AI forum 2016 . Pannels . Day I

Pannel 1

直覺{policy network}
思考{MCTS}
評估{value network}

COTS HPC:

Porblems

P1

  1. 各位老師大家好, ai big data machine learning 需要 data, value creation. 在台灣比較有機會做到這件事情?

enable, disease marker. open data. e-commerce.

狗的攝影機丟給狗吃. 固定 service. 狗叫狗臉問題. value driven.

競爭威脅 3 年噴聲音辨識.

攻擊資料

ai 會像 micro process 到處都是.

  1. AlphaGo impact, 以後怎麼教

架構都要教,順序要調整.

theory of computer games. 先教遊戲做出東西有感覺, 之後在教東西. model base, agent

P2

  1. 系統分析? uml or 白皮書? :=> 使用者對於系統分析的期待是什麼?

原本 circumstence 是一直都很穩定,如今改為可變動。 balance: {fine-tune, automation}

  1. 您怎麼看待, uiux 需要的背景知識是什麼?(田神的問題喔 XDD 真人的聲音好有磁性。)

雙向學習的過程,很難講誰比較懂 uiux 喔。 (眾笑焉)

了解懂 machine learning terminaligy 的 designer.

// industrial injuction moding

P3

  1. how do you envision ai may make iot better?

abstracted physical object to simulate device-environment.

“抽象化”設計的人講”具象化”,描述形狀,又為 affordance.

fog-computing, dynamic tuning. intelligent container enbeded iot.

  1. What is the potential pitfall of smart iot?

menipulate the smartness is a good business.

  • 會猶豫的門 ->> implicitly alarm
  1. Please predict that, how long time would let iot really understand us?

forcast bases 蒸汽機, 大量失業時.

social factor.

  1. Please talk about “nudge” of moral ~ implicitly change the behavior of users.

人難猜,為何人必須要被猜。怎麼教人了解 smart things rather than 怎麼讓 smart things 了解人:幫助人了解這複雜的世界,當世界越來越複雜時,我要有更多交互。

  1. AI 哲學,意識延伸?

hard to answer, but I would suggest to build affordance design and control