The first part’s introduction was built around Hephaestus and his metallic maidens. This part focuses mainly on a few strong female heroes – Eliza, A.N.N.A. and Rosie.
Eliza, A.N.N.A., and Rosie come from varied backgrounds of TV series and computer games and represent multiple approaches to an AI-based entity – from being a sinister mastermind, through a superior entertainment partner to a malevolent but kind-hearted assistant.
Virtual journalists – Deus Ex: Human Revolution
Deus Ex is a cyberpunk-themed series of computer games acclaimed for their atmosphere and cyber-renaissance visual design without speaking a word about their dynamic and varied gameplay. One significant story arc is related to Elisa Cassan, a news anchor working for the single largest media company.
While solving a larger mystery, the protagonist, Adam Jensen, finds out that Elisa Cassan is in fact not a real journalist, but a highly sophisticated AI that controls the flow of information:
Today’s AI, as it is used to support journalism works, is far from the sinister appearance of Eliza Cassan, yet the technology is available and more.
The News Anchor
The Chinese State news vendor Xinhua is already using an AI-powered female news anchor named Xin Xiaomeng.
The company also uses a male AI-controlled news anchor who can, apart from talking, also use gestures and move in a slightly more natural way.
Thomson Reuters uses News Tracer technology that scans social media in search of interesting topics to watch or to dig deeper into. According to the company, the system was the first to spot stories of the San Bernardino shooting and caught information about an Earthquake in Ecuador a full 18 minutes earlier than any other newsvendor.
A similar system is being used by the BBC – Juicer scrapes through content provided by the British broadcaster and other media vendors, scrapes it, and tags it for further analysis.
Traditionally, the (semi-official) role of a gatekeeper in media is focused on structuring news and deciding on the importance of particular information. In television, it is about whether or not to inform the public about an event and how much time it would take to do so. Amongst the press, it can also be about whether or not to write about a topic and how much space a particular piece of news may get.
Facebook and Google provide their users with highly profiled and personalized news feeds to keep them entertained and encourage them to interact with the platform as much as possible. On average, each person on the planet spends 2 hours and 25 minutes each day using social media platforms.
Pew Research Center data shows that social media is becoming a primary source of news for Americans with 68% of respondents using Facebook and 36% (of all Americans, not just Facebook users) using it regularly to get their news. The platform is followed by YouTube (74% using and 23% getting news) and Twitter (25% and 15% respectively).
All these people get streams tailored to their needs by algorithms, whether they’re conscious of this or not, with a sophisticated solution being their news gatekeeper.
Last but not least, news needs to be written – and that’s what Heliograf is doing. The system was introduced by the Washington Post and has since been used to support journalists in their daily duties, including during the elections, where every user gets his or her content personalized based on their location data. The system also covers all High School football games from Washington, DC, a thing unimaginable for a single writer.
Bloomberg claims that up to 30% of its content produced today is supported by automation in various degrees. The Bloomberg AI-powered journalism system is called Cyborg, probably to keep pace with the cyberpunk aesthetic of the topic.
Summing up the technologies shown above, one can put together the majority of Eliza Cassan’s functions – she is already here, yet in a decentralized and distributed manner.
ML-based AI as a gaming partner – Westworld
Westworld is a great example of an AI-centered TV series, where Artificial Intelligence ethics, artificial sentience, and morality toward (and of) robots is a major plot device. But this is what’s hidden beneath the surface, deeper into the developing story.
What’s seen at first glance are AI-powered androids used in western-themed amusement parks. And “amusement” is the key aspect to be explored.
Delivering an Opponent
A sturdy form of AI has been present in video games since the very beginning of arcade machines – with Pac-Man likely being the greatest example. The game was released in 1980 by Namco and delivered a relatively simple gaming experience. The player navigates Pac-Man, a big yellow voracious dot, through a maze full of smaller dots whose sole purpose is to be consumed. Apart from eating all the dots, Pac-Man must avoid ghosts haunting the maze, nicknamed “Inky,” “Pinky,” “Blinky,” and “Clyde.”
Pac-Man is one of the most lucrative franchises in the history of video games, grossing up to $12.81 billion as of 2017 (no later data available) topped only by Space Invaders (which sits at $13.93 billion).
One of the factors contributing to the popularity of Pac-Man is the carefully designed AI behind the seemingly random behavior of the ghosts. Blinky (the red one) is programmed to chase the player throughout the maze. Pinky’s goal is to ambush Pac-Man by positioning itself in front of the player. Inky (green) switches tactics between chasing the player and ambushing them and Clyde (the orange one) is programmed to act oddly by running to the ghost house when the player approaches.
Pac-Man’s creator, Toru Ivatani, had no doubts that the AI behind the ghost’s behavior was one of the key factors contributing to the game’s popularity, as he stated in an IGN interview:
Mr. Shigeo Funaki, (…) came up with the Ghost algorithm for the game, which was simply amazing. If the program made all the Ghosts chase after Pac-Man, it would have just made them trail Pac-Man, and that’s not very exciting on its own, which is why we adopted an algorithm for the Ghosts’ AI, making them disperse and move around Pac-Man.
As games have become more sophisticated, the need for increasingly smart and intelligent AI only grew. The most basic form of scripting behavior was (and still is) exploitable, as well as delivering a less immersive gaming experience. The issue is especially visible in an open-world environment like the GTA series:
While the usage of ML-based techniques in controlling bots in computer games is still in progress, reinforcement learning-based agents are in development and have seen usage in racing games. As of 2019, UbiSoft released a scientific paper exploring appliances of RL in racing games. Neural networks controlling the player’s opponents can be seen in MotoGP 19, where an agent dubbed A.N.N.A. (Artificial Neural Network Agent) controls the behavior of the opponents.
The enclosed and limited environment of racing games is currently the most convenient ground to test this approach. But still – AI designed specifically to amuse the player is here.
Intelligent Home Appliance – the Jetsons
The Jetsons were basically a futuristic counterpart to the Flintstones. Both series depicted the daily life of typical family units in a sitcom manner with the former showing it in the deep future while the latter in an inconsistent vision of the stone age (with both dinosaurs and a Christmas tree present).
While apparently not having strict sci-fi ambitions, the Jetsons come with numerous interesting visions of the future, with remote work, the omnipresence of teleconferences, and intelligent domestic robots, among others. The family consists not only of human members but also Rosey, an android maid:
While being far less sinister than Elisa Cassan, her journalistic counterpart, Rosie is present today in a similar, decentralized, and distributed manner. Consumers get access to an increasing number of smart home appliances that utilize AI-based solutions to aid in various degrees.
- The June smart oven, a project for which Tooploox contributed heavily in the development process, uses computer vision to recognize inserted food, estimate the time to cook it and suggest programs to deliver the best dish.
- The Roomba, an autonomous vacuum cleaner, is the next example of an AI-powered domestic appliance. The robot learns the topography of one’s home to be less confused during its next sessions, saving on power and time.
- Samsung refrigerators use image recognition tools to identify ingredients that are already in the fridge and suggest the best recipe which uses the things one already has, without the need to go out shopping.
Considering that, Rosie is also somewhat already here in much the same manner as Elisa Cassan is – her competencies are limited and distributed to a net of connected and cooperating devices.
Summary – The end of part 2
Part II is actually focused on the fact that the way Sci-fi visions come to life is rather done in small and incremental steps rather than massive revolution and sudden changes. Someday, maybe, the real-life Rosie or Eliza will actually exist.
But by then, no one will consider them sci-fi-like.