第九屆科技藝術國際學術研討會主題演講 II 邀請到Remi擔任講者，帶來演講：The Unlearning Machine《反學習機器》
演講摘要：
This talk was written without AI assistance. It will be imperfect and that is the point.
AI did not arrive in our studio all at once. It arrived like a new colleague — one task at a time — until one day we could not work without it.
hellobastworkshop is a new media and public art studio in Taoyuan. We build large outdoor metal installations and interactive light works.
The first half of this talk is about how good it has been. In three years we built tools we could not have built before. A configurator that designs a metal structure and sends the cutting files straight to the laser. A sequencer that invents light sequences for sixteen motorised LED panels. A tracking system that lets an installation see the people standing inside it. These tools are not experiments. They work every day, and the studio would be slower and smaller without them. I will show them running.
The second half is what I think about at night. The same tools changed our work, our careers, and our days. Something is being added, but something is also being taken away, and the second part is harder to see.
Learning has three steps: you study, you practise, you find out where you were wrong. Generative tools remove the middle one — the boring one, and the only one that actually builds skill. AI makes you better if you already know how. It replaces you quietly if you don’t.
What worries me is not my own work. It is the people arriving now: young fabricators, junior developers, art students. They are entering a profession where the beginner’s jobs have already been given to the machine. The bottom rungs of the ladder are gone.
AI 不是一夕之間進到我們工作室的。它比較像是新來的同事——一次接一件事——直到有一天，我們發現少了它就做不下去。
hellobastworkshop 是一間在桃園的新媒體與公共藝術工作室，做的是大型戶外金屬裝置，還有互動光影作品。
這場演講的前半段，講的是AI帶來多少好處。三年之內，我們做出了以前做不出來的工具：一套能算出金屬結構、直接輸出圖檔進雷切的參數化工具；一套會自己編排十六片馬達 LED 面板燈光序列的程式；還有一套讓裝置「看得見」站在裡面的人的追蹤系統。這些都不是做好看的實驗，是每天在跑的東西。沒有它們，工作室只會做得更慢、規模更小。我會直接用實例來給大家看。
後半段，講的是我晚上會想的事。同樣是這些工具，它們改變了我們做事的方式、我們的職涯，還有我們每一天的日子。有些東西被加進來了，但也有些東西被拿走了——而被拿走的那部分，比較不容易看見。
學一件事，不外乎三步：先學，再練，然後發現自己哪裡做錯。生成式工具把中間那一步拿掉了——那一步最無聊，卻也是唯一真正在練功的一步。你本來就會，AI 會讓你做得更好；你要是不會，它就默默地把你換掉。
我擔心的其實不是自己。是現在正要入行的人：年輕的鐵工師傅、剛出社會的工程師、還在讀書的藝術系學生。他們要進的這一行，本來屬於新手的那些活，早就被機器接走了。往上爬的那道梯子，最下面幾階已經不見了。
