Jakarta, CNBC Indonesia \u2014 The Ministry of Communication and Digital\nAffairs (Komdigi) says chatbot responses are not always accurate and has\nurged the public to check them again.<\/p>\n
\u201cAI does not necessarily contain the truth. That is why human\nintervention is important: people need to check and recheck, and confirm\ninformation against verified sources, including verified online media\nand other media,\u201d said Marroli Jeni Indarto, director of institutional\ncommunications partnerships and public relations at Komdigi, at the\nministry\u2019s office on Thursday (29 October 2026).<\/p>\n
Human oversight is needed after using AI. This means checking the\naccuracy of information, including by verifying it against established\nmedia sources.<\/p>\n
AI models should be used to help people think. Decisions and\njudgements must remain in human hands.<\/p>\n
\u201cThe point is that AI may help us think, but decisions and judgements\nmust remain with its users,\u201d Marroli said.<\/p>\n
He added that platforms themselves acknowledge that chatbots can make\nmistakes, meaning the possibility of inaccurate answers remains\nsignificant.<\/p>\n
\u201cAI can make mistakes. There is still a possibility that the platform\nwill not be entirely accurate. That is why users need to check and\nrecheck,\u201d he said.<\/p>\n
Previously, an answer from Meta AI circulated claiming that\nrestrictions on content about demonstrations and forest and land fires\n(karhutla) had been imposed at Komdigi\u2019s request. The claim went viral\non social media, where many users shared it in connection with\ncomplaints that their content had been taken down or shadowbanned.<\/p>", "url": "https:\/\/jawawa.id\/newsitem\/viral-meta-ai-response-prompts-komdigi-warning-that-chatbots-can-be-wrong-1793260275", "image": "" }, "sponsor": "Okusi Associates", "sponsor_url": "https:\/\/okusiassociates.com" }