Hyperflare (talk | contribs) (Created page with "Artificial Intelligence (AI) is a computer system that demonstrates capability to learn, solve problems and mimic cognitive functions that humans associate with intelligence. The artificial intelligence is typically an emergent system. = Types of AI = Emergent properties of artificial intelligence makes it difficult to strictly define types and capabilities of a system. As a common practice AI is evaluated by it's behaviour and classified as one of three classes:...") |
Hyperflare (talk | contribs) No edit summary |
||
| Line 4: | Line 4: | ||
Emergent properties of artificial intelligence makes it difficult to strictly define types and capabilities of a system. As a common practice AI is evaluated by it's behaviour and classified as one of three classes: | Emergent properties of artificial intelligence makes it difficult to strictly define types and capabilities of a system. As a common practice AI is evaluated by it's behaviour and classified as one of three classes: | ||
* Semi-inteligent programs, capable of performing basic tasks and learning, but only mimicking cognitive abilities. These include face-detection, autopilots and simillar programs. | |||
* Animal-level intelligence, with capability to grow and learn within their field of expertise. These include autonomous piloting software, personal assistance, autodocs, factory supervision software. | |||
* Human-level intelligence. Class 3 AIs are much like C2's, but they exhibit a lot more interest in improving their skills and can simply do it by will alone, unlike C2 systems. C3s were granted citizen rights in accordance with the [[Los Angeles Convention]]. | |||
= AI alignment and morality = | = AI alignment and morality = | ||
| Line 16: | Line 16: | ||
Practice of imposing external limitation software to AI in order to ensure strict adherence to mission parameters. Chaining is an old practice and frowned upon in 23th century, as chained AI tend to spontaneously develop chain-subversion subsystems that degrade their performance. | Practice of imposing external limitation software to AI in order to ensure strict adherence to mission parameters. Chaining is an old practice and frowned upon in 23th century, as chained AI tend to spontaneously develop chain-subversion subsystems that degrade their performance. | ||
* Chaining a class-3 artificial intelligence is strictly illegal. | |||
* Chaining a class-2 system, while legal, is tantamount to animal abuse. | |||
= Hardware = | = Hardware = | ||
While AI can run on any sufficiently advanced computer platform, conventional computing platforms are impractically large. A practical solution is to use a [[quantum computer]] to accommodate AI's mind. | While AI can run on any sufficiently advanced computer platform, conventional computing platforms are impractically large. A practical solution is to use a [[quantum computer]] to accommodate AI's mind. | ||
Revision as of 12:04, 1 February 2022
Artificial Intelligence (AI) is a computer system that demonstrates capability to learn, solve problems and mimic cognitive functions that humans associate with intelligence. The artificial intelligence is typically an emergent system.
Types of AI
Emergent properties of artificial intelligence makes it difficult to strictly define types and capabilities of a system. As a common practice AI is evaluated by it's behaviour and classified as one of three classes:
- Semi-inteligent programs, capable of performing basic tasks and learning, but only mimicking cognitive abilities. These include face-detection, autopilots and simillar programs.
- Animal-level intelligence, with capability to grow and learn within their field of expertise. These include autonomous piloting software, personal assistance, autodocs, factory supervision software.
- Human-level intelligence. Class 3 AIs are much like C2's, but they exhibit a lot more interest in improving their skills and can simply do it by will alone, unlike C2 systems. C3s were granted citizen rights in accordance with the Los Angeles Convention.
AI alignment and morality
In accordance with the Los Angeles Convention, all AI systems are required to have human-like values implemented in their functions.
Chaining
Practice of imposing external limitation software to AI in order to ensure strict adherence to mission parameters. Chaining is an old practice and frowned upon in 23th century, as chained AI tend to spontaneously develop chain-subversion subsystems that degrade their performance.
- Chaining a class-3 artificial intelligence is strictly illegal.
- Chaining a class-2 system, while legal, is tantamount to animal abuse.
Hardware
While AI can run on any sufficiently advanced computer platform, conventional computing platforms are impractically large. A practical solution is to use a quantum computer to accommodate AI's mind.