送交者: cornbug 于 2016-03-13, 09:28:43:

回答: 我也有这样的怀疑。团队是否在前三场结束后对阿尔法狗做过调试。 由 asasasa 于 2016-03-13, 08:21:57:

• after you played with AlphaGo for a while, maybe even I can beat it - conner (33 bytes) 2016-03-13, 09:59:27 (803659)
  • The problem is if it wins then it is not a "mistake" or a "bad" move - conner (71 bytes) 2016-03-13, 10:08:57 (803664)
    • that will not happen - pgss (88 bytes) 2016-03-13, 10:15:31 (803667)
      • The damage of such problems could be even larger when playing with itself. - conner (32 bytes) 2016-03-13, 10:22:43 (803670)
        • i believe those must have come up in the past and got resolved (无内容) - pgss (0 bytes) 2016-03-13, 10:33:26 (803674)
      • once the move comes into the radar screen of policy network, - pgss (197 bytes) 2016-03-13, 10:18:14 (803669)
  • it will not make the same mistake twice - pgss (35 bytes) 2016-03-13, 10:06:04 (803662)

加跟贴