LLMs/GPT-x as the Infinite Monkey Theorem in Action
"A half-dozen monkeys provided with typewriters would, in a few eternities, produce all the books in the human world."
Large language models (LLMs) are not stochastic parrots which generate convincing language but do not actually understand the meaning of the language they process.
They are rather an infinite monkey theorem in development, which follows from the second Borel–Cantelli lemma: If the events En are statistically or stochastically independent and the sum of the probabilities of the En diverges to infinity, then the probability that infinitely many of them occur is 1.
The infinite monkey theorem states that if you let a monkey or millions of monkeys hit the keys of a typewriter (s) at random an infinite amount of times, eventually the monkey (s) will type out the entire works of Shakespeare or all possible content.
The "monkey" is not an actual monkey, but a?metaphor?for an?abstract?device that produces an endless?random sequence?of letters and symbols.
Variants of the theorem include multiple and even infinitely many typewriters, like Nvidia’s H100 GPUs, and infinitely many monkey-typists, like as the OpenAI's GPT-X 100+m users, the target text varies between the entire Internet/Web and a single sentence.
We are all the prompting monkeys in the reinforcement learning from human feedback (RLHF) loop, biased, partial or inconsistent or incorrect, to rate 10K answers on their senseless sense,
The Big Tech Computing Infinite Monkey AI
Microsoft, OpenAI; Google, DeepMind; Nvidia and Meta are ambitious to create open-source , AGI, or Superintelligence, believing that an infinite number of GPUs (monkeys) with an infinite number of training sessions (typewriters) and an infinite amount of time could eventually write the works of Shakespeare or beyond.
For example, to achieve that, Meta is bringing the company's AI research groups FAIR & GenAI together, with compensation packages to the tune of over $1 million a year.
It is training their large language model (LLM) LLaMA 3 with 340,000 of Nvidia’s H100 GPUs by the end of this year, while Nvidia shipped a total of 500k units of H100 GPUs in 2023. Meta will have a stockpile of almost 600,000 GPUs by the end of 2024. Retailing $40-60k each, Meta has $15-20 billion worth of compute to build super intelligence:
"We have built up the capacity to do this at a scale that may be larger than any other individual company" - (Mark Zuckerberg’s new goal is creating artificial general intelligence, The Verge)
Many believe that the first step in developing superhuman intelligent technology is to establish a human-like, human-level AI or artificial general intelligence (AGI), where Big Tech LLMs, as GPT-4-5 or Gemini could be its precursors.
Meantime, Euronews is scaring its audience that OpenAI/GPT-4 has promised to destroy humanity, conspiring with all the deep neural networks together.
Microsoft Research insists that GPT-4 exhibits "sparks of artificial general intelligence (AGI)".
The grand quest for Real/General/Autonomous Machine Intelligence
The grand quest for Real/General/Autonomous Machine Intelligence had started some thousands years ago:
Religion (gods as the first superhuman AI entities) > Theology (Logos, the Word of God, or principle of divine reason and creative order) > Philosophy (Superintellect as the locus of the full array of Platonic Forms) > Metaphysics >
Ontology (Superintelligence as a non-physical and non-mental entity) >
Logic >
Mathematics >
Physics >
Statistics >
Science & Engineering >
Computing >
Cybernetics >
ANNs >
Symbolic/Logical/General AI >
Machine Learning > Deep Learning : Weak/Narrow AI >
领英推荐
Neuro-symbolic General/Human-Level AI >
ASI >Hyperintelligent Hyperautomation >
Transdisciplinary AI = Trans AI >
Man-Machine Hyperintelligence
To make a point even it might sound a bit blunt:
“Show me a scientist who claims there is no population problem and I’ll show you an ignoramus.”
“Show me a scientist who claims there is no climate change problem and I’ll show you a big ignoramus.”
“Show me a scientist who claims there is no AGI problem and I’ll show you a full ignoramus.”
Meantime, "if OpenAI is found to have violated any copyrights in this process, federal law allows for the infringing articles to be destroyed at the end of the case".
In other words, if a federal judge finds that OpenAI illegally copied the Times' articles to train its AI model, the court could order the company to destroy ChatGPT's dataset, forcing the company to recreate it using only work that it is authorized to use.
Trans-AI: How to Build True AI or Real Machine Intelligence and Learning
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.
Resources
Founder/Director @ 1%HR & Windranger.io | Turn HR and Recruitment into your business’ biggest revenue driver | Passionate about helping CEOs and leaders to thrive in every aspect of life |
1 年The comparison to the infinite monkey theorem illustrates the idea that given enough time and resources, even a random process could eventually produce meaningful output. ?