VH vs AI vs Robots

In the second chapter of the book we look at how virtual humans relate to Artificial Intelligence and Robots, and present a Computing/AI landscape on which to situate our analysis. Chapter 2 also covers Virtual Humans and Science Fiction. We'll use this page to share the landscape model and to refine it and to track the evolving relationships between virtual humans, AI and robots.

The AI Landscape Model

This is the model we've been using for several years now in order to help explain to people what we mean by virtual humans, and how the "traditional" or science fiction use of the term "AI" is now massively at odds with the marketing use of the term. The horizontal axis is about how sophisticated a system is, and the vertical axis about how well it presents as human.

We can then plot almost any system, AI or not, on the chart, along with all Science Fiction representations of AI. Some examples are shown below.

Virtual humans could be taken as anything which is at least about 25% up the vertical axis of the chart, with virtual humanoids far to the left, and virtual sapiens at the extreme right.

What is notable is that all current systems are effectively boxed in to the bottom left quadrant - and even machine learning type AI is only slowly crawling along the bottom axis. The reason for this, we argue, is that there are three major barriers to moving across the chart, each probably at least an order of magnitude harder to cross than the previous one. The barriers are:

  • #1 - Making a virtual human actually seem truly human (in any or all presentational aspects)
  • #2 - Achieving Artificial General Intelligence
  • #3 - Achieving Artificial Sentience

Much of the book is about how people are working to cross the first barrier, and in later chapters we explore the efforts (and implications) towards crossing the second, and the likelihood and implications of the third.

Robots and Androids

One thing to note is that the book is not that concerned with robots, robotics, or androids. The reasons for that are twofold:

  • In creating a virtual human we think that the physical world just gets in the way. A proper social virtual world (such as perhaps Second Life) should provide a rich enough environment for a virtual human to develop within and gain a sense of embodiment without having to worry about all the mechatronics.

  • Once we have well developed virtual humans it should be expected that they would be able to remotely control a robot/android body, or even place a part of themselves in it, whenever the need requires, and then re-aggregate the new memories at a later date. Why on earth would a VH trust its precious self to a single copy in a very fragile single physical robot or android body?

Supporting Material for the Book

Here are some resources to help those reading the book, providing images and quotes we couldn't fit in, relevant links and key references.