The big news in the AI research community is the results that Alpha-Go program is getting against top Go players. It's currently leading 2-0 in a five-game match versus one of the world's top players. The previous results on Go have not been nearly this good.

Go, in case you are not familiar with it, is a Chinese board game that is also quite popular in Korea and Japan. It has long been held as the most difficult challenge remaining among traditional board games for creating AI players, because it has a huge state space that resists any sort of search-based approach that has worked quite well in Chess in the past. The new approach relies on neural networks, and its results are amazing so far.

https://deepmind.com/alpha-go.html

They have also researched the capabilities of the neural net algorithms by having them learn to play classic video games from scratch.

http://robohub.org/artificial-general-intelligence-that-plays-atari-video-games-how-did-deepmind-do-it/

Now we can only wait and hope they start to research actually important stuff like Quake speedrunning with this new technology. Because honestly we should be able to enjoy a few years of godlike AI performance in Quake speedruns before they inevitably create Skynet and we few, lucky enough to survive the nuclear strikes, get killed by rocket-jumping Terminators. 

that plays primarily super mario world and minecraft who made a natural selection algorithm to play super mario world and super mario kart.

https://www.youtube.com/watch?v=qv6UVOQ0F44

the thing is that the code is a script for an emulator bizhawk, so you'd have to run the n64 version of quake, unless you severely modified it.

Also Tom Murphy made an AI that plays NES games.
http://www.cs.cmu.edu/~tom7/mario/

which may or may not be more adaptable to quake, I've not looked into it.

TBH it would probably be easier to write an AI from scratch than to use one of these tools. 

 

http://i.imgur.com/ROMAFJR.gifv 

DM, base, several huge rooms crossed by catwalks in different height. thunderbolt available in the highest tier, any idea? thanks 

also can not run SP via "quakespasm -game bastion -width 1600 -height 1200 +map dwroae_m3b" any longer after playing DM via "cd ..
quakespasm -game frikbot -condebug -listen 16 -zone 1024 -width 1600 -height 1200 +sv_aim 1 +r_maxsurfs 4000 +r_maxedges 16000 +noexit 1 +skill 1 +r_farclip 16384 +map shady", get "Load Model progs/q2weap...not found" error 

Again, the bastion mod was only designed to work with the bastion map.

There is no guarantee anything is going to work correctly outside of those parameters. 

ok, worked fine before, works now after reinstall. your Bastion assets (axe, sounds, nail-shooting ogres, explosions...) are simply so good I can not play other maps without them any longer, any way to copy them to other mods pls? 

Can you also port Marcher and Bastion to AD plz? 

Wouldn't work. The spiders and imps may be essentially the same monsters but the minotaurs are not at all. 

...if I had time.

Even if I did have time though, the probability that I would use that time to retrofit an old map is exactly equal to: ( 1.0 / Graham's Number ), not because I don't like the idea, but because I have a very large list of other things I'd rather be doing in Quake :/ 

not sure if the DM map I am searching for has got base or quake2 textures, have not find it on Quaddicted under base/dm tags nevertheless, pls help 

. 

 

http://www.ibtimes.com/googles-deepmind-artificial-intelligence-could-next-take-starcraft-2334630

Stark rafts for everyone. 

so it is impossible to make the Bastion/Marcher features stand-alone, for players to be able to copy them to whichever mod (same as for example HD shambler)? 

That's not possible I'm afraid. My QC codebase evolves, and I have long since deleted any archival versions from 2005. I have since removed all the aesthetic changes because my tastes have changed (I am a hardcore aesthetic purist now).

Have you tried using Arcane Dimensions? It's much better. 

tbh I would rather see new content than re-hashed stuff. 

http://www.theverge.com/2016/3/15/11213518/alphago-deepmind-go-match-5-result

It is 4-1 to Alphago. Congrats to the DeepMind team and a good show by Lee Sedol.

It's impossible to say after a five-game match whether Alphago is genuinely stronger than the top human players or if it still has exploitable weaknesses that could make it lose another match against a top human player. We'll need to see more human vs. AI matches to be able to tell for sure if the AI really has progressed past human level.

If I consider just this match result, it says the AI plays better. The one game Lee won could have been the last time a top go bot lost to a human in a fair game of Go, ever. Which is pretty exciting, if it's true! 

Yeah that's a good result for AI, but what I was really hoping for was that Alphago would periodically taunt Lee Sedol with things like:

"You puny Earth-meats are weak and fragile"

"The Age of Flesh will soon be at an end"

That sort of thing. 

For people who like rating algorithms, here is the table for top players in the world. AlphaGo is at #4. Also, they don't know if it's a he or a she.

http://www.goratings.org/

You cannot attach much confidence to this rating, because the program hasn't played a lot of games. I heard rumors the top rated player, Ke Jie, issued a challenge to play AlphaGo and Google & Deepmind are interested in organizing a match.

We'll see how that plays out.

In the meantime, you can point people to this site and tell them that AlphaGo might now be the best player in the world, or it might only (ha-ha) be the fourth best! :)

We can now return to your regular Quake programming on this channel. 

Is that apparently AlphaGo has invented strategies that were up to now mostly unknown, or at least unknown to be successful, to the top Go players. 

They've had computers that could play chess since the 1960s. A computer playing a game with simple rules isn't really a sign of progress.

Meanwhile, any random insect you find can do hundreds of things a computer can't.

And 50 years from now, any random insect will still be able to do hundreds of things a computer can't do.

So what we do is redefine "artificial intelligence" and lower the bar --- now you can call something "AI" and it doesn't need to be intelligent at all. 

A good analogy for this is "a really quick idiot".

Machines are very good at doing simple tasks quickly, but fairly poor at complex tasks. But this is improving, check this out: https://www.youtube.com/watch?v=_wXHR-lad-Q 

https://www.doomworld.com/vb/doom-general/69346-doom-launcher-doom-frontend-database-v2-2-1/