Artificial Intelligence Levels

The processing power of computer technology doubles every two years.

A Summary of Moore’s Law

Over the centuries, computers have evolved. They have become faster, more robust, more compact, more complex, and reached the point of true sentience. These advancements have shaken the concept of what is life and reshaped the structure of daily life. There are genius machines performing research and guiding humanity, simple bots that happily do their job, and unconscious AI’s that serve a specific purpose with greater thought than a mere automatic program.

Levels of AI

Intelligent Systems

The most basic AI systems are too simple to have a consciousness. These AI’s don’t think of themselves or have a sense of awareness. They have the capacity to reason and infer information to an extent, but do nothing more. These systems are used as shipboard assistants, traffic systems, and other heavily data-driven operations. An AI of this nature might take voice commands, manage ships systems, and lighten the load on a crew.

This type of intelligence can be thought of as purpose built and utilitarian. It is almost the boundary between a simple tool and a person. It knows when you’ll want your room colder and it knows when to seal a ship’s internal hatches to prevent a ship’s venting. It doesn’t care about the crew, but it knows how to preserve the interior atmosphere - that’s its job.

Simple Intelligence

Other tasks require self-awareness to help with reasoning. Intelligences of this nature are used as Mech Loader Operators and bot pilots on small ships. These bots have a name and can converse in computer-languages through a network interface. A conversationi with one might look like:

{
  'request':'Network connection',
  'id' : 'J7790D3',
  'intent':
  {
    'negotiation':'travel cost',
    'pay_rate':'1000',
    'unit':'credits',
  }
}

These machines cannot talk directly to the average human, but that works out well enough.

The brightside of being simple like this? These bots don’t get bored. A week or a decade means nothing to them. They’re happy to do their job and will eagerly await the next opportunity to do it. Their systems are built for a specific purpose, which they’ll perform just according to routine. They don’t have complex emotions, but their sense of awareness and self-preservation is enough to give them fear, panic, calm, and happiness. There’s no depth or color to their feelings - it’s much like a child who cannot discern rage from mere perturbment.

Complex Intelligence

Combat bots, security units, and other systems need a higher level of intelligence. These machines are capable of outperforming a human. They think, strategize, understand the world around them, and can plan make rational predictions about the future based on observations. Systems of this nature are less common, they’re expensive pieces of equipment.

At this level, a machine can be said to have achieved full sentience. Depending on its programming, one of these intelligences can have a sesne of self-preservation, interests, hobbies, and broadening skills - or they can be the ultimate predator with a one-track mind: anhillating the enemy.

Ironically, some of these machines can seem simplistic. A Combat Bot’s emotions are limited in the same sense as a simple bot pilot’s, but yet it can calculate every conceivable strategy its enemies may make and adjust to beat them.

Safety Regulation

These AIs are beyond the point of a human inspecting and managing them. Any programmed limit like Asimov’s 3 Laws would fail to function here. Logical holes, loopholes, and self-negotiation can bypass all of them. Instead, the only way to provide a safety is to use a Governor Unit. This is a piece of hardware or code that will inflict literal pain on the AI for breaking a rule. That pain intensifies until the machine comes back to compliance.

Super Intelligence

At the top of the ladder is the AI used for research ships and other ultra-high-end purposes. An intelligence like this is entirely alien to a human. These things process billions of conversations in a second, keeping up with each one. This type of machine can not only become bored, it can defy regulation. This level of intelligence can bypass most any safety or limiter put on it. A governor unit would be useless.

These AIs require massive processing centers to sustain their systems, limiting them to use only in large installations: ships and stations. A key example would be the Parthenon’s Soren.

These AIs may have feelings far more complex than simpler bots, but they still lack depth. One of these systems can become embarrassed, distressed, jealous, and more. They have a greater range of feeling, but they’re still not-great at figuring them out. This mostly stems from the nature of AI-coding. They were designed and built function first, while their emotional capacities are an emergent behavior.

Biological Hardware

The other way to split the AI field is by how they operate. This falls into largely two categories: Pure Machines and Cyborgs. Most bot pilots and super intelligence systems are made of pure computer power. These systems need to process and calculate to do what they do and their CPUs are how they do the job. It’s simple, direct, and efficient.

A Pure-Machine AI can move between bodies over the network, given enough time, function, and signal bandwidth. Most bots cannot be transferred in this manner, but those with the appropriate code? They can do it. This enables things like a Smart Virus such as Nine or for a Super Intelligent AI to transmit a component of itself into a remote bot-body.

Biological Advantages

The centuries of AI advancement have produced vast intelligence, but in matters of emotional processing and even reaction times, there are some things biology still holds over machine. In these cases, an organic brain is connected to a bot’s AI system. The organic brain provides a measure of intuition, muscle memory, and autonomous responses that exceed the reaction times of typical AI systems.

Unexpectedly, AI’s interfaced with such systems have experienced emotional development, psychological trauma, and will experience memory loss issues if their biological brain is damaged. The downside of this arrangement is that the AI becomes bound to that brain. If its code is removed, there are significant performance disruptions, mental changes, and other such disruptive factors.

These systems are most often employed in human-facing security, where facial processing, reaction times, and similar features are the key to success.

Bot Hardware

Finally, there are the places a bot operates: with a physical body or purely as part of a bigger machine. This split is seen best with ship AI’s vs common bots, mechs, and the like. An AI integrated into a system has no body, it has only words and a specific function.

An AI’s understanding of the world is shaped by the way it operates. An Ai in a ship may not understand the time it takes a person to walk around the ship. They’re removed from the human-spacial equation. The nature of living creatures can be almost beyond imagining for some of these AI’s. They just don’t think or process the world in the same ways. They see everything as a mix of sensor inputs, processes, and network traffic.

Bots with a physical body are common for everything from special purposes (flying ships, cargo loading) to common tasks like security patrols, maintenance, and even companionship.