What is Real Artificial General Intelligence (RAGI), and who prevails the AGI arms race?
To be, or not to be... AGI, or not AGI: that is a question... Either human-compete humanoid fake AGI robots or human-complete, non-human real AGI robots...
Here is a simple universal criterion (iron law) of intelligence: "something/somebody is rational or intelligent, a human, animal or machine, IFF IT KNOWS WHAT IT DOES.
Otherwise, it is simply advanced software/hardware tools, and dumb as a rock.
"Artificial General Intelligence Is Already Here"
The message of the article, Artificial General Intelligence Is Already Here, is
“The most important parts of AGI have already been achieved by the current generation of advanced AI large language models”, in topics, tasks, modalities, languages, and instructability, or "in-context learning".
Our message is to correct the wishful thinking:
"if generative AI and LLMs are to be transformed as true AGI, then a comprehensive rational world model is what that information tools or robotic machines or software agents need for real intelligence and true understanding, as of their identities and actions".
Or, bringing together computer programming and world modeling/reality simulating, not simply logical symbolic systems, mathematics, statistics, linguistics, training data, or mental models, is the only rational way to produce real AI or AGI:
Real AGI (RAGI) = True AI = the World Modeling and Reality Simulating [Causality] Engines of Intelligence + Software/Hardware Information Tools [GenAI/LLMs (OpenAI's?GPT?series of models,?GPT-3.5?and?GPT-4, ChatGPT?and?Microsoft Copilot,?Google's?PaLM?and?Gemini,,?xAI's?Grok,?Meta's?LLaMA?family of open-source models,?Anthropic's?Claude?models,?Mistral AI's?open source?models, and?Databricks' open source?DBRX...}]
Today, to lead in AI R & D and Technology is to lead in AGI R & D & T
The New Age of AGI
In the age of AI, robotics and smart automation, the ultimate test of human rationality and collective intelligence if we are capable to build true/real/rational artificial general intelligence (RAGI), and if we know when it's attained.
Here is a standard piece on AGI from "Tech Companies Want to Build Artificial General Intelligence".
In the middle of the 20th century, computer scientists dreamed of building a computer as smart as humans. Now, tech companies are in a race to build it.
Called artificial general intelligence, or AGI, the systems would do much more than current AI technology. AGI would be as good as humans in many areas of human thinking. These include planning, problem solving, and learning from experience.
Google, Meta, Microsoft, Amazon and Chat-GPT maker OpenAI are all working to develop AGI.
At the same time, governments and leading AI scientists worry about ways the technology could be dangerous for humanity".
AGI, or not AGI: that is a question?..
Today's advanced generative AI systems, from GPT-4 to Meta Llama 3, are ranging between being quasi-intelligent and pseudo-intellectual:
a quasi-intelligent person who wants to be thought of as having a lot of intelligence and knowledge but who is not really intelligent or knowledgeable
or
pseudo-intelligence, which is someone who acts like they know a lot when actually they know very little or nothing at all, just hallucinating.
Therefore they are softly classified as weak AI or narrow AI which are limited to computing specifications, algorithms, and specific tasks they are designed for. Even emerging generative AI applications with better rote memory retention are weak AI because they cannot be repurposed for other domains and require substantial training before they can handle related tasks even within the same domain. For example, you must fine-tune a pre-trained large language model (LLM) with financial/medical/Q/A datasets before it can operate consistently as a financial/medical/Q/A chatbot.?
The follow-up of generative AI, presented by large language models (LLMs), as Microsoft/OpenAI's GPT-4 or Google's Gemini, is artificial general intelligence (AGI,) or a digital superintelligence (G. Hinton), full general intelligence (Meta), or a human-like and human-level AI or strong AI
Such a general-purpose, predictive and generative AI could be of two completely different types, Irrational AGI or Rational AGI:
a computer hardware/software which REASON AND ACT HUMANLY, with human-like intelligence and the ability to self-learn, human comprehension and cognitive capabilities, as creativity, perception, learning, and memory, all to mimics the human brain/mind or behavior, with all inherent irrationalities.
a computer hardware/software which REASON AND ACT RATIONALLY, with machine-like intelligence and the ability to self-learn, world comprehension and cognitive capabilities, as creativity, perception, learning, and memory, complementing the human brain/mind or behavior.
Thus, AGI is either Irrational Human-Compete Systems or Rational Human-Complete Machines:
aligned either with the human brain/mind/intelligence, promoted as general-purpose, generative AI or a human-like and human level machine intelligence and learning
or
aligned with reality and factuality, truth and scientific knowledge, known as Real AI: Real Machine Intelligence and Learning or Trans-AI.
AI/AGI Definitions
For thousands of years, rational humanity has tried to understand how the brain/mind works, or how we perceive, learn, understand, think, predict, and manipulate a world.
As a result, today we can’t find a common definition of human intelligence, not mentioning machine intelligence, going as AI or AGI or ASI.
But according to Russel and Norvig (2016), AI can be categorized into four groups:
1. Thinking Humanly
“The exciting new effort to make computers think . . . machines with minds, in the full and literal sense.” (Haugeland, 1985)
It uses cognitive modelling. We need to get inside the actual workings of human minds. There are three ways to do this: through introspection—trying to catch our own thoughts as they go by; through psychological experiments—observing a person in action; and through brain imaging—observing the brain in action.
2. Thinking Rationally
“The study of mental faculties through the use of computational models.” (Charniak and McDermott, 1985)
Thinks with Logics: Aristotles syllogisms provided patterns for argument structures that always yielded correct conclusions when given correct premise?
3. Acting Humanly
“The art of creating machines that perform functions that require intelligence when performed by people.” (Kurzweil, 1990)
Example of this definition is Turing test: During the test, one of the humans functions as the questioner, while the second human and the computer function as respondents.?
4. Acting Rationally
“Computational Intelligence is the study of the design of intelligent agents.” (Poole et al., 1998)
Example of this definition is a Rational Agents: They operate autonomously, perceive their environment, persist over a prolonged time period, adapt to change, and create and pursue goals.
Extending the rational AI classification to AGI, it could be upgraded into broad classes of irrational/Human AI/AGI and Rational/Real AI/AGI:
Irrational/Human AI/AGI, persecuted by the big tech companies and their startups:
Looking as Humans, from sex robots to humanoid hydraulic robot models
领英推荐
Thinking Humanly, Statistical/Irrational ML/DL/ANNs, Neuro-Symbolic Machines, Humanoid AI-Robots
Acting Humanly, Statistical LLMs, Generative AI
Looking, Perceiving, Thinking, Learning, Acting Humanly, as AGI machines and quasi-intelligent humanoid robots
Rational/Real AI/AGI, persecuted by the Rational AGI/Agents promoters
Looking as Machines, AI processors
Thinking Rationally
Acting Rationally, Causal/Rational/Intelligent Agents/ Machines
Looking, Thinking, Acting Rationally, as Non-Human Machines, Rational AGI Machines, World Widely Deployed and Distributed, Cyber-Physical-Social Hyperintelligence
How Harmful Irrational Minds/AGI, or Why Rational Mind Won the Big Tech AGI Race
Among the AI companies trying to build AGI are the Google's DeepMind, Microsoft's OpenAI, Amazon's Anthropic and Musk's xAI, all of which are backed by billions of dollars.
Many leading figures at these very companies have claimed, in public, while standing in front of microphones, that one possible outcome of the technology they are explicitly trying to build is that everyone on Earth dies.
As a matter of fact, following the Rational AI/AGI Paradigm, we have decided the AGI problem by 2021 while being an active member of "the EU AI alliance, an initiative of the European Commission to establish an open policy dialogue on Artificial Intelligence. Since its launch in 2018, the AI Alliance has engaged around 6000 stakeholders through regular events, public consultations and online forum exchanges".
The basic results with technical specifications, classifications, formulas, algorithms, designs and patterns, have been kept as a proprietary IP, documented as the Corporate Confidential Report:
How to Engineer Real AI [Man-Machine Superintelligence] by 2025.
Now, where to find the RAGI pieces:
5. Trans-AI: How to Build True AI or Real Machine Intelligence and Learning (See the Supplement)
A Moral Lesson
Rationality is what humanity is in urgent need, which overrelies on social media, heuristics and biases, rather than rational principles.
Rational minds could prevail the AI/AGI race due to leveraging the machine's modeling the world (computing ontology), together with philosophy, metaphysics, mathematics, science, data science, computer science, cybernetics, engineering and technology.
Our Rational/True AI/AGI initiative is realize the new techno-scientific ambitions of the EU, to catch up China and the US, to finally lead the future AI/AGI world.
Resource
SUPPLEMENT: Trans-AI: How to Build True AI or Real Machine Intelligence and Learning
Azamat Abdoullaev/EIS Encyclopedic Intelligent Systems Ltd, EU, Cyprus-Russia
Abstract
We are at the edge of colossal changes. This is a critical moment of historical choice and opportunity. It could be the best 5 years ahead of us that we have ever had in human history or one of the worst, because we have all the power, technology and knowledge to create the most fundamental general-purpose technology (GPT), which could completely upend the whole human history. The most important GPTs were fire, the wheel, language, writing, the printing press, the steam engine, electric power, information and telecommunications technology, all to be topped by real artificial intelligence technology.
Our study refers to Why and How the Real Machine Intelligence or True AI or Real Superintelligence (RSI) could be designed and developed, deployed and distributed in the next 5 years. The whole idea of RSI took about three decades in three phases. The first conceptual model of TransAI was published in 1989. It covered all possible physical phenomena, effects and processes. The more extended model of Real AI was developed in 1999.
A complete theory of superintelligence, with its reality model, global knowledge base, NL programing language, and master algorithm, was presented in 2008. The RSI project has been finally completed in 2020, with some key findings and discoveries being published on the EU AI Alliance/Futurium site in 20+ articles.
The RSI features a unifying World Metamodel (Global Ontology), with a General Intelligence Framework (Master Algorithm), Standard Data Type Hierarchy, NL Programming Language, to effectively interact with the world by intelligent processing of its data, from the web data to the real-world data.
The basic results with technical specifications, classifications, formulas, algorithms, designs and patterns, were kept as a trade secret and documented as the Corporate Confidential Report: How to Engineer Man-Machine Superintelligence 2025.
As a member of EU AI Alliance, the author has proposed the Man-Machine RSI Platform as a key part of Transnational EU-Russia Project. To shape a smart and sustainable future, the world should invest into the RSI Science and Technology, for the Trans-AI paradigm is the way to an inclusive, instrumented, interconnected and intelligent world.
Key words: Artificial Intelligence, Machine Intelligence, Machine Learning, Trans-AI. Citation: Azamat Abdoullaev. Trans-AI: How to Build True AI or Real Machine Intelligence and Learning. Ontology of Designing. 2021; 11(4): 402-421. DOI: 10.18287/2223-9537-2021-11-4-402- 421. Introd
Resources
References:
Power BI | Tableau | Python | Data Science | AI | Machine Learner | Marketing
4 个月Exciting times ahead as we explore the potential of rational AI systems in the pursuit of Artificial General Intelligence!
blogger
4 个月When will I get a robot servant?
Sounds like a mind-bending journey ahead Count me in for some rational AGI action. ?? Azamat Abdoullaev
CEO at Cognitive.Ai | Building Next-Generation AI Services | Available for Podcast Interviews | Partnering with Top-Tier Brands to Shape the Future
4 个月It's fascinating how rationality guides digital humanity toward progress and collaboration. Exciting times ahead for those joining this ambitious initiative Azamat Abdoullaev