1. Are Video Games Artificially Intelligent？
Obviously yes. In a way.
Julia and I play Super Mario Kart on the Wii. This is a racing game where we select a vehicle and race on one of several possible courses against either ghosts, robots, or real people located elsewhere in the world. It occurs to me that we are carrying out a Turing Test every time we do this.
We know when we are racing real people connected to us via the Internet because we select this option, and because of other signals that would be impossible to miss because we know how the game works. Therefore, we are not really carrying out a Turing Test because we know in advance if we are playing Artificially Intelligent Robots built into the game vs. real humans. I imagine we could set up a test for someone who does not know the game, but that would be a lot of work and would require that we let someone else use our stuff and they might break it, so that's not gonna happen. But we can ... and as a matter of survival (in the game) do ... contemplate the difference between a robot and a human.
There are four levels of robots, in terms of their ability to play the game, and both Julia and I can handily beat, 100% of the time, robots of the lowest two levels. Julia can beat on most tracks most robots on level 3 most of the time. I have less success, but I can beat them most of the time. The fourth level robots seem to have about the same capacity to race as the third level entities, but the fourth level tracks are reversed, mirror image versions of the regular tracks. Under these conditions, the robots have no trouble ... I assume they simply reverse some chip in their brains or put their eyes in backwards or something and they can run the tracks as well in mirror mode as normal. Julia, who knows the tracks quite well, has some trouble with a small number of these "mirror mode" challenges because she had internalized the layouts. I have only a moderate amount of added difficulty because I wasn't that good to begin with. The chance of me driving off the track or running into an obstacle are, shall we say, not increased by running the course in reverse.
However, when we play actual humans, it is a very different story. The patterns of ability and strategy we see in our fellow humans is more complicated.
A small number of the humans are incompetent or distracted. We pass them right away and they never catch up. We often imagine them to be the little brothers and sisters of the actual racers. This imagery is enhanced if they have chosen toddlers as avatars. (About a third of the avatars, including the one I usually use, are "baby" versions of the adults. They are lighter weight and drive highly maneuverable vehicles .... avatar choice does affect performance and must be matched with racing style in this game.) They may, of course, be noobs, or they could be people who went to the fridge for a beer during the game.
Another subset of actual humans seem to be super humans. they have capacities that none of the robots have, and that we don't have. They seem to have earned these abilities by being very good racers and meeting certain challenges that we have not met yet. Or, they've hacked the game and are cheating. Regardless, we hate them.
The other racers are generally quite good. Julia can often beat them, I rarely beat them. Actual humans, including Julia and I, have certain courses that we are good at and others that we are not so good at (part of the game's strategy is voting for which course a race will occur on, and then the course that is used is randomly chosen among those selected by the initial vote). So as we play a set of humans on a few different course, the results change.
Simply put, we can tell the robots are robots because they don't make mistakes but also don't seem to take certain chances that humans take (with some success). The lowest level robots do not seem to drive around obstacles that we leave on the track as part of our strategy, and there are two or three trick you can do to speed up or get some other positive result that they don't seem to "know" about. The most advanced robots drive around our obstacles if possible, and they seem to know about the tricks.
The fact that the best robots are not anywhere near as skilled as the best humans is not because of some limitation of Artificial Intelligence or computing power. At least, I'm pretty sure of that. With respect to tactics, knowledge of how to play the game and basic skills, the robots can be perfect, and thus very hard to beat (though there are chance effects that matter a great deal). I'm pretty sure the robots are not programmed to have this level of perfection.
There is one way to play the game which is called "Tournament." Here, one person plays the machine, and that one person races 11 other racers all of whom are robots. One or two of the robots will be better than all the other robots, and if a robot beats you it will be one of them. But which robot this is seems to vary from game to game. This is a human-like pattern programmed into the game.
Another way to play is to race against "ghosts" from the Internet. A "ghost" is a ghostly (semi-transparent) avatar that represents an avatar from a previous race. So an avatar from the Internet is some other person who raced, with no other racers, on a track to see how fast they could go, and you race this ghost to see if you can beat it. SO, this is you against a real person, but the real person ran the race in the past.
What is interesting about this is that that the computer seems to be able to select ghosts that are about as good as you are, but not quite. Julia beats her ghosts almost every time she plays them. I beat my ghosts about 5 out of 6 times. But when I play the game using Julia's identity, the ghost that she would likely beat beats me.
So the ghosts from the Internet on this game are a kind of chimera ... they are real human experiences being selected to run against you by an artificially intelligent being.
So what distinguishes the real humans form the robots in the Wii Mario Kart game? The humans are more variable and more unpredictable. It seems to me that this should be easy to program into the robots. The humans can be imitated, I'm convinced, convincingly, and these game robots could easily pass the Turing test.
It would not be a real Turing test, because the Turing test is to be carried out in the linguistic mode. And we know the linguist mode and the racing car mode are different! But having the occasional artificially intelligent robot join us in our inter-human Internet play would still be cool. Uncanny, yes, but cool.
If such a robot existed I would name it Klatu.
2.Your Future in Cyberspace: Artificially Intelligent Journalism
This is a summary of a paper that explores, in great depth and detail, the role of technologies that use digitally managed personal information to shape the content and direct the delivery of content in a journalistic setting. The most important conclusions of this paper are: New technologies obviate the media but not the journalist or the receiver of the information; and this ... the use of your private electronic information to determine what part of Cyberspace you will experience and how ... is inevitable.
A presumption of this paper is that the technology of using private individual data (your personal "digital identity") for pushing information has certain possibilities, and those possibilities will be realized regardless of what anyone thinks should or should not happen. The analogy of nuclear technologies (peaceful energy vs. bombs) is invoked not just as a metaphor but as a model:
Attempts to defend the Internet against ISPs that profile customers by 'snooping' echoes similar attempts by the scientific pioneers of nuclear energy, following Hiroshima, to control and limit the use of nuclear energy to the peaceful service of humanity.
The issue involves trade-offs between privacy and security, between privacy and convenience, and between privacy and business opportunities. The actors include isolated hackers and criminal networks, respectable companies pursuing market-driven business trends, and NGOs.
The paper asserts that the essence of journalism (as usually defined by journalists or journalist think tanks) is retained even if the media itself is taken out of the picture and the relationship between the producers of "journalism" and the consumers (who are "digital identities" is maintained.
...it can be constructive to look for a new short definition of 'journalism', separating it from 'the media'. Such a definition should connect to the principles of journalism, and be based on the relation between journalism and its audience, rather than on its relation to the medium it uses for communicating with the audience (which is what is causing the confusion today).
Technology including and beyond Web 2.0 and/or the Semantic Web have modified the relationship between journalists and the recipient of the content through myriad technologies including social networking, content based image and video retrieval, image coding (so pictures can be searched), content analysis, and so on. The paper seems to imply that what started out as an effort at automation has given rise to an incipient self organizing creative capacity.
Central to all of this is the concept of "Digital Identity" to which content is pushed using information about that identity. This applies not just to journalism, but to society and government as well.
An additional aspect concerning the digital identity is the rapid transition to an electronic government where communication between citizen and government is electronic, through the Internet. Many countries have passed laws that compel change to a "paperless government" in which all services are provided through the web. In October 1998 the Government Paperwork Elimination Act came into effect in the United States; it induces all the authorities to develop or purchase information technologies as a substitute for paper within five years.
Transition of organizations to Cyberspace and the creation of huge amounts of digital data calls for a shift to new models of decision making in many areas previously governed by the human intelligence knowledge and intuition, for all their advantages and disadvantages.
So the role of AI in journalism will be, according to this paper, as follows:
AI engines will be used by media companies to search customers for content interests, automatically. Dependence on gaining measurable consumer attention can be expected to induce journalists in all media platforms to adjust content to maximize consumer attention and advertising dollars. New business models will be needed to reduce the intrinsic risk to journalistic freedom that the new methods will induce.
What seems to be missing here is that with a full manifestation of "digital identity" and an AI interaface to the population of digital identities, the journalists themselves will be pretty much obviated, as long as the web is free and the OpenSource movement is not squished.
And thus arrives the citizen journalist. I don't think even the bloggers have a chance...
Latar, Noam, & Nordfors, David (2009). Digital Identities and: Journalism Content How Artificial Intelligence and Journalism May Co-Develop and Why Society Should Care
The Innovation Journalism Publication Series, 6 (7)