 0 !important;||}||||@media (max-width: 768px) {|| .fullwidth-bg .et_pb_row {|| max-width: 500px !important;|| }||}||||@media (max-width: 480px) {|| .fullwidth-bg .et_pb_row {|| max-width: 250px !important;|| }||}» global_colors_info=»{}» theme_builder_area=»et_body_layout» sticky_enabled=»0″][et_pb_row _builder_version=»4.27.5″ _module_preset=»default» custom_padding=»20px||30px||false|true» global_colors_info=»{}» theme_builder_area=»et_body_layout»][et_pb_column type=»4_4″ _builder_version=»4.27.5″ _module_preset=»default» global_colors_info=»{}» theme_builder_area=»et_body_layout»][et_pb_heading title=»Manglende determinisme og forutsigbarhet\u00a0» _builder_version=»4.27.5″ _module_preset=»default» title_level=»h2″ title_text_color=»#060C31″ title_font_size=»32px» title_line_height=»1.3em» custom_margin=»||||false|false» global_colors_info=»{}» theme_builder_area=»et_body_layout»][\/et_pb_heading][et_pb_text _builder_version=»4.27.5″ _module_preset=»default» text_text_color=»#060C31″ custom_margin=»-10px||||false|false» custom_padding=»0px||||false|false» global_colors_info=»{%22gcid-7a729979-e95d-4a37-b4c2-da38eea9cb79%22:%91%22header_text_color%22,%22header_text_color%22%93}» theme_builder_area=»et_body_layout»]<\/p>\n

De fleste IT-systemer er deterministiske. Det betyr at samme input normalt gir samme output.<\/span>\u00a0<\/span><\/p>\n

Hvis en kunde bestiller et produkt, registreres ordren. Hvis en faktura er forfalt, sendes en p\u00e5minnelse. Hvis en bruker ikke har tilgang, f\u00e5r vedkommende ikke logget inn. Systemet gj\u00f8r det det er programmert til \u00e5 gj\u00f8re.<\/span>\u00a0<\/span><\/p>\n

Moderne\u00a0AI\u00a0fungerer annerledes. En spr\u00e5kmodell er ikke bygget rundt faste regler for alle mulige situasjoner. Den er trent p\u00e5 enorme mengder tekst og beregner hvilke svar som statistisk sett er mest sannsynlige basert p\u00e5 det den har l\u00e6rt.<\/span>\u00a0<\/span><\/p>\n

Resultatet er at\u00a0AI\u00a0kan l\u00f8se oppgaver den aldri eksplisitt er programmert for. Samtidig blir oppf\u00f8rselen mindre forutsigbar. To personer kan stille omtrent samme sp\u00f8rsm\u00e5l og f\u00e5 ulike svar. Sm\u00e5 endringer i formuleringer eller kontekst kan p\u00e5virke resultatet betydelig.<\/span>\u00a0<\/span><\/p>\n

Dette er ikke n\u00f8dvendigvis en feil ved teknologien. Det er en egenskap ved hvordan moderne\u00a0AI\u00a0fungerer. Men det gj\u00f8r at\u00a0AI\u00a0m\u00e5 vurderes annerledes enn tradisjonelle systemer som er bygget rundt faste regler og forutsigbare prosesser.<\/span>\u00a0<\/span><\/p>\n

[\/et_pb_text][\/et_pb_column][\/et_pb_row][\/et_pb_section][et_pb_section fb_built=»1″ module_class=»fullwidth-bg» _builder_version=»4.27.6″ _module_preset=»default» background_color=»#FFFFFF» use_background_color_gradient=»off» custom_margin=»||||true|false» custom_padding=»20px||20px||true|false» hover_enabled=»0″ custom_css_free_form=».fullwidth-bg {|| width: 100vw !important;|| position: relative;|| left: 50%;|| right: 50%;|| margin-left: -50vw;|| margin-right: -50vw;|| background-color: #ffeedd !important;|| padding-top: 40px; || padding-bottom: 40px; ||}||||.fullwidth-bg .et_pb_row {|| max-width: 870px; || margin: 0 auto; || padding-left: 0 !important;|| padding-right: 0 !important;||}||||@media (max-width: 768px) {|| .fullwidth-bg .et_pb_row {|| max-width: 500px !important;|| }||}||||@media (max-width: 480px) {|| .fullwidth-bg .et_pb_row {|| max-width: 250px !important;|| }||}» global_colors_info=»{}» theme_builder_area=»et_body_layout» sticky_enabled=»0″][et_pb_row _builder_version=»4.27.5″ _module_preset=»default» custom_padding=»20px||30px||false|true» global_colors_info=»{}» theme_builder_area=»et_body_layout»][et_pb_column type=»4_4″ _builder_version=»4.27.5″ _module_preset=»default» global_colors_info=»{}» theme_builder_area=»et_body_layout»][et_pb_heading title=»Hallusinasjoner og faktiske feil» _builder_version=»4.27.6″ _module_preset=»default» title_level=»h2″ title_text_color=»#060C31″ title_font_size=»32px» title_line_height=»1.3em» custom_margin=»||||false|false» hover_enabled=»0″ global_colors_info=»{}» theme_builder_area=»et_body_layout» sticky_enabled=»0″][\/et_pb_heading][et_pb_text _builder_version=»4.27.6″ _module_preset=»default» text_text_color=»#060C31″ custom_margin=»-10px||||false|false» custom_padding=»0px||||false|false» hover_enabled=»0″ global_colors_info=»{%22gcid-7a729979-e95d-4a37-b4c2-da38eea9cb79%22:%91%22header_text_color%22,%22header_text_color%22%93}» theme_builder_area=»et_body_layout» sticky_enabled=»0″]<\/p>\n

En av de mest omtalte risikoene ved generativ\u00a0AI\u00a0er hallusinasjoner.<\/span>\u00a0<\/span><\/p>\n

Begrepet beskriver situasjoner hvor en\u00a0AI-modell produserer informasjon som virker plausibel og overbevisende, men som faktisk er feil. Modellen kan finne p\u00e5 kilder, referanser, hendelser, sitater eller fakta uten \u00e5 forst\u00e5 at informasjonen er uriktig.<\/span>\u00a0<\/span><\/p>\n

For mange ledere er dette en uvant type feil. N\u00e5r et regnskapssystem beregner feil moms, oppfattes det som en feil i systemet. N\u00e5r en\u00a0AI-modell presenterer feil informasjon, kan den derimot gj\u00f8re det p\u00e5 en m\u00e5te som fremst\u00e5r b\u00e5de logisk, strukturert og faglig troverdig.<\/span>\u00a0<\/span><\/p>\n

Problemet er derfor ikke bare at\u00a0AI\u00a0kan ta feil. Problemet er at feilene ofte er vanskelige \u00e5 oppdage.<\/span>\u00a0<\/span><\/p>\n

Dette skyldes at spr\u00e5kmodeller ikke er bygget for \u00e5 vurdere hva som er sant. De er bygget for \u00e5 generere det mest sannsynlige svaret basert p\u00e5 m\u00f8nstrene de har l\u00e6rt gjennom trening. I de fleste tilfeller gir dette imponerende resultater. Men n\u00e5r modellen mangler informasjon eller misforst\u00e5r konteksten, kan den produsere svar som virker riktige uten \u00e5 v\u00e6re det.<\/span>\u00a0<\/span><\/p>\n

[\/et_pb_text][et_pb_image src=»https:\/\/certainqms.com\/wp-content\/uploads\/2026\/06\/Diskusjon-om-AI-risiko.webp» alt=»Diskusjon om AI risiko» title_text=»Diskusjon om AI risiko» _builder_version=»4.27.6″ _module_preset=»default» hover_enabled=»0″ global_colors_info=»{}» theme_builder_area=»et_body_layout» sticky_enabled=»0″][\/et_pb_image][\/et_pb_column][\/et_pb_row][\/et_pb_section][et_pb_section fb_built=»1″ module_class=»fullwidth-bg» _builder_version=»4.27.6″ _module_preset=»default» background_color=»#EEF3FF» use_background_color_gradient=»off» custom_margin=»||||true|false» custom_padding=»20px||20px||true|false» hover_enabled=»0″ custom_css_free_form=».fullwidth-bg {|| width: 100vw !important;|| position: relative;|| left: 50%;|| right: 50%;|| margin-left: -50vw;|| margin-right: -50vw;|| background-color: #ffeedd !important;|| padding-top: 40px; || padding-bottom: 40px; ||}||||.fullwidth-bg .et_pb_row {|| max-width: 870px; || margin: 0 auto; || padding-left: 0 !important;|| padding-right: 0 !important;||}||||@media (max-width: 768px) {|| .fullwidth-bg .et_pb_row {|| max-width: 500px !important;|| }||}||||@media (max-width: 480px) {|| .fullwidth-bg .et_pb_row {|| max-width: 250px !important;|| }||}» global_colors_info=»{}» theme_builder_area=»et_body_layout» sticky_enabled=»0″][et_pb_row _builder_version=»4.27.5″ _module_preset=»default» custom_padding=»20px||30px||false|true» global_colors_info=»{}» theme_builder_area=»et_body_layout»][et_pb_column type=»4_4″ _builder_version=»4.27.5″ _module_preset=»default» global_colors_info=»{}» theme_builder_area=»et_body_layout»][et_pb_heading title=»Datakvalitet, bias og skjevheter» _builder_version=»4.27.6″ _module_preset=»default» title_level=»h2″ title_text_color=»#060C31″ title_font_size=»32px» title_line_height=»1.3em» custom_margin=»||||false|false» hover_enabled=»0″ global_colors_info=»{}» theme_builder_area=»et_body_layout» sticky_enabled=»0″][\/et_pb_heading][et_pb_text _builder_version=»4.27.6″ _module_preset=»default» text_text_color=»#060C31″ custom_margin=»-10px||||false|false» custom_padding=»0px||||false|false» hover_enabled=»0″ global_colors_info=»{%22gcid-7a729979-e95d-4a37-b4c2-da38eea9cb79%22:%91%22header_text_color%22,%22header_text_color%22%93}» theme_builder_area=»et_body_layout» sticky_enabled=»0″]<\/p>\n

AI-modeller er et produkt av dataene de l\u00e6rer fra.\u00a0<\/span>\u00a0<\/span><\/p>\n

Dersom datagrunnlaget er mangelfullt, utdatert eller inneholder systematiske skjevheter, vil dette ofte p\u00e5virke resultatene modellen produserer. Dette gjelder b\u00e5de generelle spr\u00e5kmodeller og virksomhetsspesifikke\u00a0AI-l\u00f8sninger som trenes eller tilpasses p\u00e5 interne data.<\/span>\u00a0<\/span><\/p>\n

Bias er et av de mest brukte begrepene innen\u00a0AI-risiko. Det beskriver situasjoner hvor en modell systematisk favoriserer eller diskriminerer bestemte grupper, perspektiver eller utfall.<\/span>\u00a0<\/span><\/p>\n

I praksis er utfordringen ofte mer kompleks enn at\u00a0AI\u00a0er \u00abbiased\u00bb. Modellen gjenspeiler i stor grad m\u00f8nstrene den finner i dataene den har blitt eksponert for. Hvis historiske data inneholder skjevheter, kan\u00a0AI\u00a0videref\u00f8re eller forsterke disse.<\/span>\u00a0<\/span><\/p>\n

Derfor handler datakvalitet ikke bare om riktige data. Det handler ogs\u00e5 om representativitet, relevans og kontekst. To modeller med samme teknologi kan gi sv\u00e6rt ulike resultater avhengig av hvilke data de bygger p\u00e5.<\/span><\/p>\n

[\/et_pb_text][\/et_pb_column][\/et_pb_row][\/et_pb_section][et_pb_section fb_built=»1″ module_class=»fullwidth-bg» _builder_version=»4.27.6″ _module_preset=»default» background_color=»#FFFFFF» use_background_color_gradient=»off» custom_margin=»||||true|false» custom_padding=»20px||20px||true|false» hover_enabled=»0″ custom_css_free_form=».fullwidth-bg {|| width: 100vw !important;|| position: relative;|| left: 50%;|| right: 50%;|| margin-left: -50vw;|| margin-right: -50vw;|| background-color: #ffeedd !important;|| padding-top: 40px; || padding-bottom: 40px; ||}||||.fullwidth-bg .et_pb_row {|| max-width: 870px; || margin: 0 auto; || padding-left: 0 !important;|| padding-right: 0 !important;||}||||@media (max-width: 768px) {|| .fullwidth-bg .et_pb_row {|| max-width: 500px !important;|| }||}||||@media (max-width: 480px) {|| .fullwidth-bg .et_pb_row {|| max-width: 250px !important;|| }||}» global_colors_info=»{}» theme_builder_area=»et_body_layout» sticky_enabled=»0″][et_pb_row _builder_version=»4.27.5″ _module_preset=»default» custom_padding=»20px||30px||false|true» global_colors_info=»{}» theme_builder_area=»et_body_layout»][et_pb_column type=»4_4″ _builder_version=»4.27.5″ _module_preset=»default» global_colors_info=»{}» theme_builder_area=»et_body_layout»][et_pb_heading title=»Manglende forklarbarhet og transparens\u00a0» _builder_version=»4.27.6″ _module_preset=»default» title_level=»h2″ title_text_color=»#060C31″ title_font_size=»32px» title_line_height=»1.3em» custom_margin=»||||false|false» hover_enabled=»0″ global_colors_info=»{}» theme_builder_area=»et_body_layout» sticky_enabled=»0″][\/et_pb_heading][et_pb_text _builder_version=»4.27.6″ _module_preset=»default» text_text_color=»#060C31″ custom_margin=»-10px||||false|false» custom_padding=»0px||||false|false» hover_enabled=»0″ global_colors_info=»{%22gcid-7a729979-e95d-4a37-b4c2-da38eea9cb79%22:%91%22header_text_color%22,%22header_text_color%22%93}» theme_builder_area=»et_body_layout» sticky_enabled=»0″]<\/p>\n

I mange virksomheter er det viktig \u00e5 kunne forklare hvorfor en beslutning ble tatt. Dette gjelder s\u00e6rlig innen kvalitet, revisjon,\u00a0compliance, informasjonssikkerhet og andre omr\u00e5der hvor sporbarhet st\u00e5r sentralt.\u00a0AI\u00a0utfordrer dette prinsippet.<\/span>\u00a0<\/span><\/p>\n

Mange moderne modeller fungerer som det som ofte omtales som en \u00abblack\u00a0box\u00bb. Man kan observere hvilke data som g\u00e5r inn, og man kan se resultatet som kommer ut. Men det er ikke alltid mulig \u00e5 forklare n\u00f8yaktig hvordan modellen kom frem til konklusjonen.<\/span>\u00a0<\/span><\/p>\n

Dette betyr ikke at\u00a0AI\u00a0n\u00f8dvendigvis er ukontrollerbar. Men det betyr at forklarbarheten ofte er svakere enn i tradisjonelle regelbaserte systemer.<\/span>\u00a0<\/span><\/p>\n

Jo mer avansert modellen blir, desto vanskeligere kan det bli \u00e5 forst\u00e5 hvilke faktorer som p\u00e5virket resultatet. For virksomheter som er avhengige av dokumentasjon, revisjonsspor og etterpr\u00f8vbarhet, representerer dette en ny type utfordring som mange ikke tidligere har v\u00e6rt eksponert for.<\/span><\/p>\n

[\/et_pb_text][et_pb_image src=»https:\/\/certainqms.com\/wp-content\/uploads\/2026\/06\/AI-risikostyring-radgiver.webp» alt=»AI risikostyring ra\u030adgiver» title_text=»AI risikostyring ra\u030adgiver» _builder_version=»4.27.6″ _module_preset=»default» hover_enabled=»0″ global_colors_info=»{}» theme_builder_area=»et_body_layout» sticky_enabled=»0″][\/et_pb_image][\/et_pb_column][\/et_pb_row][\/et_pb_section][et_pb_section fb_built=»1″ module_class=»fullwidth-bg» _builder_version=»4.27.6″ _module_preset=»default» background_color=»#FFF3EB» use_background_color_gradient=»off» custom_margin=»||||true|false» custom_padding=»20px||20px||true|false» hover_enabled=»0″ custom_css_free_form=».fullwidth-bg {|| width: 100vw !important;|| position: relative;|| left: 50%;|| right: 50%;|| margin-left: -50vw;|| margin-right: -50vw;|| background-color: #ffeedd !important;|| padding-top: 40px; || padding-bottom: 40px; ||}||||.fullwidth-bg .et_pb_row {|| max-width: 870px; || margin: 0 auto; || padding-left: 0 !important;|| padding-right: 0 !important;||}||||@media (max-width: 768px) {|| .fullwidth-bg .et_pb_row {|| max-width: 500px !important;|| }||}||||@media (max-width: 480px) {|| .fullwidth-bg .et_pb_row {|| max-width: 250px !important;|| }||}» global_colors_info=»{}» theme_builder_area=»et_body_layout» sticky_enabled=»0″][et_pb_row _builder_version=»4.27.5″ _module_preset=»default» custom_padding=»20px||30px||false|true» global_colors_info=»{}» theme_builder_area=»et_body_layout»][et_pb_column type=»4_4″ _builder_version=»4.27.5″ _module_preset=»default» global_colors_info=»{}» theme_builder_area=»et_body_layout»][et_pb_heading title=»Automatiseringsbias og overavhengighet» _builder_version=»4.27.6″ _module_preset=»default» title_level=»h2″ title_text_color=»#060C31″ title_font_size=»32px» title_line_height=»1.3em» custom_margin=»||||false|false» hover_enabled=»0″ global_colors_info=»{}» theme_builder_area=»et_body_layout» sticky_enabled=»0″][\/et_pb_heading][et_pb_text _builder_version=»4.27.6″ _module_preset=»default» text_text_color=»#060C31″ custom_margin=»-10px||||false|false» custom_padding=»0px||||false|false» hover_enabled=»0″ global_colors_info=»{%22gcid-7a729979-e95d-4a37-b4c2-da38eea9cb79%22:%91%22header_text_color%22,%22header_text_color%22%93}» theme_builder_area=»et_body_layout» sticky_enabled=»0″]<\/p>\n

Den kanskje mest undervurderte\u00a0AI-risikoen handler ikke om algoritmer eller teknologi. Den handler om mennesker.<\/span>\u00a0<\/span><\/p>\n

Automatiseringsbias er et etablert begrep innen forskning p\u00e5 beslutningsst\u00f8tte og beskriver menneskers tendens til \u00e5 stole for mye p\u00e5 anbefalinger fra teknologiske systemer.<\/span>\u00a0<\/span><\/p>\n

Dette fenomenet er ikke nytt. Det har v\u00e6rt observert i alt fra luftfart til medisinsk diagnostikk. Men generativ\u00a0AI\u00a0gj\u00f8r problemstillingen mer aktuell enn noen gang.<\/span>\u00a0<\/span><\/p>\n

N\u00e5r en\u00a0AI-modell leverer raske, velformulerte og tilsynelatende intelligente svar, er det naturlig \u00e5 tillegge anbefalingene h\u00f8y verdi. Over tid kan mennesker bli mindre kritiske og bruke mindre tid p\u00e5 egne vurderinger.<\/span>\u00a0<\/span><\/p>\n

Risikoen er derfor ikke n\u00f8dvendigvis at\u00a0AI\u00a0tar beslutninger p\u00e5 egen h\u00e5nd. Risikoen er at mennesker gradvis begynner \u00e5 stole p\u00e5 teknologien uten \u00e5 gjennomf\u00f8re den faglige kvalitetssikringen som situasjonen krever.<\/span>\u00a0<\/span><\/p>\n

Jo bedre\u00a0AI\u00a0blir til \u00e5 kommunisere, desto viktigere blir det \u00e5 huske at overbevisende svar ikke n\u00f8dvendigvis er korrekte svar.<\/span>\u00a0<\/span><\/p>\n

[\/et_pb_text][\/et_pb_column][\/et_pb_row][\/et_pb_section][et_pb_section fb_built=»1″ module_class=»fullwidth-bg» _builder_version=»4.27.6″ _module_preset=»default» background_color=»#FFFFFF» use_background_color_gradient=»off» custom_margin=»||||true|false» custom_padding=»20px||20px||true|false» hover_enabled=»0″ custom_css_free_form=».fullwidth-bg {|| width: 100vw !important;|| position: relative;|| left: 50%;|| right: 50%;|| margin-left: -50vw;|| margin-right: -50vw;|| background-color: #ffeedd !important;|| padding-top: 40px; || padding-bottom: 40px; ||}||||.fullwidth-bg .et_pb_row {|| max-width: 870px; || margin: 0 auto; || padding-left: 0 !important;|| padding-right: 0 !important;||}||||@media (max-width: 768px) {|| .fullwidth-bg .et_pb_row {|| max-width: 500px !important;|| }||}||||@media (max-width: 480px) {|| .fullwidth-bg .et_pb_row {|| max-width: 250px !important;|| }||}» global_colors_info=»{}» theme_builder_area=»et_body_layout» sticky_enabled=»0″][et_pb_row _builder_version=»4.27.5″ _module_preset=»default» custom_padding=»20px||30px||false|true» global_colors_info=»{}» theme_builder_area=»et_body_layout»][et_pb_column type=»4_4″ _builder_version=»4.27.5″ _module_preset=»default» global_colors_info=»{}» theme_builder_area=»et_body_layout»][et_pb_heading title=»En teknologi med nye styrker \u2013 og nye svakheter\u00a0» _builder_version=»4.27.6″ _module_preset=»default» title_level=»h2″ title_text_color=»#060C31″ title_font_size=»32px» title_line_height=»1.3em» custom_margin=»||||false|false» hover_enabled=»0″ global_colors_info=»{}» theme_builder_area=»et_body_layout» sticky_enabled=»0″][\/et_pb_heading][et_pb_text _builder_version=»4.27.6″ _module_preset=»default» text_text_color=»#060C31″ custom_margin=»-10px||||false|false» custom_padding=»0px||||false|false» hover_enabled=»0″ global_colors_info=»{%22gcid-7a729979-e95d-4a37-b4c2-da38eea9cb79%22:%91%22header_text_color%22,%22header_text_color%22%93}» theme_builder_area=»et_body_layout» sticky_enabled=»0″]<\/p>\n

Kunstig intelligens representerer et betydelig teknologisk sprang, men den introduserer ogs\u00e5 et risikobilde som skiller seg fra tradisjonell programvare.<\/span>\u00a0<\/span><\/p>\n

Hallusinasjoner, bias, manglende forklarbarhet og automatiseringsbias er ikke n\u00f8dvendigvis tegn p\u00e5 at teknologien fungerer d\u00e5rlig. Tvert imot er de ofte en konsekvens av hvordan moderne\u00a0AI\u00a0er bygget og hvorfor den er s\u00e5 kraftig.<\/span>\u00a0<\/span><\/p>\n

For ledere handler det derfor ikke om \u00e5 bli eksperter p\u00e5 maskinl\u00e6ring eller nevrale nettverk. Det handler om \u00e5 forst\u00e5 at\u00a0AI\u00a0har andre egenskaper enn systemene virksomheten tradisjonelt har jobbet med.<\/span>\u00a0<\/span><\/p>\n

Jo bedre man forst\u00e5r disse egenskapene, desto bedre rustet er man til \u00e5 vurdere b\u00e5de mulighetene og begrensningene teknologien bringer med seg.<\/span>\u00a0<\/span><\/p>\n

[\/et_pb_text][\/et_pb_column][\/et_pb_row][\/et_pb_section][et_pb_section fb_built=»1″ module_class=»fullwidth-bg» _builder_version=»4.27.6″ _module_preset=»default» background_color=»#FFFFFF» use_background_color_gradient=»off» custom_margin=»||||true|false» custom_padding=»30px||30px||true|false» custom_css_free_form=».fullwidth-bg {|| width: 100vw !important;|| position: relative;|| left: 50%;|| right: 50%;|| margin-left: -50vw;|| margin-right: -50vw;|| background-color: #ffeedd !important;|| padding-top: 40px; || padding-bottom: 40px; ||}||||.fullwidth-bg .et_pb_row {|| max-width: 870px; || margin: 0 auto; || padding-left: 0 !important;|| padding-right: 0 !important;||}||||@media (max-width: 768px) {|| .fullwidth-bg .et_pb_row {|| max-width: 500px !important;|| }||}||||@media (max-width: 480px) {|| .fullwidth-bg .et_pb_row {|| max-width: 250px !important;|| }||}» saved_tabs=»all» global_colors_info=»{}» global_module=»5315″ theme_builder_area=»post_content»][et_pb_row _builder_version=»4.27.6″ _module_preset=»default» background_color=»#EEF3FF» custom_padding=»30px||30px||true|true» border_radii=»on|9px|9px|9px|9px» global_colors_info=»{}» theme_builder_area=»post_content»][et_pb_column type=»4_4″ _builder_version=»4.27.5″ _module_preset=»default» global_colors_info=»{}» theme_builder_area=»post_content»][et_pb_heading title=»Snakk med oss om styringssystemer for AI» _builder_version=»4.27.6″ _module_preset=»default» title_level=»h2″ title_text_align=»center» title_text_color=»#060C31″ title_font_size=»32px» title_line_height=»1.3em» custom_margin=»15px|10px|15px|10px|true|true» custom_padding=»||||false|false» global_colors_info=»{}» theme_builder_area=»post_content»][\/et_pb_heading][et_pb_text _builder_version=»4.27.6″ _module_preset=»default» text_text_color=»#060C31″ text_font_size=»20px» text_line_height=»1.8em» header_font_size=»25px» header_3_font_size=»21px» header_3_line_height=»1.3em» custom_margin=»-10px|20px||20px|false|true» custom_padding=»0px||||false|false» global_colors_info=»{%22gcid-7a729979-e95d-4a37-b4c2-da38eea9cb79%22:%91%22header_text_color%22,%22header_text_color%22%93}» theme_builder_area=»post_content»]<\/p>\n

Vi hjelper virksomheter med r\u00e5dgivning, etablering og videreutvikling av styringssystemer for AI. Ta kontakt for \u00e5 h\u00f8re hvordan vi kan hjelpe deg!<\/p>\n

[\/et_pb_text][et_pb_button button_url=»@ET-DC@eyJkeW5hbWljIjp0cnVlLCJjb250ZW50IjoicG9zdF9saW5rX3VybF9wYWdlIiwic2V0dGluZ3MiOnsicG9zdF9pZCI6IjI5In19@» button_text=»Kontakt oss» button_alignment=»center» admin_label=»Button kontakt oss» _builder_version=»4.27.5″ _dynamic_attributes=»button_url» _module_preset=»default» custom_button=»on» button_text_size=»16px» button_text_color=»#1F3BF4″ button_bg_color=»RGBA(255,255,255,0)» button_border_width=»3px» button_border_color=»#1F3BF4″ button_font=»Open Sans|600|||||||» button_use_icon=»off» custom_padding=»15px|30px|15px|30px|false|false» animation_style=»fade» global_colors_info=»{}» button_text_size__hover_enabled=»on|hover» button_text_size__hover=»18px» theme_builder_area=»post_content»][\/et_pb_button][\/et_pb_column][\/et_pb_row][\/et_pb_section][et_pb_section fb_built=»1″ module_class=»fullwidth-bg» _builder_version=»4.27.6″ _module_preset=»default» background_color=»#FFFFFF» use_background_color_gradient=»off» custom_margin=»||||true|false» custom_padding=»20px||20px||true|false» custom_css_free_form=».fullwidth-bg {|| width: 100vw !important;|| position: relative;|| left: 50%;|| right: 50%;|| margin-left: -50vw;|| margin-right: -50vw;|| background-color: #ffeedd !important;|| padding-top: 40px; || padding-bottom: 40px; ||}||||.fullwidth-bg .et_pb_row {|| max-width: 870px; || margin: 0 auto; || padding-left: 0 !important;|| padding-right: 0 !important;||}||||@media (max-width: 768px) {|| .fullwidth-bg .et_pb_row {|| max-width: 500px !important;|| }||}||||@media (max-width: 480px) {|| .fullwidth-bg .et_pb_row {|| max-width: 250px !important;|| }||}» global_colors_info=»{}» global_module=»5268″ theme_builder_area=»post_content»][et_pb_row _builder_version=»4.27.5″ _module_preset=»default» global_colors_info=»{}» theme_builder_area=»post_content»][et_pb_column type=»4_4″ _builder_version=»4.27.5″ _module_preset=»default» global_colors_info=»{}» theme_builder_area=»post_content»][et_pb_team_member name=»Mirjam Meling» position=»Marketing & Communication Manager» image_url=»https:\/\/certainqms.com\/wp-content\/uploads\/2026\/04\/Mirjam-Meling-Netpower.png» module_class=»author-box» _builder_version=»4.27.6″ _module_preset=»default» header_font_size=»25px» header_line_height=»0.7em» body_line_height=»1.8em» position_line_height=»1.8em» link_option_url=»https:\/\/netpower.no\/mirjam-meling\/» link_option_url_new_window=»on» global_colors_info=»{}» theme_builder_area=»post_content»]<\/p>\n

\n
\n
\n
\n
\n
\n
\n

Produserer innhold for Certain QMS om ledelsessystemer, kvalitetsstyring, informasjonssikkerhet og AI governance. Hun samarbeider med fagspesialister for \u00e5 formidle komplekse temaer p\u00e5 en tydelig og praktisk m\u00e5te.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n

<\/div>\n<\/div>\n<\/div>\n

[\/et_pb_team_member][\/et_pb_column][\/et_pb_row][\/et_pb_section]<\/p>\n","protected":false},"excerpt":{"rendered":"

Hva er de st\u00f8rste AI-risikoene? L\u00e6r om hallusinasjoner, bias, manglende forklarbarhet og andre utfordringer ved kunstig intelligens.<\/p>\n","protected":false},"author":8,"featured_media":5441,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"_et_pb_use_builder":"on","_et_pb_old_content":"","_et_gb_content_width":"","footnotes":""},"categories":[27],"tags":[],"class_list":["post-5425","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ressurser"],"acf":[],"yoast_head":"\nAI-risikoene ledere m\u00e5 forst\u00e5 – Ressurser – Certain<\/title>\n<meta name=\"description\" content=\"Hva er de st\u00f8rste AI-risikoene? L\u00e6r om hallusinasjoner, bias, manglende forklarbarhet og andre utfordringer ved kunstig intelligens.\" \/>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/certainqms.com\/no\/ressurser\/ai-risiko\/\" \/>\n<meta property=\"og:locale\" content=\"nb_NO\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"AI-risikoene ledere m\u00e5 forst\u00e5 – Ressurser – Certain\" \/>\n<meta property=\"og:description\" content=\"Hva er de st\u00f8rste AI-risikoene? L\u00e6r om hallusinasjoner, bias, manglende forklarbarhet og andre utfordringer ved kunstig intelligens.\" \/>\n<meta property=\"og:url\" content=\"https:\/\/certainqms.com\/no\/ressurser\/ai-risiko\/\" \/>\n<meta property=\"og:site_name\" content=\"Certain\" \/>\n<meta property=\"article:published_time\" content=\"2026-06-03T12:14:43+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2026-06-04T03:26:49+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/certainqms.com\/wp-content\/uploads\/2026\/06\/AI-risiko-1.webp\" \/>\n\t<meta property=\"og:image:width\" content=\"1600\" \/>\n\t<meta property=\"og:image:height\" content=\"1000\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/webp\" \/>\n<meta name=\"author\" content=\"Oda Hjelmeland\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"Skrevet av\" \/>\n\t<meta name=\"twitter:data1\" content=\"Oda Hjelmeland\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\\\/\\\/schema.org\",\"@graph\":[{\"@type\":\"Article\",\"@id\":\"https:\\\/\\\/certainqms.com\\\/no\\\/ressurser\\\/ai-risiko\\\/#article\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/certainqms.com\\\/no\\\/ressurser\\\/ai-risiko\\\/\"},\"author\":{\"name\":\"Oda Hjelmeland\",\"@id\":\"https:\\\/\\\/certainqms.com\\\/no\\\/#\\\/schema\\\/person\\\/5da7a414f3256257931a3acdc87ffb54\"},\"headline\":\"Hallusinasjoner, bias og\u00a0black\u00a0boxes:\u00a0AI-risikoene ledere m\u00e5 forst\u00e5\u00a0\",\"datePublished\":\"2026-06-03T12:14:43+00:00\",\"dateModified\":\"2026-06-04T03:26:49+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\\\/\\\/certainqms.com\\\/no\\\/ressurser\\\/ai-risiko\\\/\"},\"wordCount\":3624,\"publisher\":{\"@id\":\"https:\\\/\\\/certainqms.com\\\/no\\\/#organization\"},\"image\":{\"@id\":\"https:\\\/\\\/certainqms.com\\\/no\\\/ressurser\\\/ai-risiko\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/certainqms.com\\\/wp-content\\\/uploads\\\/2026\\\/06\\\/AI-risiko-1.webp\",\"articleSection\":[\"Ressurser\"],\"inLanguage\":\"nb-NO\"},{\"@type\":\"WebPage\",\"@id\":\"https:\\\/\\\/certainqms.com\\\/no\\\/ressurser\\\/ai-risiko\\\/\",\"url\":\"https:\\\/\\\/certainqms.com\\\/no\\\/ressurser\\\/ai-risiko\\\/\",\"name\":\"AI-risikoene ledere m\u00e5 forst\u00e5 – Ressurser – Certain\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/certainqms.com\\\/no\\\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\\\/\\\/certainqms.com\\\/no\\\/ressurser\\\/ai-risiko\\\/#primaryimage\"},\"image\":{\"@id\":\"https:\\\/\\\/certainqms.com\\\/no\\\/ressurser\\\/ai-risiko\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/certainqms.com\\\/wp-content\\\/uploads\\\/2026\\\/06\\\/AI-risiko-1.webp\",\"datePublished\":\"2026-06-03T12:14:43+00:00\",\"dateModified\":\"2026-06-04T03:26:49+00:00\",\"description\":\"Hva er de st\u00f8rste AI-risikoene? L\u00e6r om hallusinasjoner, bias, manglende forklarbarhet og andre utfordringer ved kunstig intelligens.\",\"breadcrumb\":{\"@id\":\"https:\\\/\\\/certainqms.com\\\/no\\\/ressurser\\\/ai-risiko\\\/#breadcrumb\"},\"inLanguage\":\"nb-NO\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\\\/\\\/certainqms.com\\\/no\\\/ressurser\\\/ai-risiko\\\/\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"nb-NO\",\"@id\":\"https:\\\/\\\/certainqms.com\\\/no\\\/ressurser\\\/ai-risiko\\\/#primaryimage\",\"url\":\"https:\\\/\\\/certainqms.com\\\/wp-content\\\/uploads\\\/2026\\\/06\\\/AI-risiko-1.webp\",\"contentUrl\":\"https:\\\/\\\/certainqms.com\\\/wp-content\\\/uploads\\\/2026\\\/06\\\/AI-risiko-1.webp\",\"width\":1600,\"height\":1000,\"caption\":\"AI risiko\"},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\\\/\\\/certainqms.com\\\/no\\\/ressurser\\\/ai-risiko\\\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Certain QMS\",\"item\":\"https:\\\/\\\/certainqms.com\\\/no\\\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"Ressurser\",\"item\":\"https:\\\/\\\/certainqms.com\\\/no\\\/category\\\/ressurser\\\/\"},{\"@type\":\"ListItem\",\"position\":3,\"name\":\"Hallusinasjoner, bias og\u00a0black\u00a0boxes:\u00a0AI-risikoene ledere m\u00e5 forst\u00e5\u00a0\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\\\/\\\/certainqms.com\\\/no\\\/#website\",\"url\":\"https:\\\/\\\/certainqms.com\\\/no\\\/\",\"name\":\"Certain\",\"description\":\"Programvare og r\u00e5dgivning for ledelsessystemer\",\"publisher\":{\"@id\":\"https:\\\/\\\/certainqms.com\\\/no\\\/#organization\"},\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\\\/\\\/certainqms.com\\\/no\\\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"nb-NO\"},{\"@type\":\"Organization\",\"@id\":\"https:\\\/\\\/certainqms.com\\\/no\\\/#organization\",\"name\":\"Netpower Business Solutions AS\",\"alternateName\":\"Netpower\",\"url\":\"https:\\\/\\\/certainqms.com\\\/no\\\/\",\"logo\":{\"@type\":\"ImageObject\",\"inLanguage\":\"nb-NO\",\"@id\":\"https:\\\/\\\/certainqms.com\\\/no\\\/#\\\/schema\\\/logo\\\/image\\\/\",\"url\":\"\",\"contentUrl\":\"\",\"caption\":\"Netpower Business Solutions AS\"},\"image\":{\"@id\":\"https:\\\/\\\/certainqms.com\\\/no\\\/#\\\/schema\\\/logo\\\/image\\\/\"}},{\"@type\":\"Person\",\"@id\":\"https:\\\/\\\/certainqms.com\\\/no\\\/#\\\/schema\\\/person\\\/5da7a414f3256257931a3acdc87ffb54\",\"name\":\"Oda Hjelmeland\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"nb-NO\",\"@id\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/26f5102b9a8cfac61010c52508397a3905dc6fb3efb6ad6471f012501b3622fd?s=96&d=mm&r=g\",\"url\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/26f5102b9a8cfac61010c52508397a3905dc6fb3efb6ad6471f012501b3622fd?s=96&d=mm&r=g\",\"contentUrl\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/26f5102b9a8cfac61010c52508397a3905dc6fb3efb6ad6471f012501b3622fd?s=96&d=mm&r=g\",\"caption\":\"Oda Hjelmeland\"},\"url\":\"https:\\\/\\\/certainqms.com\\\/no\\\/author\\\/oda\\\/\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"AI-risikoene ledere m\u00e5 forst\u00e5 – Ressurser – Certain","description":"Hva er de st\u00f8rste AI-risikoene? L\u00e6r om hallusinasjoner, bias, manglende forklarbarhet og andre utfordringer ved kunstig intelligens.","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:\/\/certainqms.com\/no\/ressurser\/ai-risiko\/","og_locale":"nb_NO","og_type":"article","og_title":"AI-risikoene ledere m\u00e5 forst\u00e5 – Ressurser – Certain","og_description":"Hva er de st\u00f8rste AI-risikoene? L\u00e6r om hallusinasjoner, bias, manglende forklarbarhet og andre utfordringer ved kunstig intelligens.","og_url":"https:\/\/certainqms.com\/no\/ressurser\/ai-risiko\/","og_site_name":"Certain","article_published_time":"2026-06-03T12:14:43+00:00","article_modified_time":"2026-06-04T03:26:49+00:00","og_image":[{"width":1600,"height":1000,"url":"https:\/\/certainqms.com\/wp-content\/uploads\/2026\/06\/AI-risiko-1.webp","type":"image\/webp"}],"author":"Oda Hjelmeland","twitter_card":"summary_large_image","twitter_misc":{"Skrevet av":"Oda Hjelmeland"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/certainqms.com\/no\/ressurser\/ai-risiko\/#article","isPartOf":{"@id":"https:\/\/certainqms.com\/no\/ressurser\/ai-risiko\/"},"author":{"name":"Oda Hjelmeland","@id":"https:\/\/certainqms.com\/no\/#\/schema\/person\/5da7a414f3256257931a3acdc87ffb54"},"headline":"Hallusinasjoner, bias og\u00a0black\u00a0boxes:\u00a0AI-risikoene ledere m\u00e5 forst\u00e5\u00a0","datePublished":"2026-06-03T12:14:43+00:00","dateModified":"2026-06-04T03:26:49+00:00","mainEntityOfPage":{"@id":"https:\/\/certainqms.com\/no\/ressurser\/ai-risiko\/"},"wordCount":3624,"publisher":{"@id":"https:\/\/certainqms.com\/no\/#organization"},"image":{"@id":"https:\/\/certainqms.com\/no\/ressurser\/ai-risiko\/#primaryimage"},"thumbnailUrl":"https:\/\/certainqms.com\/wp-content\/uploads\/2026\/06\/AI-risiko-1.webp","articleSection":["Ressurser"],"inLanguage":"nb-NO"},{"@type":"WebPage","@id":"https:\/\/certainqms.com\/no\/ressurser\/ai-risiko\/","url":"https:\/\/certainqms.com\/no\/ressurser\/ai-risiko\/","name":"AI-risikoene ledere m\u00e5 forst\u00e5 – Ressurser – Certain","isPartOf":{"@id":"https:\/\/certainqms.com\/no\/#website"},"primaryImageOfPage":{"@id":"https:\/\/certainqms.com\/no\/ressurser\/ai-risiko\/#primaryimage"},"image":{"@id":"https:\/\/certainqms.com\/no\/ressurser\/ai-risiko\/#primaryimage"},"thumbnailUrl":"https:\/\/certainqms.com\/wp-content\/uploads\/2026\/06\/AI-risiko-1.webp","datePublished":"2026-06-03T12:14:43+00:00","dateModified":"2026-06-04T03:26:49+00:00","description":"Hva er de st\u00f8rste AI-risikoene? L\u00e6r om hallusinasjoner, bias, manglende forklarbarhet og andre utfordringer ved kunstig intelligens.","breadcrumb":{"@id":"https:\/\/certainqms.com\/no\/ressurser\/ai-risiko\/#breadcrumb"},"inLanguage":"nb-NO","potentialAction":[{"@type":"ReadAction","target":["https:\/\/certainqms.com\/no\/ressurser\/ai-risiko\/"]}]},{"@type":"ImageObject","inLanguage":"nb-NO","@id":"https:\/\/certainqms.com\/no\/ressurser\/ai-risiko\/#primaryimage","url":"https:\/\/certainqms.com\/wp-content\/uploads\/2026\/06\/AI-risiko-1.webp","contentUrl":"https:\/\/certainqms.com\/wp-content\/uploads\/2026\/06\/AI-risiko-1.webp","width":1600,"height":1000,"caption":"AI risiko"},{"@type":"BreadcrumbList","@id":"https:\/\/certainqms.com\/no\/ressurser\/ai-risiko\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Certain QMS","item":"https:\/\/certainqms.com\/no\/"},{"@type":"ListItem","position":2,"name":"Ressurser","item":"https:\/\/certainqms.com\/no\/category\/ressurser\/"},{"@type":"ListItem","position":3,"name":"Hallusinasjoner, bias og\u00a0black\u00a0boxes:\u00a0AI-risikoene ledere m\u00e5 forst\u00e5\u00a0"}]},{"@type":"WebSite","@id":"https:\/\/certainqms.com\/no\/#website","url":"https:\/\/certainqms.com\/no\/","name":"Certain","description":"Programvare og r\u00e5dgivning for ledelsessystemer","publisher":{"@id":"https:\/\/certainqms.com\/no\/#organization"},"potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/certainqms.com\/no\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"nb-NO"},{"@type":"Organization","@id":"https:\/\/certainqms.com\/no\/#organization","name":"Netpower Business Solutions AS","alternateName":"Netpower","url":"https:\/\/certainqms.com\/no\/","logo":{"@type":"ImageObject","inLanguage":"nb-NO","@id":"https:\/\/certainqms.com\/no\/#\/schema\/logo\/image\/","url":"","contentUrl":"","caption":"Netpower Business Solutions AS"},"image":{"@id":"https:\/\/certainqms.com\/no\/#\/schema\/logo\/image\/"}},{"@type":"Person","@id":"https:\/\/certainqms.com\/no\/#\/schema\/person\/5da7a414f3256257931a3acdc87ffb54","name":"Oda Hjelmeland","image":{"@type":"ImageObject","inLanguage":"nb-NO","@id":"https:\/\/secure.gravatar.com\/avatar\/26f5102b9a8cfac61010c52508397a3905dc6fb3efb6ad6471f012501b3622fd?s=96&d=mm&r=g","url":"https:\/\/secure.gravatar.com\/avatar\/26f5102b9a8cfac61010c52508397a3905dc6fb3efb6ad6471f012501b3622fd?s=96&d=mm&r=g","contentUrl":"https:\/\/secure.gravatar.com\/avatar\/26f5102b9a8cfac61010c52508397a3905dc6fb3efb6ad6471f012501b3622fd?s=96&d=mm&r=g","caption":"Oda Hjelmeland"},"url":"https:\/\/certainqms.com\/no\/author\/oda\/"}]}},"_links":{"self":[{"href":"https:\/\/certainqms.com\/no\/wp-json\/wp\/v2\/posts\/5425","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/certainqms.com\/no\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/certainqms.com\/no\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/certainqms.com\/no\/wp-json\/wp\/v2\/users\/8"}],"replies":[{"embeddable":true,"href":"https:\/\/certainqms.com\/no\/wp-json\/wp\/v2\/comments?post=5425"}],"version-history":[{"count":17,"href":"https:\/\/certainqms.com\/no\/wp-json\/wp\/v2\/posts\/5425\/revisions"}],"predecessor-version":[{"id":5467,"href":"https:\/\/certainqms.com\/no\/wp-json\/wp\/v2\/posts\/5425\/revisions\/5467"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/certainqms.com\/no\/wp-json\/wp\/v2\/media\/5441"}],"wp:attachment":[{"href":"https:\/\/certainqms.com\/no\/wp-json\/wp\/v2\/media?parent=5425"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/certainqms.com\/no\/wp-json\/wp\/v2\/categories?post=5425"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/certainqms.com\/no\/wp-json\/wp\/v2\/tags?post=5425"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}