Jump to content
Forbin

The threat of Big Dog

Recommended Posts

So when Santa Petman comes down all of our chimneys, what should I leave out on the cookie tray for him?

It doesn't matter what you leave out, Petman will take what it wants.

Share this post


Link to post
Share on other sites

TmF1Z2h0eS9uaWNlIGFuYWx5c2lzIGNvbXBsZXRlLiBSZW5kZXJpbmcganVkZ2VtZW50Li4uIA==

Share this post


Link to post
Share on other sites

https://www.quora.com/Are-modern-fighter-jets-too-over-sophisticated-for-warfare

 

http://www.swarm-troopers.com/

Swarm Troopers explains why small military drones will be cheap and plentiful, and how researchers made a military-grade drone for less than $2,000 in a basic workshop using smartphone components and 3D printing. With modern combat aircraft costing upwards of $100,000,000, the military will face a choice between a single manned plane or a swarm of fifty thousand drones. Except that off-the-shelf electronics are getting more powerful and cheaper, so small drones will continue to fall in price while getting ever more capable.

Share this post


Link to post
Share on other sites

I think I totally misunderstood the comparison there between a hundred million dollar manned fighter and a swarm of locust drones. I'm like imagining a big box of bombs being carried by a hundred quadcoptors like the balloons in Up.

Share this post


Link to post
Share on other sites

The logical endpoint to project locust is Stephenson's Diamond Age.

 

Or that Star Trek episode where the computers calculated who died and asked them to report to the termination chambers.

Share this post


Link to post
Share on other sites

Yesterday Google announced that it had achieved a milestone step in AI research.  It has developed the first AI, named AlphaGo, capable of beating an expert level human player at Go beating three-time European champion Fan Hui 5-0 in a closed match.  That may not sound like much, but Go is many, MANY times more complex that Chess.  As an example, after the first round of turns in Chess there are 400 possible board positions.  In Go, after the first round of turns there are 129,960 possible board positions and it grows at an insane rate from there.  There are more total Go positions than there are atoms in the universe.  In March the AI will take on the world champion Lee Sedol in Seoul.

Share this post


Link to post
Share on other sites

Yesterday Google announced that it had achieved a milestone step in AI research.  It has developed the first AI, named AlphaGo, capable of beating an expert level human player at Go beating three-time European champion Fan Hui 5-0 in a closed match.  That may not sound like much, but Go is many, MANY times more complex that Chess.  As an example, after the first round of turns in Chess there are 400 possible board positions.  In Go, after the first round of turns there are 129,960 possible board positions and it grows at an insane rate from there.  There are more total Go positions than there are atoms in the universe.  In March the AI will take on the world champion Lee Sedol in Seoul.

I just posted this in the strategy games board, but its fitting here too.

Games like checkers, chess, go etc are pretty solvable for AI. It just requires a lot of brute force type programming. I will get a lot more worried when AI can beat champions at games that have a lot of chance involved. Being able to build a winning strategy out of a bad set of rolls or draws is much more impressive/terrifying. Daft Souls latest episode had a good discussion about how AI in 4X don't hold grudges or really think long term. You can declare war on them over and over, betray them, steal from them, but if an alliance looks good to them in the short term they will make it again.

In other news I am listening to Nick Bostrom's Superintelligence book. I am pretty unimpressed so far, he makes a lot of leaps and assumptions that I really don't think are valid. The paperclip robot example bugs me to no end. I just do see how an AI who is trying to produce 1 million paper clips just starts destroying the earth/solar system/galaxy to achieve its goal. This is a huge leap, you need an AI dumb enough not to know its going too far, but smart enough to trick all of humanity until it's too late.

Share this post


Link to post
Share on other sites

Yesterday Google announced that it had achieved a milestone step in AI research.  It has developed the first AI, named AlphaGo, capable of beating an expert level human player at Go beating three-time European champion Fan Hui 5-0 in a closed match.  That may not sound like much, but Go is many, MANY times more complex that Chess.  As an example, after the first round of turns in Chess there are 400 possible board positions.  In Go, after the first round of turns there are 129,960 possible board positions and it grows at an insane rate from there.  There are more total Go positions than there are atoms in the universe.  In March the AI will take on the world champion Lee Sedol in Seoul.

 

out of everything this is probably the closest thing to worry about...

 

@Cordeos, what...?  Go is anything but simple for traditional AI to manage.  AI in 4x games are just scripts, not remotely in the same league as AlphaGo which is far more general purpose than a set of codes written specifically for a specific game.

Share this post


Link to post
Share on other sites

out of everything this is probably the closest thing to worry about...

 

@Cordeos, what...?  Go is anything but simple for traditional AI to manage.  AI in 4x games are just scripts, not remotely in the same league as AlphaGo which is far more general purpose than a set of codes written specifically for a specific game.

Go is a two player piece placement game. This is less complex than a more than two player game with multiple victory paths and combat including random dice rolls allowing for things to just go horribly wrong due to bad luck.

I know they are scripts right now, what I am saying is I would be far more impressed if they could make a 4X AI that plays like a human using tools similar to the GoAI.

Share this post


Link to post
Share on other sites

If they use tools like AlphaGo for 4X, it'll straight up play better than any human players and hence become 'inhuman'.  Also random dice roll if anything would make it easier for the AI since it now doesn't have to discover the probabilities as it had to in Go... but riadsala already explained this and I think you are ignoring how AlphaGo is different from previous chessbots that brute forced via search treeing history of chess.

Share this post


Link to post
Share on other sites

AI for 4X games is designed to be fun to play against. I have a hard time believing that if a team with the goal of making an unbeatable 4X AI couldn't do so as easily as chess, and the main reason no one has done so is a lack of incentive and a lack of budget. I also have a hard time believing that a computer would be somehow worse at dealing with randomness than a human would.

Share this post


Link to post
Share on other sites

The best part of that clip is the guy walking past who pauses mid-stride like "whoa a robot playing golf, I'm gonna check this out" then by the end of the clip he's just as pumped as everyone else.

Share this post


Link to post
Share on other sites

Nature fights back! Dutch police are considering using eagles to take down drones:

 

http://www.wired.com/2016/02/so-dutch-cops-are-teaching-majestic-eagles-to-hunt-drones/

 

Of course the net result is merely that eagles will be the second evil species mentioned in the datapackets future robot educators will create about the terrible, drastic and necessary actions robots had to take to save the planet from its evil slaving overlords and their animal quislings.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×