{"id":29550,"date":"2025-03-29T00:59:42","date_gmt":"2025-03-28T16:59:42","guid":{"rendered":"https:\/\/www.aisharenet.com\/?p=29550"},"modified":"2025-03-29T00:59:42","modified_gmt":"2025-03-28T16:59:42","slug":"arc-agi-2-chengjijieai","status":"publish","type":"post","link":"https:\/\/www.kdjingpai.com\/ja\/arc-agi-2-chengjijieai\/","title":{"rendered":"ARC-AGI-2 \u6210\u7ee9\u63ed\u6653\uff1a\u5168\u90e8 AI \u6a21\u578b\u63a8\u7406\u80fd\u529b\u906d\u9047\u6ed1\u94c1\u5362"},"content":{"rendered":"

\"ARC-AGI-2<\/p>\n

\u8861\u91cf\u901a\u7528\u4eba\u5de5\u667a\u80fd\uff08AGI\uff09\u8fdb\u5c55\u7684\u57fa\u51c6\u81f3\u5173\u91cd\u8981\u3002\u6709\u6548\u7684\u57fa\u51c6\u80fd\u591f\u63ed\u793a\u80fd\u529b\uff0c\u800c\u4f18\u79c0\u7684\u57fa\u51c6\u66f4\u80fd\u6fc0\u53d1\u7814\u7a76\u65b9\u5411\u3002ARC Prize \u57fa\u91d1\u4f1a\u81f4\u529b\u4e8e\u901a\u8fc7\u5176\u00a0ARC-AGI<\/code>\u00a0\u7cfb\u5217\u57fa\u51c6\u626e\u6f14\u8fd9\u6837\u7684\u89d2\u8272\uff0c\u5f15\u5bfc\u7814\u7a76\u529b\u91cf\u5173\u6ce8\u771f\u6b63\u7684\u901a\u7528\u667a\u80fd\u3002\u6700\u65b0\u7684\u00a0ARC-AGI-2<\/code>\u00a0\u57fa\u51c6\u53ca\u5176\u521d\u6b65\u6d4b\u8bd5\u7ed3\u679c\uff0c\u4e3a\u5f53\u524d AI \u80fd\u529b\u7684\u8fb9\u754c\u548c\u6548\u7387\u95ee\u9898\u6572\u54cd\u4e86\u8b66\u949f\u3002<\/p>\n

ARC-AGI-1<\/code>\u00a0\u81ea 2019 \u5e74\u63a8\u51fa\u4ee5\u6765\uff0c\u5728\u8ffd\u8e2a AGI \u8fdb\u5c55\u65b9\u9762\u626e\u6f14\u4e86\u72ec\u7279\u89d2\u8272\uff0c\u66fe\u5e2e\u52a9\u8bc6\u522b AI \u4f55\u65f6\u5f00\u59cb\u8d85\u8d8a\u5355\u7eaf\u7684\u6a21\u5f0f\u8bb0\u5fc6\u3002\u968f\u540e\u7684\u00a0ARC Prize 2024<\/code>\u00a0\u7ade\u8d5b\u4e5f\u5438\u5f15\u4e86\u5927\u91cf\u7814\u7a76\u8005\u63a2\u7d22\u65b0\u7684\u6d4b\u8bd5\u65f6\u9002\u5e94\u6027\u601d\u8def\u3002<\/p>\n

\u7136\u800c\uff0c\u901a\u5f80 AGI \u7684\u9053\u8def\u4f9d\u7136\u6f2b\u957f\u3002\u5f53\u524d\u7684\u8fdb\u5c55\uff0c\u5982\u00a0OpenAI<\/code>\u00a0\u7684\u00a0o3<\/code>\u00a0\u7cfb\u7edf\u6240\u5c55\u73b0\u7684\uff0c\u6216\u8bb8\u53ea\u662f\u5728\u201c\u6d41\u4f53\u667a\u529b\u201d\u8fd9\u4e2a\u7ef4\u5ea6\u4e0a\u5b9e\u73b0\u4e86\u6709\u9650\u7a81\u7834\u3002\u8fd9\u4e9b\u7cfb\u7edf\u4e0d\u4ec5\u6548\u7387\u4f4e\u4e0b\uff0c\u4e14\u4ecd\u9700\u5927\u91cf\u4eba\u5de5\u76d1\u7763\u3002\u663e\u7136\uff0cAGI \u7684\u5b9e\u73b0\u9700\u8981\u66f4\u591a\u6e90\u5934\u4e0a\u7684\u521b\u65b0\u3002<\/p>\n

\u65b0\u6311\u6218\uff1aARC-AGI-2\uff0c\u4e13\u4e3a\u66b4\u9732 AI \u5f31\u70b9\u800c\u751f<\/h3>\n

\u4e3a\u6b64\uff0cARC Prize \u57fa\u91d1\u4f1a\u73b0\u5df2\u63a8\u51fa\u00a0ARC-AGI-2<\/code>\u00a0\u57fa\u51c6\u3002\u5176\u8bbe\u8ba1\u76ee\u6807\u660e\u786e\uff1a\u5bf9 AI\uff08\u5c24\u5176\u662f\u63a8\u7406\u7cfb\u7edf\uff09\u6765\u8bf4\u96be\u5ea6\u663e\u8457\u63d0\u5347\uff0c\u540c\u65f6\u4fdd\u6301\u5bf9\u4eba\u7c7b\u7684\u76f8\u5bf9\u6613\u5904\u7406\u6027\u3002\u8fd9\u5e76\u975e\u7b80\u5355\u7684\u96be\u5ea6\u589e\u52a0\uff0c\u800c\u662f\u9488\u5bf9\u6027\u5730\u6311\u6218\u5f53\u524d AI \u65b9\u6cd5\u96be\u4ee5\u514b\u670d\u7684\u969c\u788d\u3002<\/p>\n

\u8bbe\u8ba1\u54f2\u5b66\uff1a\u805a\u7126\u4eba\u6613 AI \u96be\u7684\u667a\u80fd\u9e3f\u6c9f<\/h3>\n

\u4e0e\u5176\u4ed6\u8bb8\u591a\u8ffd\u6c42\u8d85\u4eba\u80fd\u529b\u7684 AI \u57fa\u51c6\u4e0d\u540c\uff0cARC-AGI<\/code>\u00a0\u805a\u7126\u4e8e\u90a3\u4e9b\u5bf9\u4eba\u7c7b\u6765\u8bf4\u76f8\u5bf9\u5bb9\u6613\uff0c\u4f46\u5bf9\u5f53\u524d AI \u800c\u8a00\u5374\u6781\u5176\u56f0\u96be\u7684\u4efb\u52a1\u3002\u8fd9\u79cd\u7b56\u7565\u65e8\u5728\u63ed\u793a\u90a3\u4e9b\u65e0\u6cd5\u901a\u8fc7\u7b80\u5355\u5730\u201c\u6269\u5927\u89c4\u6a21\u201d\u5c31\u80fd\u5f25\u8865\u7684\u80fd\u529b\u7f3a\u53e3\u3002\u901a\u7528\u667a\u80fd\u7684\u6838\u5fc3\u5728\u4e8e\u4ece\u6709\u9650\u7ecf\u9a8c\u4e2d\u9ad8\u6548\u6cdb\u5316\u548c\u5e94\u7528\u77e5\u8bc6\u7684\u80fd\u529b\uff0c\u800c\u8fd9\u6b63\u662f\u5f53\u524d AI \u7684\u8f6f\u808b\u3002<\/p>\n

ARC-AGI-2\uff1a\u96be\u5ea6\u5347\u7ea7\uff0c\u76f4\u9762 AI \u63a8\u7406\u8f6f\u808b<\/h3>\n

ARC-AGI-2<\/code>\u00a0\u5728\u00a0ARC-AGI-1<\/code>\u00a0\u7684\u57fa\u7840\u4e0a\uff0c\u663e\u8457\u63d0\u9ad8\u4e86\u5bf9 AI \u7684\u8981\u6c42\uff0c\u5f3a\u8c03\u9ad8\u9002\u5e94\u6027\u4e0e\u9ad8\u6548\u7387\u7684\u7ed3\u5408\u3002\u901a\u8fc7\u5206\u6790\u524d\u6cbf AI \u5728\u5148\u524d\u4efb\u52a1\u4e0a\u7684\u5931\u8d25\u6848\u4f8b\uff0cARC-AGI-2<\/code>\u00a0\u5f15\u5165\u4e86\u66f4\u591a\u8003\u9a8c\u7b26\u53f7\u89e3\u91ca\u3001\u7ec4\u5408\u63a8\u7406\u3001\u4e0a\u4e0b\u6587\u89c4\u5219\u5e94\u7528\u7b49\u80fd\u529b\u7684\u6311\u6218\u3002\u8fd9\u4e9b\u4efb\u52a1\u65e8\u5728\u8feb\u4f7f AI \u8d85\u8d8a\u8868\u9762\u6a21\u5f0f\u5339\u914d\uff0c\u8fdb\u884c\u66f4\u6df1\u5c42\u6b21\u7684\u62bd\u8c61\u548c\u63a8\u7406\u3002<\/p>\n

ARC-AGI-2 \u6210\u7ee9\u5355\uff1a\u4e00\u4efd\u4e25\u5cfb\u7684\u73b0\u5b9e\u6620\u7167<\/h3>\n

\u6700\u65b0\u516c\u5e03\u7684\u00a0ARC-AGI<\/code>\u00a0\u6392\u884c\u699c\u6570\u636e\uff0c\u4e3a\u6211\u4eec\u63cf\u7ed8\u4e86\u4e00\u5e45\u5173\u4e8e\u5f53\u524d AI \u80fd\u529b\u7684\u4e25\u5cfb\u56fe\u666f\u3002\u8fd9\u4efd\u6570\u636e\u4e0d\u4ec5\u8bc1\u5b9e\u4e86\u00a0ARC-AGI-2<\/code>\u00a0\u7684\u6311\u6218\u6027\uff0c\u66f4\u6df1\u523b\u5730\u63ed\u793a\u4e86 AI \u5728\u901a\u7528\u63a8\u7406\u80fd\u529b\u548c\u6548\u7387\u4e0a\u7684\u5de8\u5927\u9e3f\u6c9f\u3002<\/p>\n

\"ARC-AGI-2<\/p>\n

Leaderboard \u6570\u636e\u6982\u89c8<\/strong><\/p>\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
AI System<\/th>\nOrganization<\/th>\nSystem Type<\/th>\nARC-AGI-1<\/th>\nARC-AGI-2<\/th>\nCost\/Task<\/th>\nCode \/ Paper<\/th>\n<\/tr>\n<\/thead>\n
Human Panel<\/td>\nHuman<\/td>\nN\/A<\/td>\n98.0%<\/td>\n100.0%<\/td>\n$17.00<\/td>\n\u2014<\/td>\n<\/tr>\n
o3 (low)*<\/td>\nOpenAI<\/td>\nCoT + Synthesis<\/td>\n75.7%<\/td>\n4.0%<\/td>\n$200.00<\/td>\n\ud83d\udcc4<\/td>\n<\/tr>\n
o1 (high)<\/td>\nOpenAI<\/td>\nCoT<\/td>\n32.0%<\/td>\n3.0%<\/td>\n$4.45<\/td>\n\ud83d\udcbb<\/td>\n<\/tr>\n
ARChitects<\/td>\nARC Prize 2024<\/td>\nCustom<\/td>\n56.0%<\/td>\n2.5%<\/td>\n$0.200<\/td>\n\ud83d\udcc4\ud83d\udcbb<\/td>\n<\/tr>\n
o3-mini<\/a> (medium)<\/td>\nOpenAI<\/td>\nCoT<\/td>\n29.1%<\/td>\n1.7%<\/td>\n$0.280<\/td>\n\ud83d\udcbb<\/td>\n<\/tr>\n
Icecuber<\/td>\nARC Prize 2024<\/td>\nCustom<\/td>\n17.0%<\/td>\n1.6%<\/td>\n$0.130<\/td>\n\ud83d\udcbb<\/td>\n<\/tr>\n
o3-mini (high)<\/td>\nOpenAI<\/td>\nCoT<\/td>\n35.0%<\/td>\n1.5%<\/td>\n$0.410<\/td>\n\ud83d\udcbb<\/td>\n<\/tr>\n
Gemini<\/a> 2.0 Flash<\/td>\nGoogle<\/td>\nBase LLM<\/td>\nN\/A<\/td>\n1.3%<\/td>\n$0.004<\/td>\n\ud83d\udcbb<\/td>\n<\/tr>\n
o1 (medium)<\/td>\nOpenAI<\/td>\nCoT<\/td>\n31.0%<\/td>\n1.3%<\/td>\n$2.76<\/td>\n\ud83d\udcbb<\/td>\n<\/tr>\n
Deepseek<\/a> R1<\/td>\nDeepseek<\/td>\nCoT<\/td>\n15.8%<\/td>\n1.3%<\/td>\n$0.080<\/td>\n\ud83d\udcbb<\/td>\n<\/tr>\n
Gemini-2.5-Pro-Exp-03-25 **<\/td>\nGoogle<\/td>\nCoT<\/td>\n12.5%<\/td>\n1.3%<\/td>\nN\/A<\/td>\n\ud83d\udcbb<\/td>\n<\/tr>\n
o1-pro<\/td>\nOpenAI<\/td>\nCoT + Synthesis<\/td>\n50.0%<\/td>\n1.0%<\/td>\n$39.00<\/td>\n\u2014<\/td>\n<\/tr>\n
Claude<\/a> 3.7 (8K)<\/td>\nAnthropic<\/a><\/td>\nCoT<\/td>\n21.2%<\/td>\n0.9%<\/td>\n$0.360<\/td>\n\ud83d\udcbb<\/td>\n<\/tr>\n
Gemini 1.5 Pro<\/td>\nGoogle<\/td>\nBase LLM<\/td>\nN\/A<\/td>\n0.8%<\/td>\n$0.040<\/td>\n\ud83d\udcbb<\/td>\n<\/tr>\n
GPT-4.5<\/td>\nOpenAI<\/td>\nBase LLM<\/td>\n10.3%<\/td>\n0.8%<\/td>\n$2.10<\/td>\n\ud83d\udcbb<\/td>\n<\/tr>\n
o1 (low)<\/td>\nOpenAI<\/td>\nCoT<\/td>\n25.0%<\/td>\n0.8%<\/td>\n$1.44<\/td>\n\ud83d\udcbb<\/td>\n<\/tr>\n
Claude 3.7 (16K)<\/td>\nAnthropic<\/td>\nCoT<\/td>\n28.6%<\/td>\n0.7%<\/td>\n$0.510<\/td>\n\ud83d\udcbb<\/td>\n<\/tr>\n
Claude 3.7 (1K)<\/td>\nAnthropic<\/td>\nCoT<\/td>\n11.6%<\/td>\n0.4%<\/td>\n$0.140<\/td>\n\ud83d\udcbb<\/td>\n<\/tr>\n
Claude 3.7<\/td>\nAnthropic<\/td>\nBase LLM<\/td>\n13.6%<\/td>\n0.0%<\/td>\n$0.120<\/td>\n\ud83d\udcbb<\/td>\n<\/tr>\n
GPT-4o<\/td>\nOpenAI<\/td>\nBase LLM<\/td>\n4.5%<\/td>\n0.0%<\/td>\n$0.080<\/td>\n\ud83d\udcbb<\/td>\n<\/tr>\n
GPT-4o-mini<\/td>\nOpenAI<\/td>\nBase LLM<\/td>\nN\/A<\/td>\n0.0%<\/td>\n$0.010<\/td>\n\ud83d\udcbb<\/td>\n<\/tr>\n
o3-mini (low)<\/td>\nOpenAI<\/td>\nCoT<\/td>\n11.0%<\/td>\n0.0%<\/td>\n$0.060<\/td>\n\ud83d\udcbb<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

(\u6ce8\uff1a\u8868\u683c\u4e2d * \u8868\u793a\u521d\u6b65\u4f30\u8ba1\u503c\uff0c<\/em>* \u8868\u793a\u5b9e\u9a8c\u6027\u6a21\u578b)*<\/p>\n

\u6df1\u523b\u89e3\u8bfb\uff1a\u6570\u636e\u80cc\u540e\u7684\u8b66\u793a<\/strong><\/p>\n

    \n
  1. \u4eba\u7c7b vs AI\uff1a\u96be\u4ee5\u903e\u8d8a\u7684\u9e3f\u6c9f<\/strong>
    \n\u6700\u5f15\u4eba\u6ce8\u76ee\u7684\u83ab\u8fc7\u4e8e\u4eba\u7c7b\u4e0e AI \u8868\u73b0\u7684\u60ac\u6b8a\u5bf9\u6bd4\u3002\u5728\u00a0ARC-AGI-2<\/code>\u00a0\u4e0a\uff0c\u4eba\u7c7b\u5c0f\u7ec4\u53d6\u5f97\u4e86 100% \u7684\u5b8c\u7f8e\u5f97\u5206\uff0c\u800c\u8868\u73b0\u6700\u597d\u7684 AI \u7cfb\u7edf\u2014\u2014\u00a0OpenAI<\/code>\u00a0\u7684\u00a0o3 (low)<\/code>\uff0c\u5f97\u5206\u4ec5\u4e3a 4.0%\u3002\u5176\u4ed6\u77e5\u540d\u6a21\u578b\uff0c\u5982\u00a0Gemini 2.0 Flash<\/code>\u3001Deepseek R1<\/a><\/code>\u00a0\u7b49\uff0c\u5f97\u5206\u5747\u5728 1.3% \u5de6\u53f3\u5f98\u5f8a\u3002\u66f4\u4ee4\u4eba\u9707\u60ca\u7684\u662f\uff0c\u50cf\u00a0Claude 3.7<\/code>\u3001GPT-4o<\/code>\u3001GPT-4o-mini<\/code>\u00a0\u8fd9\u4e9b\u5728\u5176\u4ed6\u9886\u57df\u8868\u73b0\u51fa\u8272\u7684\u57fa\u7840\u5927\u8bed\u8a00\u6a21\u578b\uff08Base LLM\uff09\uff0c\u5728\u00a0ARC-AGI-2<\/code>\u00a0\u4e0a\u7684\u5f97\u5206\u76f4\u63a5\u5f52\u96f6\u3002\u8fd9\u65e0\u60c5\u5730\u63ed\u793a\u4e86\uff0c\u5c3d\u7ba1 AI \u5728\u7279\u5b9a\u4efb\u52a1\u4e0a\u80fd\u529b\u60ca\u4eba\uff0c\u4f46\u5728\u9762\u5bf9\u9700\u8981\u7075\u6d3b\u3001\u62bd\u8c61\u3001\u901a\u7528\u63a8\u7406\u80fd\u529b\u7684\u65b0\u9896\u95ee\u9898\u65f6\uff0c\u4e0e\u4eba\u7c7b\u76f8\u6bd4\u4ecd\u6709\u672c\u8d28\u4e0a\u7684\u5dee\u8ddd\u3002<\/li>\n
  2. \u4ece AGI-1 \u5230 AGI-2\uff1aAI \u80fd\u529b\u7684\u65ad\u5d16\u5f0f\u4e0b\u8dcc<\/strong>
    \n\u51e0\u4e4e\u6240\u6709\u53c2\u4e0e\u6d4b\u8bd5\u7684 AI \u7cfb\u7edf\uff0c\u5728\u4ece\u00a0ARC-AGI-1<\/code>\u00a0\u8fc7\u6e21\u5230\u00a0ARC-AGI-2<\/code>\u00a0\u65f6\uff0c\u6027\u80fd\u90fd\u51fa\u73b0\u4e86\u6025\u5267\u4e0b\u6ed1\u3002\u4f8b\u5982\uff0co3 (low)<\/code>\u00a0\u4ece 75.7% \u9aa4\u964d\u81f3 4.0%\uff0co1-pro<\/code>\u00a0\u4ece\u7ea6 50% \u8dcc\u81f3 1.0%\uff0cARChitects<\/code>\u00a0\u4ece 56.0% \u964d\u81f3 2.5%\u3002\u8fd9\u4e00\u666e\u904d\u73b0\u8c61\u5f3a\u70c8\u6697\u793a\uff0cARC-AGI-2<\/code>\u00a0\u786e\u5b9e\u6210\u529f\u89e6\u53ca\u4e86\u5f53\u524d AI \u65b9\u6cd5\u8bba\u7684\u201c\u75db\u70b9\u201d\uff0c\u65e0\u8bba\u662f\u57fa\u4e8e CoT\uff08\u601d\u7ef4\u94fe\uff09\u3001Synthesis\uff08\u7efc\u5408\/\u641c\u7d22\uff09\u8fd8\u662f\u5176\u4ed6\u5b9a\u5236\u65b9\u6cd5\uff0c\u90fd\u96be\u4ee5\u6709\u6548\u5e94\u5bf9\u00a0ARC-AGI-2<\/code>\u00a0\u6240\u4ee3\u8868\u7684\u63a8\u7406\u6311\u6218\u3002<\/li>\n
  3. \u7cfb\u7edf\u7c7b\u578b\u4e0e\u6548\u7387\uff1a\u9ad8\u6210\u672c\u5e76\u672a\u5e26\u6765\u9ad8\u667a\u80fd<\/strong>
    \n\u6392\u884c\u699c\u8fdb\u4e00\u6b65\u63ed\u793a\u4e86\u4e0d\u540c AI \u7cfb\u7edf\u7c7b\u578b\u5728\u00a0ARC-AGI-2<\/code>\u00a0\u4e0a\u7684\u8868\u73b0\u5dee\u5f02\u548c\u4e25\u5cfb\u7684\u6548\u7387\u95ee\u9898\uff1a<\/p>\n
      \n
    • CoT + Synthesis \u7cfb\u7edf<\/strong>\u00a0(o3 (low)<\/code>,\u00a0o1-pro<\/code>) \u867d\u7136\u53d6\u5f97\u4e86\u76f8\u5bf9\u6700\u9ad8\u7684 AI \u5206\u6570\uff084.0% \u548c 1.0%\uff09\uff0c\u4f46\u5176\u6210\u672c\u5374\u9ad8\u5f97\u60ca\u4eba\uff08\u5206\u522b\u4e3a $200 \u548c $39 \u6bcf\u4efb\u52a1\uff09\u3002\u8fd9\u8868\u660e\u590d\u6742\u7684\u63a8\u7406\u52a0\u641c\u7d22\u7b56\u7565\u6216\u8bb8\u80fd\u201c\u6324\u201d\u51fa\u4e00\u70b9\u5206\u6570\uff0c\u4f46\u6548\u7387\u6781\u5176\u4f4e\u4e0b\u3002<\/li>\n
    • \u7eaf CoT \u7cfb\u7edf<\/strong>\u00a0\u8868\u73b0\u53c2\u5dee\u4e0d\u9f50\uff0c\u5f97\u5206\u666e\u904d\u5728 1%-3% \u4e4b\u95f4\uff0c\u6210\u672c\u4e5f\u4ece\u51e0\u7f8e\u5206\u5230\u51e0\u7f8e\u5143\u4e0d\u7b49\u3002\u8fd9\u4f3c\u4e4e\u8bf4\u660e\u4ec5\u9760 CoT \u4e0d\u8db3\u4ee5\u5e94\u5bf9\u6311\u6218\u3002<\/li>\n
    • \u57fa\u7840\u5927\u8bed\u8a00\u6a21\u578b (Base LLM)<\/strong>\u00a0(GPT-4.5<\/code>,\u00a0Gemini 1.5 Pro<\/code>,\u00a0Claude 3.7<\/code>,\u00a0GPT-4o<\/code>) \u7684\u8868\u73b0\u582a\u79f0\u6e83\u8d25\uff0c\u5f97\u5206\u591a\u4e3a 0% \u6216\u63a5\u8fd1 0%\u3002\u8fd9\u6709\u529b\u5730\u53cd\u9a73\u4e86\u201c\u89c4\u6a21\u5c31\u662f\u4e00\u5207\u201d\u7684\u89c2\u70b9\uff0c\u81f3\u5c11\u5728\u00a0ARC-AGI<\/code>\u00a0\u6240\u8861\u91cf\u7684\u901a\u7528\u6d41\u4f53\u667a\u529b\u65b9\u9762\u662f\u5982\u6b64\u3002<\/li>\n
    • \u5b9a\u5236\u7cfb\u7edf (Custom)<\/strong>\u00a0(ARChitects<\/code>,\u00a0Icecuber<\/code>) \u4f5c\u4e3a\u00a0ARC Prize 2024<\/code>\u00a0\u7684\u4ea7\u7269\uff0c\u4ee5\u6781\u4f4e\u7684\u6210\u672c\uff08\u7ea6 $0.1-$0.2 \u6bcf\u4efb\u52a1\uff09\u53d6\u5f97\u4e86\u4e0e\u5176\u4ed6 AI \u7cfb\u7edf\u76f8\u5f53\u751a\u81f3\u7565\u4f18\u7684\u6210\u7ee9\uff082.5%, 1.6%\uff09\u3002\u8fd9\u6216\u8bb8\u6697\u793a\u4e86\uff0c\u9488\u5bf9\u6027\u7684\u3001\u8f7b\u91cf\u7ea7\u7684\u7b97\u6cd5\u6216\u67b6\u6784\uff0c\u5728\u89e3\u51b3\u8fd9\u7c7b\u95ee\u9898\u4e0a\u53ef\u80fd\u6bd4\u5e9e\u5927\u800c\u901a\u7528\u7684\u6a21\u578b\u66f4\u6709\u6f5c\u529b\uff0c\u4e5f\u51f8\u663e\u4e86\u5f00\u653e\u7ade\u8d5b\u548c\u793e\u533a\u521b\u65b0\u7684\u4ef7\u503c\u3002<\/li>\n<\/ul>\n<\/li>\n
    • \u6548\u7387\u5371\u673a\uff1a\u667a\u80fd\u4e0d\u80fd\u53ea\u770b\u5206\u6570<\/strong>
      \nARC Prize<\/code>\u00a0\u5c06\u201c\u6210\u672c\/\u4efb\u52a1\u201d\u4f5c\u4e3a\u5173\u952e\u6307\u6807\u7eb3\u5165\u6392\u884c\u699c\uff0c\u610f\u4e49\u91cd\u5927\u3002\u6570\u636e\u663e\u793a\uff0c\u5373\u4f7f\u662f\u8868\u73b0\u6700\u597d\u7684 AI (o3 (low)<\/code>\u00a0\u5f97 4%)\uff0c\u5176\u5355\u4efb\u52a1\u6210\u672c ($200) \u4e5f\u662f\u4eba\u7c7b ($17 \u5f97 100%) \u7684\u5341\u500d\u4ee5\u4e0a\u3002\u800c\u4e00\u4e9b\u4f4e\u6210\u672c\u6a21\u578b\uff0c\u5982\u00a0Gemini 2.0 Flash<\/code>\uff0c\u867d\u7136\u6210\u672c\u6781\u4f4e ($0.004)\uff0c\u5f97\u5206\u4e5f\u4ec5\u6709 1.3%\u3002\u8fd9\u5f62\u6210\u4e86\u9c9c\u660e\u7684\u5bf9\u6bd4\uff1aAI \u8981\u4e48\u5f97\u5206\u6781\u4f4e\uff0c\u8981\u4e48\u6210\u672c\u9ad8\u6602\uff0c\u6216\u8005\u4e24\u8005\u517c\u5177\u3002\u667a\u80fd\u5e76\u975e\u4e0d\u8ba1\u6210\u672c\u5730\u83b7\u5f97\u6b63\u786e\u7b54\u6848\uff0c\u6548\u7387\u662f\u5176\u5185\u5728\u5c5e\u6027\u3002\u5f53\u524d AI \u5728\u00a0ARC-AGI-2<\/code>\u00a0\u4e0a\u5c55\u73b0\u7684\u201c\u80fd\u529b-\u6210\u672c\u201d\u66f2\u7ebf\uff0c\u65e0\u7591\u66b4\u9732\u4e86\u4e00\u573a\u6df1\u523b\u7684\u201c\u6548\u7387\u5371\u673a\u201d\u3002<\/li>\n<\/ol>\n

      \u6570\u636e\u96c6\u6784\u6210\u4e0e\u7ade\u8d5b\u7ec6\u8282<\/h3>\n

      ARC-AGI-2<\/code>\u00a0\u5305\u542b\u7ecf\u6821\u51c6\u7684\u8bad\u7ec3\u96c6\u548c\u8bc4\u4f30\u96c6\uff0c\u6cbf\u7528\u00a0pass@2<\/code>\u00a0\u8bc4\u5206\u673a\u5236\u3002\u4e3b\u8981\u53d8\u66f4\u5305\u62ec\u4efb\u52a1\u6570\u91cf\u589e\u52a0\u3001\u79fb\u9664\u6613\u88ab\u66b4\u529b\u7834\u89e3\u7684\u4efb\u52a1\u3001\u57fa\u4e8e\u4eba\u7c7b\u6d4b\u8bd5\u8fdb\u884c\u96be\u5ea6\u6821\u51c6\u3001\u5e76\u8bbe\u8ba1\u4e86\u9488\u5bf9\u6027\u7684\u65b0\u4efb\u52a1\u3002<\/p>\n

      ARC Prize 2025 \u7ade\u8d5b\u542f\u52a8\uff1a\u767e\u4e07\u7f8e\u5143\u60ac\u8d4f\u65b0\u601d\u8def<\/strong><\/p>\n

      \u4f34\u968f\u7740\u8fd9\u4efd\u4e25\u5cfb\u7684\u6210\u7ee9\u5355\uff0cARC Prize 2025<\/code>\u00a0\u7ade\u8d5b\u5728\u00a0Kaggle<\/code>\u00a0\u5e73\u53f0\u4e0a\u7ebf\uff083\u670826\u65e5-11\u67083\u65e5\uff09\uff0c\u603b\u5956\u91d1\u9ad8\u8fbe 100 \u4e07\u7f8e\u5143\u3002\u7ade\u8d5b\u73af\u5883\u9650\u5236\u4e86 API \u4f7f\u7528\u548c\u8ba1\u7b97\u8d44\u6e90\uff08\u7ea6 50 \u7f8e\u5143\/\u63d0\u4ea4\uff09\uff0c\u5e76\u5f3a\u5236\u8981\u6c42\u83b7\u5956\u8005\u5f00\u6e90\u89e3\u51b3\u65b9\u6848\u3002\u8fd9\u8fdb\u4e00\u6b65\u5f3a\u5316\u4e86\u5bf9\u6548\u7387\u548c\u521b\u65b0\u7684\u8981\u6c42\u3002<\/p>\n

      \u76f8\u8f83\u4e8e 2024 \u5e74\uff0c2025 \u5e74\u7ade\u8d5b\u7684\u4e3b\u8981\u53d8\u66f4\u65e5\u5fd7<\/strong>\u5305\u62ec\uff1a\u91c7\u7528\u00a0ARC-AGI-2<\/code>\u00a0\u6570\u636e\u96c6\u3001\u65b0\u7684\u6392\u884c\u699c\u62a5\u544a\u673a\u5236\u3001\u5f3a\u5316\u5f00\u6e90\u8981\u6c42\u3001\u8ba1\u7b97\u8d44\u6e90\u7ffb\u500d\u3001\u4ee5\u53ca\u989d\u5916\u7684\u9632\u8fc7\u62df\u5408\u63aa\u65bd\u3002<\/p>\n

      \u7ed3\u8bed\uff1a\u771f\u6b63\u7684\u7a81\u7834\u4e9f\u5f85\u65b0\u8303\u5f0f<\/h3>\n

      ARC-AGI-2<\/code>\u00a0\u7684\u6392\u884c\u699c\u6570\u636e\u5982\u540c\u4e00\u9762\u955c\u5b50\uff0c\u6e05\u6670\u5730\u6620\u7167\u51fa\u5f53\u524d AI \u5728\u901a\u7528\u63a8\u7406\u548c\u6548\u7387\u4e0a\u7684\u5c40\u9650\u6027\u3002\u5b83\u63d0\u9192\u6211\u4eec\uff0c\u901a\u5f80 AGI \u7684\u9053\u8def\u8fdc\u975e\u5766\u9014\uff0c\u7b80\u5355\u7684\u6a21\u578b\u653e\u5927\u6216\u589e\u52a0\u8ba1\u7b97\u8d44\u6e90\u53ef\u80fd\u4e0d\u8db3\u4ee5\u8de8\u8d8a\u773c\u524d\u7684\u9e3f\u6c9f\u3002\u771f\u6b63\u7684\u7a81\u7834\uff0c\u6216\u8bb8\u9700\u8981\u5168\u65b0\u7684\u601d\u8def\u3001\u4e0d\u540c\u7684\u67b6\u6784\uff0c\u751a\u81f3\u53ef\u80fd\u6765\u81ea\u5927\u578b\u5b9e\u9a8c\u5ba4\u4e4b\u5916\u7684\u521b\u65b0\u8005\u3002ARC Prize 2025<\/code>\u00a0\u6b63\u662f\u8fd9\u6837\u4e00\u4e2a\u547c\u5524\u65b0\u8303\u5f0f\u7684\u5e73\u53f0\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"

      \u8861\u91cf\u901a\u7528\u4eba\u5de5\u667a\u80fd\uff08AGI\uff09\u8fdb\u5c55\u7684\u57fa\u51c6\u81f3\u5173\u91cd\u8981\u3002\u6709\u6548\u7684\u57fa\u51c6\u80fd\u591f\u63ed\u793a\u80fd\u529b\uff0c\u800c\u4f18\u79c0\u7684\u57fa\u51c6\u66f4\u80fd\u6fc0\u53d1\u7814\u7a76\u65b9\u5411\u3002ARC Prize \u57fa\u91d1\u4f1a\u81f4\u529b\u4e8e\u901a\u8fc7\u5176\u00a0ARC-AGI\u00a0\u7cfb\u5217\u57fa\u51c6\u626e\u6f14\u8fd9\u6837\u7684\u89d2\u8272\uff0c\u5f15\u5bfc\u7814\u7a76\u529b\u91cf\u5173\u6ce8\u771f\u6b63\u7684\u901a\u7528\u667a\u80fd\u3002\u6700\u65b0\u7684\u00a0ARC-AGI-2\u00a0…<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[46],"tags":[],"class_list":["post-29550","post","type-post","status-publish","format-standard","hentry","category-news"],"_links":{"self":[{"href":"https:\/\/www.kdjingpai.com\/ja\/wp-json\/wp\/v2\/posts\/29550","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.kdjingpai.com\/ja\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.kdjingpai.com\/ja\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.kdjingpai.com\/ja\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.kdjingpai.com\/ja\/wp-json\/wp\/v2\/comments?post=29550"}],"version-history":[{"count":0,"href":"https:\/\/www.kdjingpai.com\/ja\/wp-json\/wp\/v2\/posts\/29550\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.kdjingpai.com\/ja\/wp-json\/wp\/v2\/media?parent=29550"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.kdjingpai.com\/ja\/wp-json\/wp\/v2\/categories?post=29550"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.kdjingpai.com\/ja\/wp-json\/wp\/v2\/tags?post=29550"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}