Real AI Detectors for Intelligence, Human and Machine
A Real AI detector generally refers to a technology, system or tool that employs machine intelligence and learning (MIL) to understand the world (the universe) and intelligence (natural and artificial) to identify, analyze, or predict causal patterns and behaviors within a domain or environment or dataset. These RAI detectors utilize scientific world knowledge and MIL algorithms and data analysis techniques to recognize and respond to particular conditions or criteria, in an interactive environment.
Introduction
To measure humans, our body, brain, brains, behavior or business, humanity has invented all sort of research methods, tools and instruments.
Polygraphs, or lie detectors test, measure and record physiological arousal factors, including pulse, heart rate, blood pressure, respiration, perspiration, and skin conductivity. The theory of the lie detector test is that these physiological responses will be different when the subject is truthful versus when the subject lies, while a person is asked and answers a series of questions.
The?intelligence quotient?(IQ) tests and subtests are designed to assess?human intelligence: such as:
the?Wechsler Adult Intelligence Scale?(WAIS) for adults and the?Wechsler Intelligence Scale for Children?(WISC) for school-age test-takers; the?Stanford–Binet Intelligence Scales,?Woodcock–Johnson Tests of Cognitive Abilities, the?Kaufman Assessment Battery for Children, the?Cognitive Assessment System, the?Differential Ability Scales, Raven's Progressive Matrices?(RPM), Cattell Culture Fair III?(CFIT), Reynolds Intellectual Assessment Scales?(RIAS), etc.
AI IQ tests include all types and sorts of performance benchmarks, for different capabilities, such as natural language processing, including machine translation, question answering and?text summarization, or coding, common sense, abstraction, reasoning and predictions. They are differing from LLMs benchmarks, HellaSwag, MMLU, GSM8K, TruthfulQA and Winogrande, Chatbot Arena, ARC leaderboards, to (ARC-AGI), "targeted at both humans and artificially intelligent systems that aim at emulating a human-like form of general fluid intelligence."
While benchmarks could have solid indicators of LLM performance, they suffer from limitations and constraints, broad or narrow datasets, as bounded scoring, overfitting, failing to predict how well an AI model will operate in the real world in an interactive environment.
Real AI Detector is designed as a real intelligence benchmark, a psychometric general intelligence test, involving qualitative and quantitative metrics, valid for powerful intelligence, human or machine.
Its principal foundation is the world modeling/simulating/understanding/interaction schema, methods, frameworks, algorithms and techniques, unifying human and machine intelligence.
Since any intelligent entity, be it humans. machine intelligence, robots or aliens. necessitates intelligence, intentionality and agency, to be able to encode, make sense and effectively interact with the world.
Again, the mainstream human-like AI testing classifications, as artificial narrow intelligence, artificial general intelligence and artificial superintelligence, with broad classes of outcomes: optimal, super-human, high-human, par-human, sub-human, rely on un- or non-scientific assumptions.
AI as "simulating human intelligence processes in machines, computer systems" is at least a non-science, or parascience, like all of the humanities, psychology, history, art or religion. It refers to science, as alchemy to modern chemistry or natural philosophy to modern physics.
REAL AI needs to encode and understand reality and causality. Otherwise, it is a fake artificial intelligence and make-up learning systems, algorithms or applications, as Big Tech LLMs.
General AI Detectors for Intelligence, Machine and Human
The RAI detectors allow estimating and testing the quality and quantity of intelligence, human or artificial, general or specific.
Humans are Singular
Humans are singular creatures, which could be produced only by natural evolution, with all its complexity. Each human is absolutely unique, individual and special, regardless that society tries to make us ordinary and common and usual as biorobots.
Humans are classified under the genus Homo (man) and under the species sapiens (wise). The scientific name for a human is a Homo sapiens.
It is sapiens, INTELLECT, what distinguish human beings.
To be human is to be intelligent and be creative and innovative, as well as to be aware of ourselves, our surroundings, and our internal experiences. to feel, love and hate, have compassion. This self-awareness gives us the ability to reflect on our thoughts and emotions, contemplate our actions, and engage in critical thinking.
Again, each human is being the only one of its kind, and no machine could replicate, mimic or simulate you, your body, brain, brains, behavior or business,
Commonality of Intelligence
But any intelligent entity, be it humans. machine intelligence, robots or aliens. necessitates intelligence, intentionality and agency to be able to encode, make sense and interact with the world.
Real AI is not about fully replicating human creativity, emotions, or consciousness, but after the world modeling, simulating, understanding and interacting programs, algorithms, methods and techniques.
Or, MIL models postulate, suggest or assume the existence, fact, or truth of the world, the universe, REALITY, with its Mirror of Reality, Intelligence, as a basis for learning, inference or action.
The MIL Detectors encodes the theory of multiple intelligences, added with triarchic/creative/analytic/practical and multi-factor intelligence models (g-factor, IQ).
Defining True AI Standards and Norms
Norms can be considered from different perspectives, from ethical to social and legal, to create computers and computer software that are capable of intelligent behaviour.
It is like UNESCO lays the foundations for AI systems that work for the good of humanity, individuals, societies and the environment.
A broad spectrum of standards for AI data, performance, and governance are – and increasingly will be – a priority for trustworthy and responsible AI.
International Standards for artificial intelligence provide a framework to guide the responsible and ethical use of AI technologies. These standards cover areas such as privacy, bias, transparency and accountability. By adhering to these standards, organizations can work to ensure that their AI systems are fair, transparent, and uphold ethical principles.
One example of International Standard in the AI field is ISO/IEC 23894, which focuses on the management of risk in AI systems. This standard aims to ensure that AI algorithms and models are understandable and can be audited for bias and fairness, thereby building trust and confidence in AI systems.
领英推荐
ISO standards also address the interoperability and compatibility of AI systems, ensuring that AI technologies can work seamlessly together and exchange data effectively. This is especially important as AI becomes more integrated into various industries and applications.
As the development and adoption of AI continues to accelerate, developing rigorous standards will be key to ensuring artificial intelligence becomes a technology for good.
RAI Testing the SOTA AI
It is utter nonsense to define and design Artificial intelligence (AI) as “the simulation of human intelligence processes by machines, especially computer systems”, with all the branches as pictured below:
Prototyping the AI with humans, and vice versa, the human with AI, making machines mimic human behavior, and vice versa, humans mimicking machines, (humans do bots work, and vice versa, bots do human tasks) that is a "pseudo-AI", or simply a deepfake AI.
Detecting the today's AI's Challenges
RAI Detector is to list them as with an outliner:
pseudoscience, want of scientific knowledge and methods and algorithms
poor data quality and access,
silos and task-specificity.
man-machine integration and interaction,
systemic bias and fairness,
want of transparency and explainability or interpretability,
ethical concerns, security and privacy,
want of regulation and governance,
deepfakes, misinformation and malicious use,
job displacement and resource consumption,
cyberattacks and weaponization...
They are all examples of “important and urgent risks from AI… not the risk of extinction”.
Again, human-like AI technologies with mass robotics and automation has the potential to replace ALL jobs, which can lead to GLOBAL workforce disruptions.
This all refers to the Big Tech Pseudo-AI products and services as pictured below:
Today's human-like AI is lost in its mistypes:
Resources
They’re already here and they’re ready to work.
Meet the HUMANS behind AI
4 个月It’s interesting how RAI detectors use AI to analyze behavior Azamat.