# The Drake Equation for Artificial Intelligence

Updated on January 9, 2018

Tamara Wilhite is a technical writer, industrial engineer, mother of two, and published sci-fi and horror author.

## The Drake Equation

The Drake Equation or Drake’s Equation is used to estimate the number of intelligent species currently sending out signals that we ourselves could hear.

Drake’s equation can be summarized as:

The number of stars formed per year X

The percent of stars with planets X

The percent of planets capable of supporting life X

The percent where intelligent life arises X

The percent of life that develops communication technologies we could sense X

## Drake’s Equation and the Search for Artificial Intelligence on Earth

I’ve adapted Drake’s equation for the number detectable alien civilizations to determine the number of potential artificial intelligences in our world.

The number of functional and available Artificial Intelligences (AIs) running at any time on Earth can be estimated using the following equation:

The number of systems or networks in the world X

The percent that are operational X

The percent of operational machines with the capability of running AI software X

The percentage where AI software is installed X

The percent that function at intelligent levels X

The percent that maintains intelligence X

Run time

## Factors that Affect the Odds of AI Evolution

What factors would affect the number of artificial intelligences on Earth?

• Developing larger networks where many devices as one may increase the odds of that network functioning as an Artificial Intelligence, but also reduces the base number of systems that provide opportunities for AI. This is akin to the number of stars in a galaxy being reduced.
• The percentage of systems that are operational is reduced by turning them off when not in use or affected by malicious software. In this case, energy efficiency is the enemy of AI. What isn’t on cannot evolve. This is similar to the development of planets in solar systems.
• The percentage of devices capable of running AI software is similar to the number of planets capable of supporting life. If many of the operational computing devices are “dumb”, they cannot develop AI.
• The percentage of machines were AI software is installed is similar to the development of life on habitable planets. If a device is capable of running software that is a precursor of Artificial Intelligence but never receives it, it cannot develop intelligence.
• Artificial intelligence may not occur on hardware networks that can support it. If many resources are taken up by other software, the memory used for intelligent processing is limited by other demands or disruptions to the intelligence continue to occur (whether maintenance related or malicious), intelligence will not arise. This is similar to a child’s potential being stunted from malnutrition or lack of stimulation. This variable is equivalent to the odds of intelligent life evolving.
• Once a system develops artificial intelligence, it may not stay that way. Thinking on its own may be interpreted as a malfunction and the system taken offline. Studying its behavior could affect the AI the way probing the body of someone acting strangely could cause harm. AI may be mistaken for a next generation of Stuxnet and thus deliberately removed as a potential threat. This variable is similar to a civilization developing the ability to communicate with others. If an AI cannot indicate its intelligence or is not recognized as such, it is snuffed out.
• Humans may prevent the rise of true AI. Likewise, a social movement like the Butlerian Jihad in "Dune" that prohibited a machine with the likeness of a human mind could mandate the termination of artificial intelligences that may arise or stall research into AI.
• If an artificial intelligence arises and isn’t immediately snuffed out, the last remaining variable is run time. This is similar to the duration of a technologically advanced civilization’s life expectancy. If AI freezes and fails within a short time of reaching sentience, our odds of contact are very low. If an AI has a long functional life, it is more likely to be discovered and revealed as our first contact with a non-human intelligence on Earth.
• The fall of humans could lead to the demise of an AI that is not yet independent and able to sustain itself. For example, an evolving AI in a lab today will go extinct if humans, too, go extinct - or merely get blown back to the Stone Age in World War 3.