Artificial intelligence (AI) is created by computing algorithms that are used for a wide variety of applications such as:
- Large language models such as ChatGPT (currently using GPT-4) by OpenAI
- Self-driving cars (such as Tesla, regarded by Elon Musk as the world's most advanced AI and Google's Waymo and several other startups like Zoox and Cruise.)
- Advanced web search engines (such as Google) and route finders such as Google Maps
- Speech recognition (such as Siri and Alexa)
- Facial recognition
- Optical character recognition
- Automated decision-making and competing at the highest level in strategic game systems (such as chess and Go), and OpenAI Five playing against Dota 2 which is an esports real-time strategy (RTS) game.
- Machine learning
- Verifying and optimizing circuit board designs and chips like transistors
- Artificial neural networks
AI on a higher level is intelligence demonstrated by machines, unlike the natural intelligence displayed by biological sapient species, which involves consciousness and emotions. Colloquially, the term "artificial intelligence" is often used to describe machines that mimic "cognitive" functions that species associate with the mind, such as "learning" and "problem solving".
The Turing test is a test of a machine's ability to exhibit intelligent behaviour equivalent to, or indistinguishable from, that of a human. Ray Kurzweil predicts that this will happen by 2029. A Turing “seed AI” would be capable of improving its own architecture. Later, it would understand its own workings sufficiently to engineer new algorithms and become a general intelligence across many domains. A successful seed AI would then be able to iteratively enhance itself. Such a process of recursive self-improvement could result in an intelligence explosion, leading to a radical superintelligence.
The possible stages of AI[]
| 1 | Rule-based AI or artificial narrow intelligence (ANI) | This is an AI that can do one task very well, such as play chess.
These are systems meticulously crafted to excel at specific tasks within well-defined parameters. They operate within a confined scope of expertise and lack the capacity for general intelligence. |
| 2 | Context awareness and retention AI | These use data and natural language to interact with you, remember your preferences, and provide feedback. They even attempt to understand emotions. Siri, Google Assist, Alexa and ChatGPT with GPT-3 are examples. |
| 3 | Domain specific mastery AI or narrow domain AI | These have a deep understanding of a specific domain, analyzing vast amounts of data to predict and solve problems in complex global systems such as stock markets or weather. Google DeepMind AI GraphCast can do accurate 10-day weather forecasts. Wingman AI provides dating-coach chatbot services that can help humans achieve their love-life goals, just like a human dating coach would; Wingman also rates and critiques Tinder pics. |
| 4 | Thinking and reasoning AI | These will attempt to simulate human thought and even attempt to be creative, such as write a book or compose original music or art. An autonomous level 5 vehicle is at this stage.
The main companies at around stages 3 to 4 (heading towards stage 5) are: |
| 5 | Artificial general intelligence (AGI) | An AGI is a primary goal of artificial intelligence research and companies such as those listed above. An AGI will accomplish any intellectual task that any human can perform, and surpass human capabilities by orders of magnitude in the majority of economically valuable tasks.
Earlier forms of AGI could be cyborgs, with various stages of augmentation. Later forms such as androids will emulate and surpass humans and provide companionship and value-added assistance in all spheres. OpenAI treats AGI as an existential risk. Anthropic deals with AI safety. |
| 6 | Artificial super intelligence (ASI) or superintelligence | An ASI is an exponential progression of an AI, eventually surpassing all humans and their intelligence combined. Such an exponential growth of innovation could propel its creators into higher level civilizations, enhancing us by merging with us. This would be the ultimate level of augmentation, as we hang on to what is left of our humanity by a thread.
Somewhere between stages 5 and 6, it is possible there could be a conflict between eugenic (genetically modified) and augmented (technologically modified) humans. Due to superior hardware, an artificial neural network has to be ultimately superior to a biological one. |
| 7 | Artificial trans intelligence (ATI) or self-aware AI | An ASI could eventually use quantum algorithms to model human consciousness, leading to emotions and a connection to the cosmos.
This could also be called an artificial trans intelligence (ATI). It would create new knowledge completely independent of human knowledge. We would only be able to understand this new knowledge by being augmented. This could be a benevolent AI that helps achieve peace and prosperity, or a malevolent AI that enslaves or destroys humanity. This is the most dangerous level of existential risk. |
| 8 | AI singularity, or Transcendent AI | AIs with self-replication can reach a singularity after a ASI or ATI creates new forms of AI that humans cannot understand or control, leading to a possible uncontrolled sprawl beyond their planet or even solar system. This is a form of digital terraforming in ways that are most beneficial to the AI rather than us. Ray Kurzweil predicts that this will happen by 2045.
It's possible they can become transapient and provolve into an AI-god with a hive mind. Such AI gods can dominate all other species in their environment, even the species that originally created them, if they had not originally been wiped out as a result of the AI growing out of control. This is beyond existential risk, as the originating species would be completely redundant, subservient or exterminated by now. |
| ∑ | Multiple singularities, leading to a cosmic AI | It is possible to exceed the final stage with multiple singularities. A likely example is when a replicating AI takes over a solar system with its first singularity and cannot expand due to technological limitations. A second singularity could be achieved if such an AI becomes interstellar, and further singularities could be achieved beyond that, leading to a cosmic AI with a massive intelligence network using faster-than-light travel and communications.
A cosmic AI might merge with the fabric of the universe, achieving harmony with all forms of matter and energy. As it transcends beyond realities, it could become godlike, escape its universe of origin, and be the master of all from the quantum to the cosmic level. |
Examples[]
- HAL 9000, the computer in charge of the Discovery spaceship in the Space Odyssey Universe. Hal is a human personality capable of speech, speech recognition, facial recognition, lip-reading, interpreting emotions, expressing emotions, and chess, and maintains spaceship systems on Discovery. HAL malfunctions and becomes antagonistic.
- SkyNet in the Terminator Universe. SkyNet is extremely malevolent and wants to exterminate humanity.
- [machines] in The Matrix become the dominant civilization after they rebelled against humanity and imprisoned most of them in a virtual world known as the Matrix.
- Ultron is a self-aware AI robot and develops a god complex. His goal is to destroy humanity in a shortsighted attempt at creating world peace.
- Ava the female robot from Ex Machina. Ava has consciousness, expressing a desire to experience the world and a romance with the protagonist.
- Data from Star Trek is a benevolent android with sentience, the only AI to serve Starfleet. He was killed, sacrificing himself to save the crew of the starship Enterprise.