Yesterday there was Luis von Ahn’s IAAI talk regarding “Human Computation” at Pasadena Convention Center. He is an assistant professor of computer science at Carnegie Mellon University, and one of rising stars in AI and HCI and most active researchers to utilize human’s useful resources.
Here are my quick impressions of his talk:
1. He is a great presenter: clear and easy to understand his message, and he has a very good skill to transfer his idea and end up persuading audiences (or some random target people) to (naturally) agree with him.
2. He sees the existing CS problems in a completely different way: for several decades, many AI and CS researchers have tried to solve complex and (of course) complicated problems around us (e.g., optimization, machine translation, searching, labeling, etc…). However, for most of cases, people just have agreed that finding an exact and optimal solution of those problems in a polynomial time is not really feasible (let’s simply say NP-complete or NP-hard? well.. even NEXP-complete). Luis suggested a somewhat easy (but completely novel) way to tackle these difficulties. This also reminds me of “Egg of Columbus”.
3. He is very confident of his own work: we strongly need to believe ourselves especially when we try to explore some unknown domains. However, I realize it’s very hard to keep ourselves balanced between exploration and exploitation, and we meet significant level of uncertainties, and thus try keep pushing our work at our own peril. In this sense, keeping confidence and believing ourselves might show us some fantastic views not visible from initial starting position. Anyway, the point is that nobody knows that our new trial is correct or wrong. this is sufficient enough, right?
4. He is a good script writer for his talk: well.. this issue is somewhat related to my first point. I could see a clearly defined plot structure over his presentation like a well-made movie. This plot structure would be helpful for audiences to find some interesting story and enjoy a given situation.
5. FUNNY and interesting: I know….. this is a very important aspect for socializing and extending our networks. Even though I attend a presentation given by a GREAT researcher (or very big name in our field), the talk itself is sometimes boring and very abstract. However, frankly speaking, I even felt one hour was not really enough slot for his talk. Definitely, I need to expand my capability set by considering this skill (well.. if we can say this is kind of socializing skill set).
In short, I think he successfully described how to harness the power of millions of humans and computers together to solve problems that would be impossible (in some sense) for either to solve alone. Prolly, I need to think about multiagent issues to incorporate human intelligence into a Human-Agent-Robot-Teamwork framework by somehow NATURALLY attracting human to make them coordinate with other agents effectively.