Fact Or Fiction?
But is this just paranoia? The claims carry more than a passing resemblance to science fiction movies, such as The Matrix and 2001: A Space Odyssey, but are Musk and Bostrom voicing valid causes for concern?
The case that we are not living in a simulation is strongly supported by resource arguments. Consider the sheer computing power needed to run such a simulation. A simulation system would need to manage all the entities in the world and all their interactions. This would require a vast amount of processing. Further support can be found in arguments relating to quantum mechanics – to run a truly lifelike simulation of a city, with all its trillions of interactions, would require a city-sized computer. This makes the case for our existence in a simulation very unlikely.
Even if a machine existed that could simulate our existence then there would be a high probability that we would experience so-called “realism imperfections”. These bugs in the simulation would be seen or heard due to glitches in the model. For example, stars may or may not exist when viewed through telescopes of varying magnification. Such errors would be inevitable in a simulation of this scale, but they have never been observed by humans.
Machines that use self-learning, super intelligent software are still very distant from the current state of the art, and systems that do make use of AI operate in very tightly defined fields. Current systems learn to optimise their performance within specific domains of work – not to take over the world.
For example, neural networks, which are sometimes referred to as electronic models of the brain, are used to predict changes in stock markets. These systems can be trained using existing data from stock trading to learn and identify patterns in live data streams that may indicate that something will happen. The result of this is that traders can take action to mitigate any negative outcomes.
Equally, there are systems that are developed using AI techniques to alleviate workloads by applying programmed rules and facts. These are known as knowledge-based systems. While the human users of these systems may not realise that they are interacting with a machine – such as Jill the online tutor of AI, which answers questions and provides feedback to students on an AI course – they are also developed to work on or within well-defined problems or domains.