{"id":1960,"date":"2026-06-18T19:06:50","date_gmt":"2026-06-18T19:06:50","guid":{"rendered":"https:\/\/1975mag.com\/?p=1960"},"modified":"2026-06-18T19:07:30","modified_gmt":"2026-06-18T19:07:30","slug":"mo-hinh-the-gioi-world-models-huong-di-moi-cua-tri-tue-nhan-tao-ai","status":"publish","type":"post","link":"https:\/\/1975mag.com\/ro\/mo-hinh-the-gioi-world-models-huong-di-moi-cua-tri-tue-nhan-tao-ai\/","title":{"rendered":"M\u00f4 H\u00ecnh Th\u1ebf Gi\u1edbi (World Models): H\u01b0\u1edbng \u0110i M\u1edbi C\u1ee7a Tr\u00ed Tu\u1ec7 Nh\u00e2n T\u1ea1o (AI)"},"content":{"rendered":"
\"\"
Image by Kyrasys@pixabay.com<\/span><\/em><\/figcaption><\/figure>\n

T\u00f3m T\u1eaft<\/strong><\/span><\/h3>\n

Tr\u00ed tu\u1ec7 nh\u00e2n t\u1ea1o (AI) \u0111ang tr\u1ea3i qua m\u1ed9t b\u01b0\u1edbc ngo\u1eb7t quan tr\u1ecdng. Sau nhi\u1ec1u n\u0103m th\u1ed1ng tr\u1ecb b\u1edfi c\u00e1c M\u00f4 H\u00ecnh Ng\u00f4n Ng\u1eef L\u1edbn<\/em> (Large Language Models – LLM) nh\u01b0 ChatGPT, Claude, hay Gemini, nhi\u1ec1u nh\u00e0 nghi\u00ean c\u1ee9u h\u00e0ng \u0111\u1ea7u th\u1ebf gi\u1edbi \u0111ang \u0111\u1eb7t c\u00e2u h\u1ecfi: li\u1ec7u ng\u00f4n ng\u1eef c\u00f3 th\u1ef1c s\u1ef1 l\u00e0 con \u0111\u01b0\u1eddng d\u1eabn \u0111\u1ebfn tr\u00ed th\u00f4ng minh t\u1ed5ng qu\u00e1t<\/em> (AGI)? C\u00e2u tr\u1ea3 l\u1eddi ng\u00e0y c\u00e0ng nghi\u00eang v\u1ec1 ph\u00eda “kh\u00f4ng” \u2013 v\u00e0 h\u01b0\u1edbng \u0111i m\u1edbi \u0111ang \u0111\u01b0\u1ee3c h\u00e0ng t\u1ef7 \u0111\u00f4 la \u0111\u1ea7u t\u01b0 h\u01b0\u1edbng \u0111\u1ebfn ch\u00ednh l\u00e0 m\u00f4 h\u00ecnh th\u1ebf gi\u1edbi (world models)<\/strong>: c\u00e1c h\u1ec7 th\u1ed1ng AI c\u00f3 kh\u1ea3 n\u0103ng m\u00f4 ph\u1ecfng, hi\u1ec3u v\u00e0 d\u1ef1 \u0111o\u00e1n th\u1ebf gi\u1edbi v\u1eadt l\u00fd th\u1ef1c s\u1ef1 xung quanh ch\u00fang ta.<\/p>\n

1. B\u1ed1i C\u1ea3nh: K\u1ef7 Nguy\u00ean B\u00f9ng N\u1ed5 \u0110\u1ea7u T\u01b0 V\u00e0o AI<\/strong><\/span><\/h3>\n

\u0110\u1ec3 hi\u1ec3u t\u1ea1i sao \u201cworld models\u201d \u0111ang thu h\u00fat s\u1ef1 ch\u00fa \u00fd l\u1edbn nh\u01b0 v\u1eady, c\u1ea7n nh\u00ecn v\u00e0o b\u1ee9c tranh \u0111\u1ea7u t\u01b0 AI to\u00e0n c\u1ea7u. Theo d\u1eef li\u1ec7u t\u1eeb c\u00f4ng ty ph\u00e2n t\u00edch Quid, n\u0103m 2025 thi\u1ebft l\u1eadp k\u1ef7 l\u1ee5c m\u1edbi v\u1edbi h\u01a1n 581 t\u1ef7 USD<\/strong> \u0111\u1ea7u t\u01b0 v\u00e0o AI \u2013 g\u1ea5p \u0111\u00f4i con s\u1ed1 253 t\u1ef7 USD c\u1ee7a n\u0103m 2024. Hoa K\u1ef3 d\u1eabn \u0111\u1ea7u v\u1edbi h\u01a1n 344 t\u1ef7 USD, chi\u1ebfm g\u1ea7n 60% t\u1ed5ng \u0111\u1ea7u t\u01b0 to\u00e0n c\u1ea7u.<\/p>\n

Tuy nhi\u00ean, \u0111\u1eb1ng sau con s\u1ed1 g\u00e2y nhi\u1ec1u \u1ea5n t\u01b0\u1ee3ng \u0111\u00f3 l\u00e0 m\u1ed9t nh\u1eadn th\u1ee9c ng\u00e0y c\u00e0ng r\u00f5 r\u00e0ng trong gi\u1edbi nghi\u00ean c\u1ee9u: c\u00e1c m\u00f4 h\u00ecnh AI hi\u1ec7n t\u1ea1i, d\u00f9 m\u1ea1nh m\u1ebd \u0111\u1ebfn \u0111\u00e2u, v\u1eabn \u0111ang thi\u1ebfu m\u1ed9t th\u1ee9 g\u00ec \u0111\u00f3 r\u1ea5t c\u01a1 b\u1ea3n, \u0111\u00f3 l\u00e0: kh\u1ea3 n\u0103ng nh\u1eadn hi\u1ec3u th\u1ebf gi\u1edbi th\u1ef1c<\/em>.<\/p>\n

2. Gi\u1edbi H\u1ea1n C\u1ee7a C\u00e1c M\u00f4 H\u00ecnh AI Hi\u1ec7n T\u1ea1i (LLM)<\/strong><\/span><\/h3>\n

2.1 AI Bi\u1ebft N\u00f3i Nh\u01b0ng Kh\u00f4ng Nh\u00ecn Th\u1ea5y<\/strong><\/p>\n

H\u00e3y ngh\u0129 \u0111\u1ebfn m\u1ed9t \u0111\u1ee9a tr\u1ebb m\u1ed9t tu\u1ed5i ch\u01a1i tr\u00f2 \u00fa tim. Khi ng\u01b0\u1eddi l\u1edbn che m\u1eb7t r\u1ed3i b\u1ea5t ng\u1edd hi\u1ec7n ra, \u0111\u1ee9a tr\u1ebb kh\u00f4ng ch\u1ec9 gi\u1eadt m\u00ecnh \u2013 n\u00f3 \u0111ang h\u1ecdc r\u1eb1ng th\u1ee9 g\u00ec \u0111\u00f3 v\u1eabn t\u1ed3n t\u1ea1i d\u00f9 kh\u00f4ng nh\u00ecn th\u1ea5y \u0111\u01b0\u1ee3c. \u0110\u00f3 l\u00e0 n\u1ec1n t\u1ea3ng c\u1ee7a nh\u1eadn th\u1ee9c v\u1eadt l\u00fd: s\u1ef1 v\u1eadt ti\u1ebfp t\u1ee5c t\u1ed3n t\u1ea1i v\u00e0 v\u1eadn \u0111\u1ed9ng ngay c\u1ea3 khi ch\u00fang ta kh\u00f4ng quan s\u00e1t ch\u00fang<\/strong>.<\/p>\n

C\u00e1c h\u1ec7 th\u1ed1ng AI hi\u1ec7n nay, d\u00f9 c\u00f3 kh\u1ea3 n\u0103ng tr\u00f2 chuy\u1ec7n v\u00e0 nh\u1eadn d\u1ea1ng m\u1eabu \u0111\u00e1ng kinh ng\u1ea1c, kh\u00f4ng th\u1ec3 l\u00e0m \u0111i\u1ec1u n\u00e0y m\u1ed9t c\u00e1ch \u0111\u00e1ng tin c\u1eady. Ch\u00fang c\u00f3 th\u1ec3 m\u00f4 t\u1ea3 nh\u1eefng g\u00ec \u0111ang hi\u1ec7n di\u1ec7n tr\u01b0\u1edbc m\u1eaft, nh\u01b0ng l\u1ea1i g\u1eb7p kh\u00f3 kh\u0103n v\u1edbi c\u00e1c kh\u00e1i ni\u1ec7m nh\u01b0 v\u1eadt th\u1ec3 b\u1ecb che khu\u1ea5t l\u00e0 g\u00ec, ho\u1eb7c \u0111i\u1ec1u g\u00ec s\u1ebd x\u1ea3y ra ti\u1ebfp theo trong m\u1ed9t chu\u1ed7i h\u00e0nh \u0111\u1ed9ng v\u1eadt l\u00fd.<\/p>\n

2.2 LLM: Thao T\u00fang Ng\u00f4n Ng\u1eef, Kh\u00f4ng Ph\u1ea3i Hi\u1ec3u Th\u1ef1c T\u1ea1i<\/strong><\/p>\n

Yann LeCun, ng\u01b0\u1eddi \u0111\u01b0\u1ee3c m\u1ec7nh danh l\u00e0 “cha \u0111\u1ee1 \u0111\u1ea7u c\u1ee7a AI”, \u0111\u00e3 ph\u00e1t bi\u1ec3u th\u1eb3ng th\u1eafn trong m\u1ed9t b\u00e0i gi\u1ea3ng g\u1ea7n \u0111\u00e2y:<\/p>\n

“Ch\u00fang ta c\u00f3 nh\u1eefng h\u1ec7 th\u1ed1ng c\u00f3 th\u1ec3 thao t\u00fang ng\u00f4n ng\u1eef, v\u00e0 ch\u00fang \u0111\u00e1nh l\u1eeba ch\u00fang ta khi\u1ebfn ch\u00fang ta ngh\u0129 r\u1eb1ng ch\u00fang th\u00f4ng minh v\u00ec ch\u00fang thao t\u00fang ng\u00f4n ng\u1eef. Nh\u01b0ng th\u1ef1c ra, ch\u00fang ho\u00e0n to\u00e0n b\u1ea5t l\u1ef1c khi \u0111\u1ed1i m\u1eb7t v\u1edbi th\u1ebf gi\u1edbi v\u1eadt l\u00fd.”<\/em><\/p>\n

\u0110\u00e2y l\u00e0 v\u1ea5n \u0111\u1ec1 c\u1ed1t l\u00f5i c\u1ee7a LLM: ch\u00fang \u0111\u01b0\u1ee3c hu\u1ea5n luy\u1ec7n tr\u00ean h\u00e0ng t\u1ef7 t\u1eeb ng\u1eef l\u1ea5y t\u1eeb internet, nh\u01b0ng t\u1eeb ng\u1eef ch\u1ec9 l\u00e0 bi\u1ec3u t\u01b0\u1ee3ng<\/strong> c\u1ee7a th\u1ef1c t\u1ea1i, kh\u00f4ng ph\u1ea3i b\u1ea3n th\u00e2n th\u1ef1c t\u1ea1i. M\u1ed9t m\u00f4 h\u00ecnh ng\u00f4n ng\u1eef c\u00f3 th\u1ec3 gi\u1ea3i th\u00edch ho\u00e0n h\u1ea3o v\u1ec1 tr\u1ecdng l\u1ef1c, nh\u01b0ng kh\u00f4ng “bi\u1ebft” r\u1eb1ng m\u1ed9t qu\u1ea3 b\u00f3ng s\u1ebd r\u01a1i xu\u1ed1ng n\u1ebfu kh\u00f4ng c\u00f3 tay \u0111\u1ee1.<\/p>\n

2.3 Nh\u1eefng Gi\u1edbi H\u1ea1n C\u1ee5 Th\u1ec3<\/strong><\/p>\n

C\u00e1c gi\u1edbi h\u1ea1n c\u1ee7a LLM th\u1ec3 hi\u1ec7n r\u00f5 nh\u1ea5t trong ba l\u0129nh v\u1ef1c quan tr\u1ecdng:<\/p>\n

Robotics v\u00e0 t\u1ef1 \u0111\u1ed9ng h\u00f3a v\u1eadt l\u00fd:<\/strong> M\u1ed9t robot lau nh\u00e0 c\u1ea7n hi\u1ec3u r\u1eb1ng khi n\u00f3 \u0111\u1ea9y m\u1ed9t chi\u1ebfc gh\u1ebf, chi\u1ebfc gh\u1ebf s\u1ebd di chuy\u1ec3n theo m\u1ed9t h\u01b0\u1edbng nh\u1ea5t \u0111\u1ecbnh v\u00e0 c\u00f3 th\u1ec3 \u0111\u1ed5 xu\u1ed1ng. LLM kh\u00f4ng cung c\u1ea5p \u0111\u01b0\u1ee3c lo\u1ea1i hi\u1ec3u bi\u1ebft nh\u00e2n qu\u1ea3 n\u00e0y.<\/p>\n

Xe t\u1ef1 l\u00e1i:<\/strong> \u0110\u1ec3 an to\u00e0n, xe t\u1ef1 l\u00e1i c\u1ea7n d\u1ef1 \u0111o\u00e1n: “N\u1ebfu \u0111\u1ee9a tr\u1ebb kia ti\u1ebfp t\u1ee5c ch\u1ea1y theo h\u01b0\u1edbng \u0111\u00f3, 2 gi\u00e2y n\u1eefa n\u00f3 s\u1ebd b\u01b0\u1edbc ra \u0111\u01b0\u1eddng.” \u0110\u00e2y l\u00e0 b\u00e0i to\u00e1n d\u1ef1 \u0111o\u00e1n t\u01b0\u01a1ng lai v\u1eadt l\u00fd m\u00e0 LLM kh\u00f4ng \u0111\u01b0\u1ee3c thi\u1ebft k\u1ebf \u0111\u1ec3 gi\u1ea3i quy\u1ebft.<\/p>\n

Y t\u1ebf v\u00e0 ph\u1eabu thu\u1eadt:<\/strong> M\u00f4 ph\u1ecfng quy tr\u00ecnh ph\u1eabu thu\u1eadt \u0111\u00f2i h\u1ecfi hi\u1ec3u bi\u1ebft v\u1ec1 c\u1ea5u tr\u00fac gi\u1ea3i ph\u1eabu ba chi\u1ec1u (3D), t\u00ednh ch\u1ea5t v\u1eadt li\u1ec7u c\u1ee7a m\u00f4, v\u00e0 ph\u1ea3n \u1ee9ng c\u1ee7a c\u01a1 th\u1ec3 v\u1edbi t\u1eebng t\u00e1c \u0111\u1ed9ng \u2013 nh\u1eefng th\u1ee9 kh\u00f4ng th\u1ec3 h\u1ecdc t\u1eeb v\u0103n b\u1ea3n \u0111\u01a1n thu\u1ea7n.<\/p>\n

3. World Models L\u00e0 G\u00ec?<\/strong><\/span><\/h3>\n

\"\"<\/p>\n

3.1 \u0110\u1ecbnh Ngh\u0129a<\/strong><\/p>\n

M\u00f4 H\u00ecnh Th\u1ebf Gi\u1edbi<\/em> (world models) l\u00e0 c\u00e1c h\u1ec7 th\u1ed1ng AI \u0111\u01b0\u1ee3c thi\u1ebft k\u1ebf kh\u00f4ng ch\u1ec9 \u0111\u1ec3 nh\u1eadn d\u1ea1ng c\u00e1c m\u1eabu trong v\u0103n b\u1ea3n hay h\u00ecnh \u1ea3nh, m\u00e0 c\u00f2n \u0111\u1ec3 m\u00f4 ph\u1ecfng c\u00e1ch th\u1ebf gi\u1edbi v\u1eadt l\u00fd ho\u1ea1t \u0111\u1ed9ng<\/strong>. B\u1eb1ng c\u00e1ch hu\u1ea5n luy\u1ec7n tr\u00ean h\u00e0ng tri\u1ec7u gi\u1edd video, c\u00e1c m\u00f4 h\u00ecnh n\u00e0y x\u00e2y d\u1ef1ng m\u1ed9t phi\u00ean b\u1ea3n h\u00ecnh \u1ea3nh\u00a0 ch\u00ednh x\u00e1c v\u1ec1 c\u00e1ch th\u1ebf gi\u1edbi v\u1eadn h\u00e0nh \u2013 bao g\u1ed3m v\u1eadt l\u00fd, quan h\u1ec7 nh\u00e2n qu\u1ea3, v\u00e0 di\u1ec5n ti\u1ebfn theo th\u1eddi gian.<\/p>\n

N\u00f3i m\u1ed9t c\u00e1ch \u0111\u01a1n gi\u1ea3n: n\u1ebfu LLM h\u1ecdc t\u1eeb nh\u1eefng g\u00ec ng\u01b0\u1eddi ta n\u00f3i v\u1ec1 th\u1ebf gi\u1edbi<\/strong>, th\u00ec world models h\u1ecdc t\u1eeb nh\u1eefng g\u00ec th\u1ef1c s\u1ef1 x\u1ea3y ra trong th\u1ebf gi\u1edbi<\/strong>.<\/p>\n

3.2 Ngu\u1ed3n G\u1ed1c H\u1ecdc Thu\u1eadt<\/strong><\/p>\n

Ph\u1ea7n l\u1edbn c\u00e1c n\u1ed7 l\u1ef1c x\u00e2y d\u1ef1ng world models<\/em> ng\u00e0y nay c\u00f3 th\u1ec3 truy nguy\u00ean v\u1ec1 m\u1ed9t b\u00e0i b\u00e1o n\u0103m 2018 c\u1ee7a David Ha v\u00e0 J\u00fcrgen Schmidhuber. B\u00e0i b\u00e1o \u0111\u1ec1 xu\u1ea5t m\u1ed9t c\u00e1ch ti\u1ebfp c\u1eadn kh\u00e1c bi\u1ec7t: tr\u01b0\u1edbc khi AI c\u00f3 th\u1ec3 h\u00e0nh \u0111\u1ed9ng th\u00f4ng minh, n\u00f3 c\u1ea7n h\u1ecdc c\u00e1ch th\u1ebf gi\u1edbi ho\u1ea1t \u0111\u1ed9ng th\u00f4ng qua m\u1ed9t phi\u00ean b\u1ea3n m\u00f4 ph\u1ecfng c\u1ee7a m\u00f4i tr\u01b0\u1eddng \u0111\u00f3<\/strong>.<\/p>\n

David Ha, hi\u1ec7n l\u00e0 \u0111\u1ed3ng s\u00e1ng l\u1eadp c\u00f4ng ty AI Nh\u1eadt B\u1ea3n Sakana, m\u00f4 t\u1ea3 c\u01a1 ch\u1ebf n\u00e0y l\u00e0 cho ph\u00e9p AI luy\u1ec7n t\u1eadp b\u00ean trong c\u00e1c phi\u00ean b\u1ea3n m\u00f4 ph\u1ecfng th\u1ef1c t\u1ea1i \u2013 c\u00e1i m\u00e0 \u00f4ng g\u1ecdi l\u00e0 “nh\u1eefng gi\u1ea5c m\u01a1 \u1ea3o gi\u00e1c”<\/em> \u2013 n\u01a1i m\u1ed9t t\u00e1c nh\u00e2n c\u00f3 th\u1ec3 th\u1ef1c h\u00e0nh v\u00e0 l\u1eadp k\u1ebf ho\u1ea1ch tr\u01b0\u1edbc khi h\u00e0nh \u0111\u1ed9ng trong th\u1ebf gi\u1edbi th\u1ef1c.<\/p>\n

3.3 Ph\u00e9p \u1ea8n D\u1ee5 T\u1eeb \u0110i\u1ec7n \u1ea2nh<\/strong><\/p>\n

Ming-Yu Liu, Ph\u00f3 Ch\u1ee7 t\u1ecbch t\u1ea1i Cosmos Lab c\u1ee7a Nvidia, \u0111\u00e3 \u0111\u01b0a ra m\u1ed9t ph\u00e9p so s\u00e1nh sinh \u0111\u1ed9ng: h\u00e3y ngh\u0129 \u0111\u1ebfn b\u1ed9 phim The Matrix<\/em>, trong \u0111\u00f3 nh\u00e2n v\u1eadt ch\u00ednh h\u1ecdc v\u00f5 thu\u1eadt b\u00ean trong m\u1ed9t th\u1ebf gi\u1edbi m\u00f4 ph\u1ecfng. World models ch\u00ednh l\u00e0 “c\u01a1 s\u1edf hu\u1ea5n luy\u1ec7n t\u1ea1o sinh” nh\u01b0 v\u1eady \u2013 “c\u00f3 ph\u1ea3n h\u1ed3i v\u00e0 h\u01b0\u1edbng d\u1eabn \u0111\u1ec3 AI li\u00ean t\u1ee5c n\u00e2ng cao k\u1ef9 n\u0103ng c\u1ee7a m\u00ecnh.”<\/p>\n

4. Nh\u1eefng T\u00ean Tu\u1ed5i L\u1edbn \u0110ang \u0110\u1ea7u T\u01b0 H\u00e0ng T\u1ef7 \u0110\u00f4<\/strong><\/span><\/h3>\n

4.1 Fei-Fei Li v\u00e0 World Labs<\/strong><\/p>\n

Fei-Fei Li, gi\u00e1o s\u01b0 \u0110\u1ea1i h\u1ecdc Stanford v\u00e0 \u0111\u01b0\u1ee3c m\u1ec7nh danh l\u00e0 “m\u1eb9 \u0111\u1ee1 \u0111\u1ea7u c\u1ee7a AI”, \u0111\u00e3 s\u00e1ng l\u1eadp World Labs<\/strong> v\u00e0o n\u0103m 2024. M\u1ee5c ti\u00eau c\u1ee7a startup n\u00e0y l\u00e0 mang l\u1ea1i cho AI nh\u1eadn th\u1ee9c phong ph\u00fa h\u01a1n v\u1ec1 kh\u00f4ng gian 3D v\u00e0 v\u1ec1 c\u00e1ch c\u00e1c v\u1eadt th\u1ec3 t\u1ed3n t\u1ea1i v\u00e0 t\u01b0\u01a1ng t\u00e1c trong \u0111\u00f3. T\u00ednh \u0111\u1ebfn \u0111\u1ea7u n\u0103m 2026, World Labs \u0111\u00e3 huy \u0111\u1ed9ng \u0111\u01b0\u1ee3c kho\u1ea3ng 1 t\u1ef7 USD<\/strong> v\u1ed1n \u0111\u1ea7u t\u01b0.<\/p>\n

4.2 Yann LeCun v\u00e0 AMI Labs<\/strong><\/p>\n

Sau 12 n\u0103m g\u1eafn b\u00f3 v\u1edbi Meta (7 n\u0103m v\u1edbi t\u01b0 c\u00e1ch Tr\u01b0\u1edfng nh\u00e0 khoa h\u1ecdc AI), Yann LeCun \u0111\u00e3 r\u1eddi c\u00f4ng ty v\u00e0o th\u00e1ng 12\/2025 \u0111\u1ec3 th\u00e0nh l\u1eadp Advanced Machine Intelligence (AMI) Labs<\/strong>. \u0110\u00e2y l\u00e0 n\u1ed7 l\u1ef1c tham v\u1ecdng nh\u1ea5t t\u1eeb tr\u01b0\u1edbc \u0111\u1ebfn nay nh\u1eb1m th\u01b0\u01a1ng m\u1ea1i h\u00f3a nghi\u00ean c\u1ee9u v\u1ec1 world models.<\/p>\n

AMI Labs b\u01b0\u1edbc v\u00e0o v\u00f2ng g\u1ecdi v\u1ed1n v\u1edbi m\u1ee9c \u0111\u1ecbnh gi\u00e1 3 t\u1ef7 euro tr\u01b0\u1edbc khi ra m\u1eaft b\u1ea5t k\u1ef3 s\u1ea3n ph\u1ea9m n\u00e0o<\/strong> \u2013 m\u1ed9t trong nh\u1eefng \u0111\u1ee3t g\u1ecdi v\u1ed1n tr\u01b0\u1edbc ra m\u1eaft l\u1edbn nh\u1ea5t trong l\u1ecbch s\u1eed AI. C\u00f4ng ty \u0111\u1eb7t m\u1ee5c ti\u00eau t\u1ea1o ra c\u00e1c h\u1ec7 th\u1ed1ng AI c\u00f3 th\u1ec3: Hi\u1ec3u v\u1eadt l\u00fd v\u00e0 t\u00ednh nh\u00e2n qu\u1ea3; duy tr\u00ec b\u1ed9 nh\u1edb li\u00ean t\u1ee5c (persistent memory); l\u1eadp k\u1ebf ho\u1ea1ch cho c\u00e1c h\u00e0nh \u0111\u1ed9ng ph\u1ee9c t\u1ea1p – thay v\u00ec ch\u1ec9 \u0111\u01a1n gi\u1ea3n d\u1ef1 \u0111o\u00e1n chu\u1ed7i v\u0103n b\u1ea3n.<\/p>\n

4.3 Google v\u00e0 Project Genie<\/strong><\/p>\n

Google g\u1ea7n \u0111\u00e2y \u0111\u00e3 ti\u1ebft l\u1ed9 b\u1ea3n xem tr\u01b0\u1edbc nghi\u00ean c\u1ee9u c\u00f3 t\u00ean Project Genie<\/strong> \u2013 c\u00f3 kh\u1ea3 n\u0103ng t\u1ea1o ra c\u00e1c m\u00f4i tr\u01b0\u1eddng t\u01b0\u01a1ng t\u00e1c, si\u00eau th\u1ef1c t\u1eeb c\u00e1c g\u1ee3i \u00fd \u0111\u01a1n gi\u1ea3n, sau \u0111\u00f3 d\u1ef1 \u0111o\u00e1n c\u00e1ch nh\u1eefng th\u1ebf gi\u1edbi \u0111\u00f3 ph\u00e1t tri\u1ec3n v\u00e0 ph\u1ea3n h\u1ed3i v\u1edbi h\u00e0nh \u0111\u1ed9ng c\u1ee7a ng\u01b0\u1eddi d\u00f9ng.<\/p>\n

4.4 Alibaba v\u00e0 Shengshu\/Vidu<\/strong><\/p>\n

T\u1eeb ph\u00eda Trung Qu\u1ed1c, Alibaba Cloud d\u1eabn \u0111\u1ea7u m\u1ed9t kho\u1ea3n \u0111\u1ea7u t\u01b0 tr\u1ecb gi\u00e1 2 t\u1ef7 nh\u00e2n d\u00e2n t\u1ec7<\/strong> (kho\u1ea3ng 290 tri\u1ec7u USD) v\u00e0o ShengShu \u2013 startup \u0111\u1ee9ng sau c\u00f4ng c\u1ee5 t\u1ea1o video AI Vidu. Kho\u1ea3n \u0111\u1ea7u t\u01b0 n\u00e0y nh\u1eb1m v\u01b0\u1ee3t ra ngo\u00e0i AI d\u1ef1a tr\u00ean v\u0103n b\u1ea3n \u0111\u1ec3 h\u01b0\u1edbng \u0111\u1ebfn c\u00f4ng ngh\u1ec7 m\u00f4 ph\u1ecfng th\u1ef1c t\u1ebf, m\u1ed9t h\u01b0\u1edbng \u0111i c\u1ea7n thi\u1ebft \u0111\u1ec3 th\u00fac \u0111\u1ea9y robotics.<\/p>\n

5. \u1ee8ng D\u1ee5ng Th\u1ef1c Ti\u1ec5n<\/strong><\/span><\/h3>\n

World models kh\u00f4ng ch\u1ec9 l\u00e0 l\u00fd thuy\u1ebft nghi\u00ean c\u1ee9u \u2013 ch\u00fang h\u01b0\u1edbng \u0111\u1ebfn nh\u1eefng \u1ee9ng d\u1ee5ng c\u1ee5 th\u1ec3 v\u00e0 c\u00f3 t\u00e1c \u0111\u1ed9ng l\u1edbn:<\/p>\n

Xe t\u1ef1 h\u00e0nh:<\/strong> D\u1ef1 \u0111o\u00e1n h\u00e0nh vi c\u1ee7a ng\u01b0\u1eddi \u0111i b\u1ed9, xe \u0111\u1ea1p v\u00e0 c\u00e1c ph\u01b0\u01a1ng ti\u1ec7n kh\u00e1c trong t\u00ecnh hu\u1ed1ng ph\u1ee9c t\u1ea1p, bao g\u1ed3m c\u00e1c k\u1ecbch b\u1ea3n ch\u01b0a t\u1eebng g\u1eb7p trong d\u1eef li\u1ec7u hu\u1ea5n luy\u1ec7n.<\/p>\n

Robotics gia \u0111\u00ecnh v\u00e0 c\u00f4ng nghi\u1ec7p:<\/strong> D\u1ea1y robot c\u00e1c k\u1ef9 n\u0103ng m\u1edbi ch\u1ec9 t\u1eeb m\u1ed9t v\u00e0i v\u00ed d\u1ee5 \u2013 k\u1ec3 c\u1ea3 nh\u1eefng k\u1ef9 n\u0103ng m\u00e0 robot ch\u01b0a bao gi\u1edd th\u1ef1c hi\u1ec7n tr\u01b0\u1edbc \u0111\u00f3 \u2013 r\u1ed3i th\u1ef1c hi\u1ec7n ch\u00fang m\u1ed9t c\u00e1ch nh\u1ea5t qu\u00e1n.<\/p>\n

Ph\u1eabu thu\u1eadt v\u00e0 y t\u1ebf:<\/strong> M\u00f4 ph\u1ecfng quy tr\u00ecnh ph\u1eabu thu\u1eadt tr\u01b0\u1edbc khi th\u1ef1c hi\u1ec7n, d\u1ef1 \u0111o\u00e1n ph\u1ea3n \u1ee9ng c\u1ee7a c\u01a1 th\u1ec3.<\/p>\n

Game v\u00e0 gi\u1ea3i tr\u00ed t\u01b0\u01a1ng t\u00e1c:<\/strong> T\u1ea1o ra c\u00e1c th\u1ebf gi\u1edbi game ph\u1ea3n \u1ee9ng v\u00e0 thay \u0111\u1ed5i theo th\u1eddi gian th\u1ef1c d\u1ef1a tr\u00ean h\u00e0nh \u0111\u1ed9ng c\u1ee7a ng\u01b0\u1eddi ch\u01a1i.<\/p>\n

Gi\u00e1o d\u1ee5c v\u00e0 \u0111\u00e0o t\u1ea1o:<\/strong> M\u00f4i tr\u01b0\u1eddng m\u00f4 ph\u1ecfng \u0111\u1ec3 luy\u1ec7n t\u1eadp c\u00e1c k\u1ef9 n\u0103ng v\u1eadt l\u00fd trong kh\u00f4ng gian \u1ea3o an to\u00e0n.<\/p>\n

6. Th\u00e1ch Th\u1ee9c K\u1ef9 Thu\u1eadt v\u00e0 Kinh T\u1ebf<\/strong><\/span><\/h3>\n

6.1 Chi Ph\u00ed T\u00ednh To\u00e1n Kh\u1ed5ng L\u1ed3<\/strong><\/p>\n

\u0110\u00e0o t\u1ea1o c\u00e1c h\u1ec7 th\u1ed1ng c\u00f3 th\u1ec3 m\u00f4 ph\u1ecfng th\u1ebf gi\u1edbi th\u1ef1c \u0111\u00f2i h\u1ecfi s\u1ee9c m\u1ea1nh t\u00ednh to\u00e1n l\u1edbn h\u01a1n nhi\u1ec1u so v\u1edbi c\u00e1c m\u00f4 h\u00ecnh ng\u00f4n ng\u1eef hi\u1ec7n t\u1ea1i, v\u00ec ch\u00fang c\u1ea7n x\u1eed l\u00fd kh\u00f4ng ch\u1ec9 t\u1eeb ng\u1eef m\u00e0 c\u00f2n c\u1ea3 video \u0111\u1ed9 ph\u00e2n gi\u1ea3i cao.<\/p>\n

Odyssey cho bi\u1ebft c\u1ea7n m\u1ed9t chip H200 c\u1ee7a Nvidia<\/strong> \u2013 m\u1ed7i chip c\u00f3 gi\u00e1 l\u00ean \u0111\u1ebfn 40.000 USD<\/strong> \u2013 cho m\u1ed7i ng\u01b0\u1eddi d\u00f9ng truy c\u1eadp m\u00f4 h\u00ecnh Odyssey 2 th\u00f4ng qua API. \u0110\u00e2y l\u00e0 m\u1ee9c chi ph\u00ed c\u1ef1c k\u1ef3 cao so v\u1edbi vi\u1ec7c ph\u1ee5c v\u1ee5 ng\u01b0\u1eddi d\u00f9ng LLM th\u00f4ng th\u01b0\u1eddng.<\/p>\n

6.2 D\u1eef Li\u1ec7u: B\u00e0i To\u00e1n Nan Gi\u1ea3i<\/strong><\/p>\n

Kh\u00f4ng gi\u1ed1ng nh\u01b0 LLM \u0111\u01b0\u1ee3c hu\u1ea5n luy\u1ec7n tr\u00ean d\u1eef li\u1ec7u v\u0103n b\u1ea3n kh\u1ed5ng l\u1ed3 c\u00f3 th\u1ec3 thu th\u1eadp d\u1ec5 d\u00e0ng t\u1eeb internet, world models ph\u1ee5 thu\u1ed9c n\u1eb7ng n\u1ec1 v\u00e0o video<\/strong> \u2013 th\u1ee9 ph\u1ee9c t\u1ea1p h\u01a1n nhi\u1ec1u v\u00e0 kh\u00f3 thu th\u1eadp, g\u00e1n nh\u00e3n, v\u00e0 hu\u1ea5n luy\u1ec7n \u1edf quy m\u00f4 l\u1edbn.<\/p>\n

\u0110\u00e2y l\u00e0 l\u00fd do c\u00e1c c\u00f4ng ty nh\u01b0 Niantic Spatial c\u00f3 l\u1ee3i th\u1ebf \u0111\u1eb7c bi\u1ec7t: h\u00e0ng tri\u1ec7u ng\u01b0\u1eddi ch\u01a1i Pok\u00e9mon Go<\/em> \u0111\u00e3 kh\u00f4ng ch\u1ee7 \u00fd \u0111\u00f3ng g\u00f3p 30 t\u1ef7 h\u00ecnh \u1ea3nh v\u00e0 b\u1ea3n qu\u00e9t 3D c\u1ee7a th\u1ebf gi\u1edbi th\u1ef1c qua nhi\u1ec1u n\u0103m.<\/p>\n

6.3 V\u1ea5n \u0110\u1ec1 T\u1ed1c \u0110\u1ed9<\/strong><\/p>\n

Kh\u00f4ng ch\u1ec9 c\u1ea7n t\u1ea1o ra m\u00f4 ph\u1ecfng ch\u00ednh x\u00e1c \u2013 m\u00f4 ph\u1ecfng \u0111\u00f3 ph\u1ea3i x\u1ea3y ra \u0111\u1ee7 nhanh \u0111\u1ec3 c\u00f3 \u00edch trong th\u1ef1c t\u1ebf. \u0110\u00e2y l\u00e0 “\u0111i\u1ec3m m\u1edf kh\u00f3a then ch\u1ed1t” m\u00e0 Decart \u0111ang c\u1ed1 g\u1eafng gi\u1ea3i quy\u1ebft khi h\u1ecd t\u1ea1o ra video nhanh \u0111\u1ebfn m\u1ee9c kh\u1edbp v\u1edbi h\u00e0nh \u0111\u1ed9ng c\u1ee7a ng\u01b0\u1eddi d\u00f9ng t\u1eebng khung h\u00ecnh.<\/p>\n

7. Tranh Lu\u1eadn Trong Gi\u1edbi Nghi\u00ean C\u1ee9u<\/strong><\/span><\/h3>\n

Kh\u00f4ng ph\u1ea3i t\u1ea5t c\u1ea3 m\u1ecdi ng\u01b0\u1eddi \u0111\u1ec1u \u0111\u1ed3ng \u00fd r\u1eb1ng World Models l\u00e0 b\u01b0\u1edbc \u0111\u1ed9t ph\u00e1 ti\u1ebfp theo. C\u00f3 hai tr\u01b0\u1eddng ph\u00e1i ch\u00ednh \u0111ang tranh lu\u1eadn:<\/p>\n

Tr\u01b0\u1eddng ph\u00e1i “Scale is all you need”:<\/strong> Nhi\u1ec1u nh\u00e0 nghi\u00ean c\u1ee9u tin r\u1eb1ng c\u00e1c LLM, n\u1ebfu \u0111\u01b0\u1ee3c m\u1edf r\u1ed9ng \u0111\u1ee7 l\u1edbn v\u1edbi \u0111\u1ee7 d\u1eef li\u1ec7u \u0111a ph\u01b0\u01a1ng th\u1ee9c (bao g\u1ed3m c\u1ea3 video), cu\u1ed1i c\u00f9ng s\u1ebd t\u1ef1 ph\u00e1t tri\u1ec3n kh\u1ea3 n\u0103ng hi\u1ec3u th\u1ebf gi\u1edbi v\u1eadt l\u00fd.<\/p>\n

Tr\u01b0\u1eddng ph\u00e1i “World Models First” (LeCun, Li v\u00e0 nh\u1eefng ng\u01b0\u1eddi kh\u00e1c):<\/strong> H\u1ecd l\u1eadp lu\u1eadn r\u1eb1ng ki\u1ebfn tr\u00fac LLM v\u1ec1 c\u01a1 b\u1ea3n b\u1ecb gi\u1edbi h\u1ea1n b\u1edfi b\u1ea3n ch\u1ea5t d\u1ef1 \u0111o\u00e1n token c\u1ee7a n\u00f3 \u2013 v\u00e0 s\u1ebd kh\u00f4ng bao gi\u1edd \u0111\u1ea1t \u0111\u1ebfn tr\u00ed th\u00f4ng minh t\u1ed5ng qu\u00e1t n\u1ebfu kh\u00f4ng c\u00f3 m\u1ed9t c\u01a1 ch\u1ebf l\u1eadp m\u00f4 h\u00ecnh th\u1ebf gi\u1edbi ri\u00eang bi\u1ec7t.<\/p>\n

Nh\u01b0 blog Introl ghi nh\u1eadn, paradigm world models \u0111\u00e3 b\u00f9ng n\u1ed5 v\u00e0o d\u00f2ng ch\u00ednh c\u1ee7a ph\u00e1t tri\u1ec3n AI v\u00e0o cu\u1ed1i n\u0103m 2025 v\u00e0 \u0111\u1ea7u n\u0103m 2026, nh\u01b0ng cu\u1ed9c tranh lu\u1eadn c\u01a1 b\u1ea3n v\u1eabn ch\u01b0a \u0111\u01b0\u1ee3c gi\u1ea3i quy\u1ebft.<\/p>\n

8. T\u00edn Hi\u1ec7u T\u1eeb Th\u1ecb Tr\u01b0\u1eddng: “Kho\u1ea3nh Kh\u1eafc ChatGPT” \u0110ang \u0110\u1ebfn G\u1ea7n?<\/strong><\/span><\/h3>\n

B\u1ea5t ch\u1ea5p nh\u1eefng tranh lu\u1eadn h\u1ecdc thu\u1eadt, t\u00edn hi\u1ec7u t\u1eeb th\u1ecb tr\u01b0\u1eddng r\u1ea5t r\u00f5 r\u00e0ng. Ming-Yu Liu c\u1ee7a Nvidia nh\u1eadn \u0111\u1ecbnh r\u1eb1ng m\u1ed9t “kho\u1ea3nh kh\u1eafc ChatGPT”<\/strong> cho AI World Models<\/em> \u0111ang g\u1ea7n k\u1ec1 \u2013 m\u1ed9t b\u01b0\u1edbc nh\u1ea3y v\u1ecdt v\u1ec1 n\u0103ng l\u1ef1c c\u00f3 th\u1ec3 thay \u0111\u1ed5i to\u00e0n b\u1ed9 c\u00e1ch m\u1ecdi ng\u01b0\u1eddi ngh\u0129 v\u1ec1 robotics v\u00e0 t\u1ef1 \u0111\u1ed9ng h\u00f3a v\u1eadt l\u00fd.<\/p>\n

M\u1ee5c ti\u00eau c\u00e1 nh\u00e2n c\u1ee7a \u00f4ng l\u00e0 d\u1ea1y robot nh\u1eefng k\u1ef9 n\u0103ng m\u1edbi ch\u1ec9 t\u1eeb m\u1ed9t v\u00e0i v\u00ed d\u1ee5 \u2013 k\u1ec3 c\u1ea3 nh\u1eefng v\u00ed d\u1ee5 m\u00e0 robot ch\u01b0a t\u1eebng g\u1eb7p \u2013 r\u1ed3i th\u1ef1c hi\u1ec7n ch\u00fang m\u1ed9t c\u00e1ch nh\u1ea5t qu\u00e1n. “T\u00f4i th\u1ef1c s\u1ef1 tin r\u1eb1ng m\u1ecdi ng\u01b0\u1eddi \u0111ang d\u1ea7n t\u00ecm ra c\u00f4ng th\u1ee9c \u0111\u00fang \u0111\u1eafn,” \u00f4ng n\u00f3i.<\/p>\n

\"\"
Image by The Digital artist@pixabay.com<\/em><\/span><\/figcaption><\/figure>\n

9. K\u1ebft Lu\u1eadn<\/strong><\/span><\/h3>\n

AI World Models<\/strong> \u0111\u1ea1i di\u1ec7n cho m\u1ed9t s\u1ef1 thay \u0111\u1ed5i nh\u1eadn th\u1ee9c s\u00e2u s\u1eafc trong c\u00e1ch c\u1ed9ng \u0111\u1ed3ng AI t\u01b0 duy v\u1ec1 tr\u00ed th\u00f4ng minh. C\u00e2u h\u1ecfi kh\u00f4ng c\u00f2n l\u00e0 “l\u00e0m th\u1ebf n\u00e0o \u0111\u1ec3 AI n\u00f3i chuy\u1ec7n gi\u1ecfi h\u01a1n?” m\u00e0 l\u00e0 “l\u00e0m th\u1ebf n\u00e0o \u0111\u1ec3 AI hi\u1ec3u<\/strong> v\u00e0 t\u01b0\u01a1ng t\u00e1c<\/strong> v\u1edbi th\u1ebf gi\u1edbi th\u1ef1c?”<\/p>\n

C\u00e1c m\u00f4 h\u00ecnh ng\u00f4n ng\u1eef l\u1edbn \u0111\u00e3 mang l\u1ea1i nh\u1eefng b\u01b0\u1edbc ti\u1ebfn phi th\u01b0\u1eddng trong th\u1eadp k\u1ef7 qua \u2013 nh\u01b0ng ch\u00fang \u0111\u01b0\u1ee3c x\u00e2y d\u1ef1ng tr\u00ean n\u1ec1n t\u1ea3ng ng\u00f4n ng\u1eef, kh\u00f4ng ph\u1ea3i v\u1eadt l\u00fd. Th\u1ebf gi\u1edbi th\u1ef1c chuy\u1ec3n \u0111\u1ed9ng, thay \u0111\u1ed5i, v\u00e0 ph\u1ea3n \u1ee9ng theo nh\u1eefng c\u00e1ch m\u00e0 v\u0103n b\u1ea3n kh\u00f4ng th\u1ec3 n\u1eafm b\u1eaft \u0111\u1ea7y \u0111\u1ee7.<\/p>\n

World models l\u00e0 n\u1ed7 l\u1ef1c \u0111\u1ec3 l\u1ea5p \u0111\u1ea7y kho\u1ea3ng c\u00e1ch \u0111\u00f3 \u2013 v\u00e0 v\u1edbi h\u00e0ng t\u1ef7 \u0111\u00f4 la \u0111ang \u0111\u1ed5 v\u00e0o, c\u00f9ng v\u1edbi s\u1ef1 tham gia c\u1ee7a nh\u1eefng t\u00ean tu\u1ed5i h\u00e0ng \u0111\u1ea7u nh\u01b0 Yann LeCun, Fei-Fei Li, Google, Nvidia v\u00e0 Alibaba, \u0111\u00e2y kh\u00f4ng c\u00f2n l\u00e0 m\u1ed9t h\u01b0\u1edbng nghi\u00ean c\u1ee9u ngo\u00e0i l\u1ec1. World Models \u0111ang tr\u1edf th\u00e0nh m\u1eb7t tr\u1eadn ti\u1ebfp theo c\u1ee7a AI<\/strong>.<\/p>\n

\u00a0<\/strong><\/p>\n

Ph\u1ee5 L\u1ee5c: C\u00e1c C\u00f4ng Ty v\u00e0 Nh\u00e2n V\u1eadt Ch\u00ednh<\/strong><\/p>\n\n\n\n\n\n\n\n\n\n\n\n\n\n
T\u00ean<\/strong><\/td>\nVai tr\u00f2<\/strong><\/td>\nHo\u1ea1t \u0111\u1ed9ng ch\u00ednh<\/strong><\/td>\n<\/tr>\n<\/thead>\n
Fei-Fei Li<\/strong><\/td>\n“M\u1eb9 \u0111\u1ee1 \u0111\u1ea7u c\u1ee7a AI”, GS Stanford<\/td>\nS\u00e1ng l\u1eadp World Labs (2024), ~1 t\u1ef7 USD v\u1ed1n<\/td>\n<\/tr>\n
Yann LeCun<\/strong><\/td>\n“Cha \u0111\u1ee1 \u0111\u1ea7u c\u1ee7a AI”<\/td>\nS\u00e1ng l\u1eadp AMI Labs (2026), 3 t\u1ef7 euro \u0111\u1ecbnh gi\u00e1<\/td>\n<\/tr>\n
Google DeepMind<\/strong><\/td>\nG\u00e3 kh\u1ed5ng l\u1ed3 c\u00f4ng ngh\u1ec7<\/td>\nProject Genie \u2013 m\u00f4i tr\u01b0\u1eddng t\u01b0\u01a1ng t\u00e1c th\u1ef1c t\u1ebf<\/td>\n<\/tr>\n
Nvidia Cosmos Lab<\/strong><\/td>\nPh\u1ea7n c\u1ee9ng AI<\/td>\nC\u01a1 s\u1edf h\u1ea1 t\u1ea7ng v\u00e0 nghi\u00ean c\u1ee9u AI v\u1eadt l\u00fd<\/td>\n<\/tr>\n
Alibaba Cloud<\/strong><\/td>\nT\u1eadp \u0111o\u00e0n c\u00f4ng ngh\u1ec7 TQ<\/td>\n\u0110\u1ea7u t\u01b0 290M USD v\u00e0o Shengshu\/Vidu<\/td>\n<\/tr>\n
Niantic Spatial<\/strong><\/td>\nStartup (San Francisco)<\/td>\nM\u00f4 h\u00ecnh \u0111\u1ecba kh\u00f4ng gian l\u1edbn, 30 t\u1ef7 h\u00ecnh \u1ea3nh 3D<\/td>\n<\/tr>\n
Decart<\/strong><\/td>\nStartup (Tel Aviv)<\/td>\nWorld models th\u1eddi gian th\u1ef1c<\/td>\n<\/tr>\n
Odyssey<\/strong><\/td>\nStartup (Palo Alto)<\/td>\nD\u1ef1 \u0111o\u00e1n t\u01b0\u01a1ng lai v\u1eadt l\u00fd t\u1eeb video<\/td>\n<\/tr>\n
Sakana AI<\/strong><\/td>\nC\u00f4ng ty R&D (Nh\u1eadt B\u1ea3n)<\/td>\nNghi\u00ean c\u1ee9u n\u1ec1n t\u1ea3ng, \u0111\u1ed3ng s\u00e1ng l\u1eadp David Ha<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

","protected":false},"excerpt":{"rendered":"

Sau k\u1ef7 nguy\u00ean ChatGPT, AI h\u01b0\u1edbng \u0111\u1ebfn world models \u2014 h\u1ec7 th\u1ed1ng hi\u1ec3u th\u1ebf gi\u1edbi v\u1eadt l\u00fd. Tu\u1ec7 Quang ph\u00e2n t\u00edch l\u00e0n s\u00f3ng \u0111\u1ea7u t\u01b0 h\u00e0ng t\u1ef7 \u0111\u00f4 v\u00e0 cu\u1ed9c tranh lu\u1eadn l\u1edbn nh\u1ea5t trong gi\u1edbi nghi\u00ean c\u1ee9u AI hi\u1ec7n nay.<\/p>","protected":false},"author":14,"featured_media":1961,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"_gspb_post_css":"","footnotes":""},"categories":[101],"tags":[],"class_list":["post-1960","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-science-technology"],"blocksy_meta":[],"acf":[],"_links":{"self":[{"href":"https:\/\/1975mag.com\/ro\/wp-json\/wp\/v2\/posts\/1960","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/1975mag.com\/ro\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/1975mag.com\/ro\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/1975mag.com\/ro\/wp-json\/wp\/v2\/users\/14"}],"replies":[{"embeddable":true,"href":"https:\/\/1975mag.com\/ro\/wp-json\/wp\/v2\/comments?post=1960"}],"version-history":[{"count":3,"href":"https:\/\/1975mag.com\/ro\/wp-json\/wp\/v2\/posts\/1960\/revisions"}],"predecessor-version":[{"id":2065,"href":"https:\/\/1975mag.com\/ro\/wp-json\/wp\/v2\/posts\/1960\/revisions\/2065"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/1975mag.com\/ro\/wp-json\/wp\/v2\/media\/1961"}],"wp:attachment":[{"href":"https:\/\/1975mag.com\/ro\/wp-json\/wp\/v2\/media?parent=1960"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/1975mag.com\/ro\/wp-json\/wp\/v2\/categories?post=1960"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/1975mag.com\/ro\/wp-json\/wp\/v2\/tags?post=1960"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}