{"id":30765,"date":"2023-03-23T00:17:09","date_gmt":"2023-03-22T16:17:09","guid":{"rendered":"https:\/newsletter.sinica.edu.tw\/?p=30765"},"modified":"2023-04-20T00:43:34","modified_gmt":"2023-04-19T16:43:34","slug":"%e3%80%90%e5%b0%88%e6%ac%84%e3%80%91%e7%94%9f%e6%88%90%e5%bc%8f-ai-%e6%b5%aa%e6%bd%ae%e4%b8%8b%ef%bc%8c%e7%9c%bc%e8%a6%8b%e6%98%af%e5%90%a6%e7%82%ba%e6%86%91%ef%bc%9f","status":"publish","type":"post","link":"https:\/newsletter.sinica.edu.tw\/30765\/","title":{"rendered":"\u3010\u5c08\u6b04\u3011\u751f\u6210\u5f0f AI \u6d6a\u6f6e\u4e0b\uff0c\u773c\u898b\u662f\u5426\u70ba\u6191\uff1f"},"content":{"rendered":"

\u751f\u6210\u5f0f AI \u7684\u84ec\u52c3\u767c\u5c55\u8207\u5176\u5e36\u4f86\u7684\u6311\u6230<\/strong><\/p>\n

2022 \u5e74\u5e95\uff0c ChatGPT \u6a6b\u7a7a\u51fa\u4e16 1<\/sup>\u3002<\/p>\n

\u300cChatGPT\u300d\u662f\u4e00\u500b\u7531 OpenAI \u958b\u767c\u7684\u81ea\u7136\u8a9e\u8a00\u8655\u7406\u6a21\u578b\uff0c\u5b83\u80fd\u5920\u900f\u904e\u6587\u5b57\u4ecb\u9762\u8207\u4f7f\u7528\u8005\u9032\u884c\u81ea\u7136\u800c\u6d41\u66a2\u5730\u5c0d\u8a71\uff0c\u5354\u52a9\u4f7f\u7528\u8005\u9ad8\u54c1\u8cea\u5730\u5b8c\u6210\u5404\u9805\u5de5\u4f5c\uff0c\u8209\u51e1\u5beb\u4f5c\u3001\u5beb\u6b4c\u3001\u5beb\u7a0b\u5f0f\u3001\u6539\u6587\u6cd5\u932f\u8aa4\u7b49\u81ea\u7136\u8a9e\u8a00\u4efb\u52d9\u90fd\u96e3\u4e0d\u5012\u5b83\uff0c\u5728\u8f38\u5165\u95dc\u9375\u5b57\u7684\u5e7e\u79d2\u9418\u5f8c\uff0c\u5c31\u53ef\u7522\u751f\u4f7f\u7528\u8005\u8981\u6c42\u7684\u5167\u5bb9\u3002\u5b83\u7684\u51fa\u73fe\u53d7\u5230\u5ee3\u6cdb\u7684\u6b61\u8fce\u8207\u8ffd\u6367\uff0c\u4e0a\u7dda 2 \u500b\u6708\u5f8c\u6d3b\u8e8d\u7528\u6236\u5373\u7a81\u7834\u4e00\u5104\uff0c\u666e\u7f85\u5927\u773e\u90fd\u5728\u70ba\u5b83\u7684\u8868\u73fe\u9a5a\u5606\uff0c\u4e26\u7d1b\u7d1b\u7814\u7a76\u66f4\u597d\u7684\u63d0\u793a\u8a5e\u4ee5\u63d0\u9ad8\u4f7f\u7528\u6548\u7387\u3002<\/p>\n

\u5230\u4e86 2023 \u5e74\uff0c OpenAI \u63a8\u51fa\u4e86\u65b0\u7684 GPT-4 \u6a21\u578b 2<\/sup>\uff0c\u5177\u5099\u66f4\u7cbe\u78ba\u7684\u751f\u6210\u7d50\u679c\u548c\u8655\u7406\u591a\u6a21\u614b\u8cc7\u6599\u7684\u80fd\u529b\uff0c\u80fd\u5920\u5c07\u624b\u7e6a\u7684\u7db2\u9801\u8349\u7a3f\u8f49\u63db\u6210\u76f8\u5c0d\u61c9\u7684\u7a0b\u5f0f\u78bc\uff0c\u6548\u679c\u5341\u5206\u9a5a\u4eba\u3002\u53e6\u5916\u4e00\u65b9\u9762\uff0c\u5fae\u8edf\u8eab\u70ba OpenAI \u7684\u6700\u5927\u5408\u4f5c\u5925\u4f34\uff0c\u66f4\u8d81\u52e2\u5c07 ChatGPT \u6574\u5408\u81f3\u81ea\u5bb6\u7684 Bing \u641c\u5c0b\u5f15\u64ce\u8207 Office \u7b49\u5404\u9805\u7522\u54c1\uff0c\u63d0\u4f9b\u5167\u5bb9\u81ea\u52d5\u751f\u6210\u8207\u6458\u8981\u7684\u529f\u80fd\uff0c\u8c37\u6b4c\u8207\u5176\u4ed6\u79d1\u6280\u516c\u53f8\u4e5f\u63a8\u51fa\u985e ChatGPT \u670d\u52d9\u7dca\u8ffd\u5728\u5f8c\u3002<\/p>\n

\u800c\u9664\u4e86 ChatGPT \u5916\uff0c\u66f4\u65e9\u4e4b\u524d\uff0c AI \u7e6a\u5716\u4e00\u6a23\u5438\u5f15\u4e0d\u5c11\u5927\u773e\u7684\u76ee\u5149\uff0c OpenAI \u64f4\u6563\u751f\u6210\u6a21\u578b\uff08Diffusion Model\uff09\u7684\u6587\u5b57\u751f\u6210\u5716\u7247\u7cfb\u7d71\u300cDALL\uff0eE-2\u300d3<\/sup>\uff0c\u80fd\u5920\u8b93\u4f7f\u7528\u8005\u7d66\u5b9a\u4efb\u610f\u6587\u5b57\u6558\u8ff0\uff0c\u5feb\u901f\u751f\u6210\u9ad8\u54c1\u8cea\u7684\u5716\u7247\u8207\u756b\u4f5c\u3002\u540c\u6a23\u5730\uff0c\u8c37\u6b4c\u3001Meta\u3001\u8f1d\u9054\u7b49\u79d1\u6280\u516c\u53f8\u4e5f\u7d1b\u7d1b\u63a8\u51fa\u81ea\u5df1\u7684\u6587\u5b57\u751f\u6210\u5f71\u50cf\u8207\u5f71\u7247\u7b49\u751f\u6210\u6a21\u578b\uff08\u5305\u542b\uff1aImagen 4<\/sup>\u3001Imagen Video 5<\/sup>\u3001Phenaki 6<\/sup>\u3001Make-a-Scene 7<\/sup>\u3001Make-a-Video 8<\/sup>\u3001eDiff-I 9<\/sup>\uff09\u3002\u53ef\u4ee5\u8aaa\u96a8\u8457ChatGPT\u8207AI\u7e6a\u5716\u7684\u6210\u529f\uff0c\u4e00\u6642\u9593\uff0c\u751f\u6210\u5f0f AI \uff08Generative Artificial Intelligence\uff09\u8207\u76f8\u95dc\u6280\u8853\u6210\u70ba\u4eba\u5de5\u667a\u6167\u76ee\u524d\u6700\u71b1\u9580\u7684\u7814\u7a76\u65b9\u5411\uff0c\u5404\u5f0f\u5167\u5bb9\u751f\u6210\u7684\u5de5\u5177\uff0c\u5982\u96e8\u5f8c\u6625\u7b4d\u4e0d\u65b7\u63a8\u51fa\uff0c\u84ec\u52c3\u767c\u5c55\u3002<\/p>\n

\u9019\u4e9b\u751f\u6210\u5f0f\u4eba\u5de5\u667a\u6167\u8b93\u5167\u5bb9\u5275\u4f5c\u8b8a\u5f97\u66f4\u7c21\u55ae\uff0c\u8b93\u4eba\u5011\u53ef\u4ee5\u66f4\u5feb\u901f\u5927\u91cf\u5730\u5275\u9020\u51fa\u9ad8\u54c1\u8cea\u7684\u5f71\u50cf\u8207\u6587\u5b57\u5167\u5bb9\uff0c\u4e00\u4f4d\u540d\u70ba Sincarnate \u7684\u7db2\u53cb\u5c31\u5229\u7528\u985e\u4f3c OpenAI \u7684 DALL\uff0eE-2 \u7684 AI \u7e6a\u5716\u670d\u52d9 Midjourney 10<\/sup> \u751f\u6210\u540d\u70ba\u300c\u592a\u7a7a\u6b4c\u5287\u9662\u300d\u7684\u4f5c\u54c1\uff0c\u4e26\u5728\u85dd\u8853\u6bd4\u8cfd\u62ff\u4e0b\u9996\u734e\uff0c\u53ef\u898b\u9019\u4e9b\u6280\u8853\u7684\u512a\u8d8a\u6027\u3002<\/p>\n

\u4f46\u9019\u4e9b\u6280\u8853\u7684\u80cc\u5f8c\u4e5f\u727d\u6d89\u8a31\u591a\u95dc\u65bc\u96b1\u79c1\u8207\u8457\u4f5c\u6b0a\u7684\u554f\u984c\uff0c\u56e0\u70ba\u9019\u4e9b\u751f\u6210\u6a21\u578b\u7684\u8a13\u7df4\u8cc7\u6599\uff0c\u5927\u591a\u6578\u90fd\u662f\u7d93\u7531\u7db2\u8def\u722c\u87f2\u4e0b\u8f09\uff0c\u5927\u90e8\u5206\u8cc7\u6599\u4e26\u6c92\u6709\u7d93\u904e\u539f\u4f5c\u8005\u7684\u6388\u6b0a\uff0c\u5982 ChatGPT \u5c31\u4f7f\u7528\u8a31\u591a\u672a\u7d93\u6388\u6b0a\u4e3b\u6d41\u5a92\u9ad4\u7684\u6587\u7ae0\u5167\u5bb9\u9032\u884c\u8a13\u7df4\uff0c\u81f3\u65bc AI \u7e6a\u5716\uff0c\u4e5f\u662f\u4f7f\u7528\u5927\u91cf\u7db2\u8def\u8207\u7e6a\u5e2b\u7b49\u672a\u6388\u6b0a\u5716\u7247\u9032\u884c\u8a13\u7df4\u3002\u53e6\u5916\uff0c\u6709\u7db2\u53cb\u6536\u96c6\u6578\u5341\u5f35\u67d0\u77e5\u540d\u7e6a\u5e2b\u7684\u4f5c\u54c1\u5716\uff0c\u5fae\u8abf\u958b\u6e90\u6587\u5b57\u751f\u6210\u5716\u50cf\u6a21\u578b Stable Diffusion 11\u300112<\/sup>\uff0c\u5c31\u53ef\u4f7f\u7528\u5fae\u8abf\u904e\u7684\u65b0\u6a21\u578b\uff0c\u914d\u5408\u95dc\u9375\u5b57\u8f15\u9b06\u5feb\u901f\u5730\u9032\u884c\u985e\u4f3c\u98a8\u683c\u7684 AI \u7e6a\u5716\u5275\u4f5c 13<\/sup>\uff0c\u88ab\u7e6a\u5e2b\u5011\u8a8d\u70ba\u662f\u7aca\u53d6\u7e6a\u756b\u98a8\u683c\u3002\u9019\u4e9b\u672a\u7d93\u6388\u6b0a\u64c5\u81ea\u76dc\u7528\u8cc7\u6599\u7684\u884c\u70ba\uff0c\u5df2\u7d93\u9020\u6210\u8a31\u591a\u4e3b\u6d41\u5a92\u9ad4\u8207\u4e0a\u5343\u4f4d\u85dd\u8853\u5bb6\u5c0d\u63d0\u4f9b\u76f8\u95dc\u670d\u52d9\u7684\u516c\u53f8\u9032\u884c\u6297\u8b70\u8207\u63d0\u544a\u3002<\/p>\n

\u9664\u4e86\u96b1\u79c1\u8207\u667a\u8ca1\u6b0a\u7684\u554f\u984c\uff0c\u53ef\u4ee5\u9810\u671f\u5c07\u4f86\u6703\u6709\u66f4\u591a\u7684\u624b\u6a5f\u61c9\u7528\u7a0b\u5f0f\u8207\u958b\u6e90\u8edf\u9ad4\u5167\u5efa\u9019\u4e9b\u529f\u80fd\uff0c\u4e00\u822c\u5927\u773e\u6703\u66f4\u5bb9\u6613\u63a5\u89f8\u8207\u4f7f\u7528\u9019\u4e9b\u6280\u8853\u3002\u4e0d\u50cf\u50b3\u7d71\u8907\u96dc\u7684\u7de8\u8f2f\u8edf\u9ad4\uff0c\u4f7f\u7528\u8005\u53ea\u8981\u900f\u904e\u5e7e\u500b\u7c21\u55ae\u7684\u64cd\u4f5c\u8207\u95dc\u9375\u5b57\u5c31\u53ef\u4ee5\u5b8c\u6210\u8907\u96dc\u4e14\u9ad8\u54c1\u8cea\u7684\u7de8\u4fee\uff0c\u96a8\u8457\u5167\u5bb9\u751f\u6210\u7684\u9580\u6abb\u5927\u5e45\u964d\u4f4e\uff0c\u9019\u4e9b\u5f37\u5927\u6df1\u5ea6\u751f\u6210\u6a21\u578b\uff0c\u4e5f\u53ef\u80fd\u88ab\u6709\u5fc3\u4eba\u58eb\u7528\u65bc\u751f\u6210\u5927\u91cf\u96e3\u8fa8\u771f\u507d\u7684\u507d\u9020\u5167\u5bb9\u3001\u5047\u65b0\u805e\u3001\u548c\u5167\u5bb9\u8fb2\u5834\u6587\u7ae0\uff0c\u5982\u4e4b\u524d\u53d7\u5230\u5404\u754c\u95dc\u6ce8\u7684\u6df1\u507d\u6280\u8853\uff08Deepfake\uff09\uff0c\u53ef\u4ee5\u628a\u5f71\u7247\u7684\u4eba\u81c9\u7f6e\u63db\u6210\u4efb\u4f55\u4ed6\u5011\u60f3\u8981\u7684\u4eba\u81c9\u8207\u8868\u60c5\uff0c\u6216\u76f4\u63a5\u5c0d\u4eba\u81c9\u7684\u7279\u5fb5\uff0c\u5982\uff1a\u5e74\u9f61\u3001\u6027\u5225\u7b49\u7279\u5fb5\u76f4\u63a5\u9032\u884c\u64cd\u4f5c\u3002<\/p>\n

Facebook \u8ddf Instagram \u7b49\u793e\u7fa4\u8edf\u9ad4\u5c31\u66fe\u591a\u6b21\u767c\u73fe\u4e0d\u5c11\u4f7f\u7528\u6df1\u507d\u6280\u8853\u81ea\u52d5\u751f\u6210\u7167\u7247\u7533\u8acb\u7684\u4eba\u982d\u5e33\u865f\u3001\u7c89\u7d72\u5c08\u9801\u8207\u793e\u5718\uff0c\u9032\u884c\u8a50\u9a19\u6216\u6563\u64ad\u654f\u611f\u8a00\u8ad6\u3002\u4ee5\u4e0b\u70ba\u771f\u5be6\u6848\u4f8b\uff1a\u82f1\u570b\u4e00\u9593\u80fd\u6e90\u516c\u53f8\u7684 CEO\uff0c\u63a5\u5230\u5fb7\u570b\u6bcd\u516c\u53f8\u4e0a\u53f8\u4e00\u901a\u96fb\u8a71\uff0c\u8981\u6c42\u4ed6\u7dca\u6025\u5c07 24 \u842c 3 \u5343\u7f8e\u91d1\u532f\u81f3\u6307\u5b9a\u4f9b\u61c9\u5546\u5e33\u6236\uff0c\u8a72 CEO \u81ea\u8a8d\u5c0d\u65bc\u4e0a\u53f8\u8072\u97f3\u76f8\u7576\u719f\u6089\uff0c\u6240\u4ee5\u63a5\u5230\u96fb\u8a71\u4e4b\u5f8c\u4e0d\u7591\u6709\u5b83\uff0c\u5c31\u6d3e\u4eba\u53bb\u8f49\u5e33\uff0c\u4f46\u5176\u5be6\u9019\u901a\u96fb\u8a71\u662f\u900f\u904e\u8072\u97f3\u8f49\u63db\uff08Voice Conversion\uff09\u6280\u8853\u507d\u9020\u7684\u8a50\u9a19\u96fb\u8a71 14<\/sup>\u3002<\/p>\n

\u6df1\u507d\u6280\u8853\u4e0d\u65b7\u63a8\u9673\u51fa\u65b0\uff0c\u5982\u679c\u518d\u914d\u5408\u4f7f\u7528\u9019\u4e9b\u6700\u65b0\u751f\u6210\u5f0f AI \u6280\u8853\uff0c\u5247\u53ef\u7522\u751f\u66f4\u771f\u5be6\u7684\u507d\u9020\u5167\u5bb9\u8207\u5f71\u7247\uff0c\u56e0\u6b64\uff0c\u5728\u9019\u6ce2\u751f\u6210\u5f0f AI \u7684\u6d6a\u6f6e\uff0c\u8209\u51e1\u4f60\u8eab\u908a\u7684\u8cc7\u8a0a\uff0c\u7686\u4e0d\u518d\u662f\u773c\u898b\u70ba\u6191\uff0c\u5c24\u5176\u662f\u7576\u9019\u4e9b\u751f\u6210\u5f0f AI\uff0c\u53ef\u80fd\u6703\u4e00\u672c\u6b63\u7d93\u5730\u7522\u751f\u865b\u69cb\u7684\u932f\u8aa4\u5167\u5bb9\uff0c\u66f4\u9700\u8981\u4f60\u6211\u6ce8\u610f\u3002\u56e0\u6b64\uff0c\u9700\u8981\u793e\u6703\u5404\u754c\u5c0d\u76f8\u95dc\u8b70\u984c\u7684\u91cd\u8996\uff0c\u57f9\u990a\u826f\u597d\u7684\u5a92\u9ad4\u8b58\u8b80\u80fd\u529b\uff0c\u8207\u958b\u767c\u76f8\u95dc\u7684\u5075\u6e2c\u53ca\u4f86\u6e90\u8ffd\u8e64\u5de5\u5177\uff0c\u624d\u80fd\u907f\u514d\u53ef\u80fd\u767c\u751f\u7684\u793e\u6703\u554f\u984c\u3002<\/p>\n

\u507d\u9020\u5167\u5bb9\u5075\u6e2c\u8207\u667a\u8ca1\u6b0a\u4fdd\u8b77\u7b49\u5176\u4ed6\u53ef\u80fd\u7684\u53cd\u5236\u624b\u6bb5<\/strong><\/p>\n

\u96a8\u8457\u751f\u6210\u5f0f AI \u7684\u6280\u8853\u4e0d\u65b7\u9032\u5316\uff0c\u53ef\u80fd\u5c0e\u81f4\u507d\u9020\u5167\u5bb9\u8207\u667a\u8ca1\u6b0a\u76dc\u7528\u7684\u6c3e\u6feb\uff0c\u56b4\u91cd\u7684\u8a71\uff0c\u4e5f\u53ef\u80fd\u6703\u6f14\u8b8a\u6210\u4e0d\u53ef\u5ffd\u8996\u7684\u8cc7\u8a0a\u5b89\u5168\u8207\u793e\u6703\u4fe1\u4efb\u5a01\u8105\u3002\u7f8e\u570b\u570b\u9632\u9ad8\u7b49\u7814\u7a76\u8a08\u5283\u7f72\uff08DARPA\uff09\u65e9\u5728 2016 \u5e74\u5373\u958b\u59cb Media Forensics \u7814\u7a76\u8a08\u756b\uff0c\u958b\u767c\u95dc\u9375\u6280\u8853\u4ee5\u904f\u6b62\u76f8\u95dc\u591a\u5a92\u9ad4\u507d\u9020\u5167\u5bb9\u7684\u6c3e\u6feb\u3002\u5728\u8a08\u756b\u7d50\u675f\u5f8c\uff0cDARPA \u66f4\u5728 2020 \u5e74\u7e7c\u7e8c\u57f7\u884c Semantic Forensics \u7814\u7a76\u8a08\u756b\uff0c\u76ee\u6a19\u70ba\u8b93\u6a5f\u5668\u80fd\u7406\u89e3\u5716\u50cf\u53ca\u5f71\u7247\u4e2d\u6240\u860a\u542b\u7684\u8a9e\u610f\uff0c\u505a\u51fa\u66f4\u7cbe\u6e96\u7684\u507d\u9020\u5167\u5bb9\u8fa8\u8b58\u548c\u4e8b\u4ef6\u5075\u6e2c\u3002\u540c\u6642\u9593\uff0c\u7531\u65bc\u528d\u6a4b\u5206\u6790\u4e8b\u4ef6\u5f71\u97ff\u7f8e\u570b\u7e3d\u7d71\u5927\u9078\u7684\u4e8b\u4ef6\uff0cMeta\u3001\u8c37\u6b4c\u3001\u5fae\u8edf\u7b49\u5927\u516c\u53f8\u4e5f\u6301\u7e8c\u6295\u5165\u5927\u91cf\u8cc7\u6e90\u65bc\u76f8\u95dc\u9451\u8b58\u7814\u7a76\uff0c\u53ef\u898b\u5176\u91cd\u8981\u6027\u3002<\/p>\n

\u8981\u5075\u6e2c\u67d0\u500b\u5f71\u50cf\u6216\u6587\u7ae0\u662f\u5426\u70ba AI \u751f\u6210\uff0c\u6700\u7c21\u55ae\u7684\u65b9\u5f0f\uff0c\u5c31\u662f\u6536\u96c6\u6b63\u8ca0\u6a23\u672c\u8a13\u7df4\u8cc7\u6599\uff0c\u5373\u771f\u5be6\u8207 AI \u751f\u6210\u7684\u8cc7\u6599\uff0c\u900f\u904e\u76e3\u7763\u5b78\u7fd2\u7684\u65b9\u5f0f\uff0c\u4f7f\u7528\u6df1\u5ea6\u5b78\u7fd2\u6216\u5176\u5b83\u6a5f\u5668\u5b78\u7fd2\u7684\u6a21\u578b\u8a13\u7df4\u4e00\u500b\u4e8c\u5143\u5206\u985e\u5668\u4f86\u9032\u884c\u5224\u8b80\u3002\u5728 ChatGPT \u516c\u958b\u7684\u4e00\u6bb5\u6642\u9593\u5f8c\uff0c OpenAI \u91cb\u51fa\u76f8\u95dc\u7684 AI \u751f\u6210\u6587\u7ae0\u6aa2\u6e2c\u5668 15<\/sup>\uff0c\u4f7f\u7528\u8005\u53ef\u81f3\u7db2\u7ad9\u8f38\u5165\u76f8\u95dc\u6587\u7ae0\u9032\u884c\u5075\u6e2c\uff0c\u4f46\u9084\u662f\u6703\u6709\u7121\u6cd5\u5224\u5225\u7684\u60c5\u6cc1\u3002<\/p>\n

\u8a13\u7df4\u4e8c\u5143\u5206\u985e\u5668\u9700\u8981\u6536\u96c6\u5927\u91cf\u8cc7\u6599\u4ee5\u76e1\u53ef\u80fd\u5305\u542b\u5404\u7a2e\u60c5\u6cc1\uff0c\u4e0d\u50c5\u6536\u96c6\u8207\u6a19\u8a3b\u8cbb\u6642\u8cbb\u529b\uff0c\u5be6\u52d9\u4e0a\u57f7\u884c\u4e5f\u6709\u76f8\u7576\u7684\u56f0\u96e3\u5ea6\u3002\u7f8e\u570b\u53f2\u4e39\u4f5b\u5927\u5b78\u7814\u7a76\u5718\u968a\u63d0\u51fa\u4e00\u500b\u96f6\u6a23\u672c AI \u6587\u7ae0\u5075\u6e2c\u5668 16<\/sup>\uff0c\u91dd\u5c0d AI \u751f\u6210\u7684\u6587\u7ae0\u8207\u4eba\u985e\u5beb\u7684\u6587\u7ae0\u5176\u5728\u8a9e\u8a00\u6a21\u578b\u4e2d\u9810\u6e2c\u5206\u6578\u7684\u5206\u5e03\u5728\u901a\u5e38\u4e0d\u540c\u5340\u57df\u7684\u7279\u6027\u9032\u884c\u8a2d\u8a08\uff0c\u5728\u4e0d\u9700\u8981\u6536\u96c6\u5927\u91cf\u8cc7\u6599\u7684\u60c5\u6cc1\u4e0b\uff0c\u4e5f\u80fd\u9054\u5230\u975e\u5e38\u4e0d\u932f\u7684\u6548\u679c\u3002<\/p>\n

\u53e6\u5916\uff0c\u4e5f\u6709\u7814\u7a76\u8005\u8a66\u8457\u5728\u8a9e\u8a00\u6a21\u578b\u4e2d\u52a0\u5165\u6d6e\u6c34\u5370 17<\/sup>\uff0c\u8b93\u751f\u6210\u5167\u5bb9\u5305\u542b\u4e00\u4e9b\u7279\u5b9a\u7684\u6a21\u5f0f\uff08Pattern\uff09\uff0c\u85c9\u6b64\u8b93\u5176\u66f4\u5bb9\u6613\u88ab\u6aa2\u6e2c\u5668\u5224\u65b7\u662f\u5426\u70ba AI \u751f\u6210\u7684\u5167\u5bb9\u3002\u800c\u95dc\u65bc AI \u7e6a\u5716\uff0c\u6709\u7814\u7a76\u5b78\u8005\u767c\u73fe 18\u300119<\/sup>\uff0c\u53ef\u4ee5\u5f9e\u8a13\u7df4\u597d\u7684\u64f4\u6563\u6a21\u578b\u4e2d\uff0c\u62bd\u53d6\u539f\u5148\u8a13\u7df4\u7684\u6a23\u672c\uff0c\u9019\u4e9b\u7814\u7a76\u66f4\u5f15\u8d77 AI \u7e6a\u5716\u662f\u5426\u6709\u6284\u8972\u7684\u7591\u616e\uff0c\u800c\u8981\u8fa8\u5225\u5716\u7247\u662f\u5426\u662f AI \u751f\u6210\uff0c\u540c\u6a23\u53ef\u4ee5\u8a13\u7df4\u4e8c\u5143\u5206\u985e\u5668\u9032\u884c\u5224\u8b80\uff0c\u4e5f\u80fd\u5920\u6dfb\u52a0\u6d6e\u6c34\u5370\u9032\u884c\u8457\u4f5c\u6b0a\u4fdd\u8b77 20<\/sup>\u3002<\/p>\n

\u4f46\u70ba\u4e86\u907f\u514d\u6709\u5fc3\u4eba\u58eb\u5229\u7528\u958b\u6e90\u7684 AI \u7e6a\u5716\u6a21\u578b\uff0c\u76dc\u7528\u7279\u5b9a\u85dd\u8853\u5bb6\u7684\u756b\u4f5c\u6a21\u4eff\u5176\u85dd\u8853\u98a8\u683c\uff0c\u5927\u91cf\u7522\u751f\u985e\u4f3c\u98a8\u683c\u7684\u7e6a\u5716\uff0c\u900f\u904e\u6cd5\u898f\u7684\u6a21\u7cca\u5730\u5e36\u640d\u5bb3\u85dd\u8853\u5bb6\u7684\u667a\u8ca1\u6b0a\uff0c\u7f8e\u570b\u829d\u52a0\u54e5\u5927\u5b78 Ben Zhao \u6559\u6388\u9818\u5c0e\u7684\u5718\u968a\uff0c\u958b\u767c\u540d\u70ba GLAZE \u7684 AI \u7e6a\u5716\u9632\u79a6\u5de5\u5177 21<\/sup>\u3002\u5176\u60f3\u6cd5\u4e43\u57fa\u65bc\u5c0d\u6297\u6a23\u672c\u7684\u6982\u5ff5\uff0c\u5373\u6dfb\u52a0\u4eba\u773c\u4e0d\u53ef\u5206\u8fa8\u7684\u5fae\u5c0f\u7684\u8e81\u8072\uff0c\u8b93\u6df1\u5ea6\u5b78\u7fd2\u6a21\u578b\u7684\u9810\u6e2c\u7522\u751f\u932f\u8aa4\u7684\u7d50\u679c\uff0c\u56e0\u6b64\uff0c\u900f\u904e GLAZE \u4fdd\u8b77\u7684\u5716\u7247\uff0c\u7576\u60e1\u610f\u4f7f\u7528\u8005\u8981\u4f7f\u7528\u5b83\u5011\u5fae\u8abf AI \u7e6a\u5716\u7684\u6a21\u578b\uff0c\u9032\u884c\u985e\u4f3c\u98a8\u683c\u7684\u5275\u4f5c\uff0c\u5c07\u7121\u6cd5\u7522\u751f\u4ed6\u5011\u60f3\u8981\u7684\u7d50\u679c\u3002<\/p>\n

\u4f46\u662f\uff0c\u6dfb\u52a0\u9019\u4e9b\u5c0d\u6297\u64fe\u52d5\u6703\u964d\u4f4e\u539f\u672c\u5716\u7247\u7684\u54c1\u8cea\uff0c\u800c\u4e14\u9700\u8981\u85dd\u8853\u5bb6\u91dd\u5c0d\u6bcf\u4e00\u5f35\u5716\u7247\uff0c\u9032\u884c\u540c\u6a23\u6b65\u9a5f\u3002\u6709\u9451\u65bc\u6b64\uff0c\u7f8e\u570b\u6771\u5317\u5927\u5b78 Bau \u6559\u6388\u5718\u968a\u958b\u767c\u300c\u751f\u6210\u6a21\u578b\u6982\u5ff5\u53bb\u9664\u6280\u8853\u300d22<\/sup>\uff0c\u53ef\u4ee5\u6709\u6548\u6291\u5236\u64f4\u6563\u751f\u6210\u6a21\u578b\u751f\u6210\u67d0\u500b\u6982\u5ff5\u8207\u98a8\u683c\u7684 AI \u7e6a\u5716\u3002\u5176\u539f\u7406\u70ba\u5229\u7528\u4e00\u500b\u5df2\u7d93\u8a13\u7df4\u597d\u958b\u6e90\u7684 AI \u7e6a\u5716\u6a21\u578b\uff08\u8001\u5e2b\uff09\uff0c\u900f\u904e\u514d\u5206\u985e\u5668\u5f15\u5c0e\u8207\u77e5\u8b58\u84b8\u993e\u7684\u65b9\u5f0f\u8a13\u7df4\u53e6\u4e00\u500b AI \u7e6a\u5716\u6a21\u578b\uff08\u5b78\u751f\uff09\uff0c\u6291\u5236\u6307\u5b9a\u6982\u5ff5\u8207\u98a8\u683c\u7684\u5716\u7247\u751f\u6210\u3002\u76ee\u524d AI \u7e6a\u5716\u5c31\u5982\u540c\u6df1\u507d\u5167\u5bb9\u4e00\u6a23\uff0c\u4ecd\u53ef\u80fd\u6709\u4e00\u4e9b\u6975\u9650\uff0c\u56e0\u6b64\u751f\u6210\u7684\u5167\u5bb9\u53ef\u80fd\u5305\u542b\u4e00\u4e9b\u8996\u89ba\u7455\u75b5\u548c\u4e0d\u5408\u7406\u7684\u5730\u65b9\uff0c\u6700\u5e38\u898b\u7684\u7455\u75b5\u662f\u624b\u90e8\u7684\u90e8\u5206\uff0c\u5f62\u72c0\u5947\u602a\u6216\u662f\u53ea\u6709 4 \u500b\u6307\u982d\u6216\u591a\u65bc 5 \u500b\u6307\u982d\uff0c\u5716\u7247\u5c31\u6709\u5f88\u9ad8\u6a5f\u6703\u662f AI \u751f\u6210\u7684\u3002<\/p>\n

\u751f\u6210\u5f0f AI \u7684\u6280\u8853\u5c07\u4e0d\u65b7\u5730\u9032\u5316\u8207\u5feb\u901f\u5730\u63a8\u9673\u51fa\u65b0\uff0c\u4e0a\u8ff0\u7684\u4fdd\u8b77\u65b9\u6cd5\uff0c\u4e5f\u8a31\u5f88\u5feb\u5c31\u6703\u88ab\u6709\u5fc3\u4eba\u58eb\u7834\u89e3\uff0c\u56e0\u6b64\uff0c\u7576\u4eba\u5011\u6c89\u6d78\u65bc\u751f\u6210\u5f0f AI \u5e36\u4f86\u7684\u4fbf\u5229\uff0c\u4e5f\u61c9\u8a72\u8003\u616e\u5b83\u53ef\u80fd\u5e36\u4f86\u7684\u5a01\u8105\uff0c\u4e26\u5c0d\u65bc\u76f8\u95dc\u7684\u4fdd\u8b77\u8207\u9632\u885b\u6280\u8853\u540c\u6a23\u6295\u5165\u5927\u91cf\u7684\u7814\u7a76\uff0c\u5305\u62ec\uff1a\u751f\u6210\u5167\u5bb9\u4f9d\u64da\u7684\u4f86\u6e90\u8ffd\u8e64\u3001AI \u751f\u6210\u5167\u5bb9\u5075\u6e2c\u3001\u66f4\u7a69\u5065\u7684\u6578\u4f4d\u6d6e\u6c34\u5370\u6216\u653b\u64ca\u529b\u66f4\u5f37\u5c0d\u6297\u6a23\u672c\u7b49\u7814\u7a76\uff0c\u8b93\u5927\u773e\u90fd\u80fd\u653e\u5fc3\u5730\u4f7f\u7528\u751f\u6210\u5f0f AI \u7684\u670d\u52d9\u3002<\/p>\n

\n
    \n
  1. https:\/openai.com\/blog\/chatgpt<\/a><\/li>\n
  2. OpenAI, GPT-4 Technical Report, arXiv preprint arXiv:2303.08774v2<\/a><\/em>, 2023.<\/li>\n
  3. https:\/openai.com\/dall-e-2\/<\/a><\/li>\n
  4. Saharia, Chitwan, William Chan, Saurabh Saxena, Lala Li, Jay Whang, Emily Denton, Seyed Kamyar Seyed Ghasemipour et al. “Photorealistic text-to-image diffusion models with deep language understanding.” arXiv preprint arXiv:2205.11487<\/em>, 2022.<\/li>\n
  5. Jonathan Ho, William Chan, Chitwan Saharia, Jay Whang, Ruiqi Gao, Alexey Gritsenko, Diederik P. Kingma et al. “Imagen video: High definition video generation with diffusion models.” arXiv preprint arXiv:2210.02303, 2022<\/em>.<\/li>\n
  6. Ruben Villegas, Mohammad Babaeizadeh, Pieter-Jan Kindermans, Hernan Moraldo, Han Zhang, Mohammad Taghi Saffar, Santiago Castro, Julius Kunze, and Dumitru Erhan. “Phenaki: Variable length video generation from open domain textual description.” arXiv preprint arXiv:2210.02399, 2022.<\/li>\n
  7. Oran Gafni, Adam Polyak, Oron Ashual, Shelly Sheynin, Devi Parikh, and Yaniv Taigman. “Make-a-scene: Scene-based text-to-image generation with human priors.” arXiv preprint arXiv:2203.13131, 2022.<\/li>\n
  8. Uriel Singer, Adam Polyak, Thomas Hayes, Xi Yin, Jie An, Songyang Zhang, Qiyuan Hu et al. “Make-a-video: Text-to-video generation without text-video data.” arXiv preprint arXiv:2209.14792, 2022.<\/li>\n
  9. Yogesh Balaji, Seungjun Nah, Xun Huang, Arash Vahdat, Jiaming Song, Karsten Kreis, Miika Aittala et al. “ediffi: Text-to-image diffusion models with an ensemble of expert denoisers.” arXiv preprint arXiv:2211.01324, 2022<\/em>.<\/li>\n
  10. https:\/www.midjourney.com\/home\/?callbackUrl=%2Fapp%2F<\/a><\/li>\n
  11. https:\/github.com\/CompVis\/stable-diffusion<\/a><\/li>\n
  12. Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, Bj\u00f6rn Ommer, \u201cHigh-Resolution Image Synthesis with Latent Diffusion Models,\u201d IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR)<\/em> 2022.<\/li>\n
  13. Ruiz, Nataniel, Yuanzhen Li, Varun Jampani, Yael Pritch, Michael Rubinstein, and Kfir Aberman. “Dreambooth: Fine tuning text-to-image diffusion models for subject-driven generation.” arXiv preprint arXiv:2208.12242<\/em>, 2022.<\/li>\n
  14. https:\/qz.com\/1699819\/a-new-kind-of-cybercrime-uses-ai-and-your-voice-against-you\/<\/a><\/li>\n
  15. https:\/openai.com\/blog\/new-ai-classifier-for-indicating-ai-written-text<\/a><\/li>\n
  16. Mitchell, Eric, Yoonho Lee, Alexander Khazatsky, Christopher D. Manning, and Chelsea Finn. “DetectGPT: Zero-Shot Machine-Generated Text Detection using Probability Curvature.” arXiv preprint arXiv:2301.11305, 2023.<\/li>\n
  17. Kirchenbauer, John, Jonas Geiping, Yuxin Wen, Jonathan Katz, Ian Miers, and Tom Goldstein. “A watermark for large language models.” arXiv preprint arXiv:2301.10226, 2023.<\/li>\n
  18. Somepalli, Gowthami, Vasu Singla, Micah Goldblum, Jonas Geiping, and Tom Goldstein. “Diffusion Art or Digital Forgery? Investigating Data Replication in Diffusion Models.” arXiv preprint arXiv:2212.03860, 2022.<\/li>\n
  19. Carlini, Nicholas, Jamie Hayes, Milad Nasr, Matthew Jagielski, Vikash Sehwag, Florian Tram\u00e8r, Borja Balle, Daphne Ippolito, and Eric Wallace. “Extracting training data from diffusion models.” arXiv preprint arXiv:2301.13188<\/em>, 2023.<\/li>\n
  20. Yu, Ning, Vladislav Skripniuk, Sahar Abdelnabi, and Mario Fritz. \u201cArtificial fingerprinting for generative models: Rooting deepfake attribution in training data.\u201d In IEEE\/CVF International Conference on Computer Vision (CVPR)<\/em>, pp. 14448-14457. 2021.<\/li>\n
  21. Shan, Shawn, Jenna Cryan, Emily Wenger, Haitao Zheng, Rana Hanocka, and Ben Y. Zhao. “GLAZE: Protecting Artists from Style Mimicry by Text-to-Image Models.” arXiv preprint arXiv:2302.04222<\/em> (2023).<\/li>\n
  22. Gandikota, Rohit, Joanna Materzynska, Jaden Fiotto-Kaufman, and David Bau. “Erasing Concepts from Diffusion Models.” arXiv preprint arXiv:2303.07345<\/em> (2023).<\/li>\n<\/ol>\n","protected":false},"excerpt":{"rendered":"

    \u751f\u6210\u5f0f AI \u7684\u84ec\u52c3\u767c\u5c55\u8207\u5176\u5e36\u4f86\u7684\u6311\u6230 2022 \u5e74\u5e95\uff0c ChatGPT \u6a6b\u7a7a\u51fa\u4e16 1\u3002 \u300cChatGPT\u300d\u662f […]<\/p>\n","protected":false},"author":1,"featured_media":30797,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"_et_pb_use_builder":"","_et_pb_old_content":"","_et_gb_content_width":"","footnotes":""},"categories":[990,983],"tags":[],"acf":[],"_links":{"self":[{"href":"https:\/newsletter.sinica.edu.tw\/wp-json\/wp\/v2\/posts\/30765"}],"collection":[{"href":"https:\/newsletter.sinica.edu.tw\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/newsletter.sinica.edu.tw\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/newsletter.sinica.edu.tw\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/newsletter.sinica.edu.tw\/wp-json\/wp\/v2\/comments?post=30765"}],"version-history":[{"count":1,"href":"https:\/newsletter.sinica.edu.tw\/wp-json\/wp\/v2\/posts\/30765\/revisions"}],"predecessor-version":[{"id":30864,"href":"https:\/newsletter.sinica.edu.tw\/wp-json\/wp\/v2\/posts\/30765\/revisions\/30864"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/newsletter.sinica.edu.tw\/wp-json\/wp\/v2\/media\/30797"}],"wp:attachment":[{"href":"https:\/newsletter.sinica.edu.tw\/wp-json\/wp\/v2\/media?parent=30765"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/newsletter.sinica.edu.tw\/wp-json\/wp\/v2\/categories?post=30765"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/newsletter.sinica.edu.tw\/wp-json\/wp\/v2\/tags?post=30765"}],"curies":[{"name":"wp","href":"https:\/api.w.org\/{rel}","templated":true}]}}