Lit Review Pow!

Its not exactly a talk, but I've done a ~20 page literature review. I'll post it up here soon, or talk at it. I kept imagining that I was explaining it out loud to someone and using that to work out how to construct it. I think its the first piece of writing I've enjoyed doing since early high school.

I've also been skiing (yay), written the specs for 2 assignments and by this time next week I'll have gone on 8 flights in 3 months. I'm quite frustrated at all of this. Before I did my lit review I had this beautiful period where I was actually getting research done.

Distractions

Its amazing how much time simple distractions can take up.

On wednesdays I theoretically take 3 hours out to teach (11-2). In reality, I wake up, have a quick breakfast, prepare to teach (read over notes, work out which questions my students are likely to ask me, etc). I don't get any work done before I teach -- and once I've taught its 2 and I haven't had lunch. By the time I've eaten and made it back to my desk I'm tired and its 3:30 in the afternoon.

I haven't really been getting anything done on wednesdays or fridays. Thursday I'm learning.

Then on monday and tuesdays I've been finding other interesting projects to do. Its amazing what seems interesting when your main project is really hard :)

My current side projects:

• Code editor project, where Joseph is learning how to write desktop apps in JavaScript. (Javascript is way faster than python and ruby!)
• iProgress; a little iphone game
• Operating system project. When I finish uni, I think I'm going to try and write my own OS. Planning starts now.

Next session is time to knuckle down. I have been making progress the last month or so but its not impressive progress. Its background reading and learning kind of progress. I've also started explaining my algorithm in code; but thats a story for another day ;)

For now, here's a procedurally generated map I wrote for a little toy game I'm working on for the iphone:

Episode 4: Architecture

What are biases? Why are they important?

Episode 3: Biases

What are biases? Why are they important?

Sellout!

I decided to show a little restraint about what I post here.

Its not a decision I made lightly; but basically every researcher I talked to said that it really is a concern and it really does happen. Meh.

The effect for all of you is:

• I'll stop stuffing around and get back to posting videos
• As we get more technical, there's some implementation details I'll leave out. I'll record everything, just hold video.
• Once I publish my findings in journals, I'll post the previously redacted material online.

I wish there was a good solution for online publication where I could include these videos as research or something.

On sharing

I had a big 'discussion' with my supervisor yesterday about sharing my ideas online. He's worried that other researchers (perhaps subconsciously) will take my ideas and incorporate them into their own work. If they do that and publish before I do, it will be very difficult for me to publish and get my PhD.

The whole idea of hording ideas seems so wrong to me... I'm never going to run out of good ideas. But maybe he's right about the academic stuff. And it would kinda suck if someone published my research.

Sam's take on it is "If they really do beat you to it, they deserve it." He's probably right... but... Urgh.

My options:

• Keep doing my video blog anyway
  
• Only talk about what I'm reading about and background stuff. Self-censor the meaty details
• Hold some / all posts until I've published the work (or at least, some work so others can cite me)
  
• Abandon the video blog entirely

I'm nervous about all of those options. I see a great amount of value in the video blog for myself and hopefully other people too. It would be a real shame to remove that out of fear.

Also, apparently even if other academics wanted to cite the work I'm doing, they'll have a hard time with it until I publish my work in a journal. Whats with academia completely failing to discover the humble URL? Cool URLs don't change and the URL was invented exactly for the purpose of interrelated papers.

Sigh.

Episode 2: People

How do people factor in to my intelligence framework? What is our reward function

Episode 1: Intelligence and Rewards

... in which I talk about what intelligence is and what rewards are. What problem am I trying to solve?

First post!

Hello there!

I'm Joseph. I just started a PhD in Artificial Intelligence. I want to make thinking machines - I think that would be _TOTALLY_ rockin'.

- Yes; I've watched battlestar and 'AI' and iRobot. But, AI doesn't really work like that. It needs a reward function which we give it. Imagine having robots make all the food, and build all our houses and clean up after us. We could do so much good with strong artificial intelligences. It'll completely change how our society works, but it would do so for the better.

Anyway, thats the goal; but I'm a long way from it.

I'm going to have a video blog where I explain all the AI bits as I work them out. First episode uploading now!