5 Fictional Artificial Intelligences We’re Working on Today
Science fiction movies often deal with malevolent AI bent on the destruction of humanity, but for every HAL or Terminator, there is an example of artificial intelligences working alongside, or for, humanity. From android companions who flawlessly mimic humans to semi-sentient security drones, AI is everywhere in science fiction.
So how close are we to creating that technology today? Here’s a list of five technologies we’re actively working towards, and a guess about how far away we might be from wide-spread implementation.
Fiction: The Culture. A series of novels by Iain M. Banks.
Technology Highlight: Sentient highly intelligence spaceships.
Description: The artificial intelligences of the Culture series range from proto-sentient androids who perform menial service tasks to hyper-intelligent “Minds” which are part of ships. Minds are tremendously powerful, capable of running all of the functions of a ship or habitat while holding potentially billions of simultaneous conversations with the citizens that live aboard them. To allow them to perform at such a high degree, they exist partially in hyperspace to get around hindrances to computing power such as the speed of light.
Current Projects: The spaceships of The Culture require two things: first, the creation of an artificial general intelligence (AGI), and second, integration of that intelligence into a ship. The feasibility of AGI is hotly debated, with some scientists claiming we will have achieved true AGI in as little as 25 years, while others say that it may never happen. Integrating basic artificial intelligence into ships, however, is already in the works. DARPA is currently researching a chip which will allow them to put AI on drones that can be sent into war zones, the US Navy is automating ships, sensors, and weapons, and of course cars can drive themselves.
Estimated Timeline: 2040/Never
Fiction: Red Dwarf. A television show about life on a spaceship.
Technology Highlight: Sentient android.
Description: Kryten is a service mechanoid — an artificially intelligent android who must follow his programming, and has a limited understanding of human concepts. He has been shown to have the ability to break his programming, but has significant difficulty doing so. Because Red Dwarf is a comedy, Kryten’s programming may often not make sense. He doesn’t understand lying, for instance, but he has enough emotions to want to date.
Current Projects: When we see fictional characters like Kryten we often assume they must be a fully operational general intelligence, but the truth is that we’ve proven it doesn’t take true intelligence to trick humans into thinking something is sentient; all it really takes is conversation skills. Students studying AI were tricked into thinking their TA was human, when in fact she was a simple chatbot. And multiple times researchers have downloaded conversations, letters, and writing from real people into networks to create androids that closely mirror their human counterparts. Unfortunately they still sound like ‘computers’ when they talk, but the things they’re saying are eerily human. (Listen at 1.57 of that first link if you want goosebumps.)
Estimated Timeline: 5 years.
Fiction: WWW Trilogy. A series of novels by Robert K. Sawyer.
Technology Highlight: Internet Consciousness.
Description: Webmind is a self-aware consciousness that grows and evolves through the Internet.
Current Projects: It might not be a positive example, but everyone is now familiar with Tay, Microsoft’s chatbot that was designed to learn from the input of Twitter users. Tay drew only from Twitter and Webmind drew from the entire internet, but whether that would help or hinder an artificial intelligence remains to be seen — I’m not sure I want to meet the artificial mind raised on memes and 4chan. Sawyer imagines a highly benevolent general intelligence birthed from the total sum of human knowledge, and while we’re still very far away from that, we are working towards it. The vast majority of current projects in AI use machine learning algorithms to teach machines to teach themselves, often training them with the vast quantities of “free” data available on the internet.
Estimated Timeline: 25+ years.
Fiction: Star Trek. A series of television shows.
Technology Highlight: Computers driven by artificial intelligence that can recognize verbal and physical input.
Description: The computers on the various Enterprises (and Deep Space Nine) were powered by artificial intelligence. They had no personalities or names, but were vocalized by a bland, pleasing female speaker (familiar, anyone?) and could respond to simple verbal input (Earl Grey tea — hot).
Current Projects: Siri is an excellent example of the early iterations of this technology. Capable of responding to voice commands (most of the time), Siri can activate many of the commands on your phone — but for some functions, it’s easier to just do it yourself, as evidenced on Star Trek by the fact that most of the stations had human operators. In fact, Siri’s limited ability might be an excellent parallel for the technology level evidenced in Star Trek; but if you’re hoping for more, you’ll have it soon. Siri’s replacement, Viv, has been announced by Apple, with the promise of much more sophisticated vocal recognition software and integration into all the functions of your smartphone, not just search and music. No word on when Vivi will launch, though it could be as early as this winter. And Google is loathe to be left behind; their Google Assistant has a less human touch, but an equally sophisticated voice and search system.
Estimated Timeline: A year.
Fiction: Snowcrash. A novel by Neal Stephenson.
Technology Highlight: Semi-autonomous security robots.
Description: Rat Things are robots with the neural networks of dogs embeded into them — they ‘remember’ their previous lives as dogs, and can communicate with other Rat Things by “barking” in the Metaverse. Although their minds are largely controlled by their implants, they can sometimes act independently of their programming.
Current Projects: There are, thankfully, no current projects to transmit the consciousness of animals into machines. Weirdly, there is a huge amount of research (and some experimentation) into the possibility of uploading human consciousness into machines, but the rest of the animal kingdom is so far being left alone. That said, we are creating semi-autonomous attack drones capable of firing a 80,000 volt taser into a human intruder.
Estimated Timeline: Never/no current research.