Interactive AI = Real AGI = AI + GenAI/LLM + IAI World Model
AI, GenAI, LLMs, AGI and ASI bound to humans' capabilities are learning or inspiring subjects for computer scientists and programmers, statisticians and neuroscientists, scientists and technologists, businessmen and laymen, philosophers and politicians, mass media and sci-fi community.?
We shall seek AI/AGI/ASI in the very sense of "intelligent" and "general", not in the sense of "being like a human", as human-like, superhuman AI systems, with comprehensive, human-like understanding.?
And this necessarily implies having a general theory of "intelligence" in terms of reality and interaction, its causal/mathematical/scientific modeling and world knowledge integration.
The world models and causal patterns and scientific laws and rules are the essence of intelligence, natural and artificial. ?? ?????????? ?????????? ???? ???????????????? ???? ????????????????????, ?????????????? ???? ?????????????????????? ????? ?????????? ???? ?????????? ?????? ???? ????????????, ?????? ???????????? ?????? ????????????????????????, ???????????????????? ?????? ??????????????????, ???????????? ?????? ???????????????s, ???? ?????? ?????? ????????????, scopes ?????? ????????????.
What is a Human-Like AI/ML/DL?
?AI system: An AI system is a machine-based system that, for explicit or implicit objectives, infers, from the input it receives, how to generate outputs such as predictions, content, recommendations, or decisions that can influence physical or virtual environments. Different AI systems vary in their levels of autonomy and adaptiveness after deployment.
?AI system lifecycle: AI system lifecycle phases involve: i) ‘design, data and models’; which is a context-dependent sequence encompassing planning and design, data collection and processing, as well as model building; ii) ‘verification and validation’; iii) ‘deployment’; and iv) ‘operation and monitoring’. These phases often take place in an iterative manner and are not necessarily sequential. The decision to retire an AI system from operation may occur at any point during the operation and monitoring phase.
?AI knowledge: AI knowledge refers to the skills and resources, such as data, code, algorithms, models, research, know-how, training programmes, governance, processes and best practices, required to understand and participate in the AI system lifecycle.
?AI actors: AI actors are those who play an active role in the AI system lifecycle, including organisations and individuals that deploy or operate AI.
?Stakeholders: Stakeholders encompass all organisations and individuals involved in, or affected by, AI systems, directly or indirectly. AI actors are a subset of stakeholders.
The Generative AI Building Blocks
Generative AI (genAI) can produce new text, images, video, code, art, or audio clips learning patterns from?training data?and generates outputs with the same statistical properties.
Its building blocks might involve foundational models, compute, frameworks, compute, orchestration & vector databases, fine-tuning, labeling, synthetic data, AI observability and model safety, as figured below.
How Does Generative AI Work?
Generative AI models use?neural networks?to learn patterns in data and generate new content. Once trained, the neural network can generate content similar to the data it was trained on. For example, a neural network trained on a dataset of text can be used to generate new text, and depending on the model’s input, the text output can take the form of a poem, a story,?a complex mathematical calculation, or even?programming code for software applications.
The usefulness of genAI outputs depends heavily on the quality and comprehensiveness of the training data, the model’s architecture, the processes used to train the model, and the??prompts human users give the model.
Data quality?is essential because that’s what genAI models use to learn how to generate high-quality outputs. The more diverse and comprehensive the training data, the more patterns and nuances the model will potentially be able to understand and replicate. When a model is trained on inconsistent,?biased, or noisy data, it’s likely to produce outputs that mirror these flaws.?
Training methodologies and evaluation strategies are also crucial. During training, the model uses feedback to adjust values within the model’s architecture (internal parameters).
The complexity of the model’s architecture can also play a significant role in output usefulness because the model’s architecture determines how the genAI processes and learns from training data.
In essence, the building blocks of AI/GenAI are as the following:
GenAI Models = Training Data + Word Embeddings + Vector Data Base + Statistical Algorithms
To become general and interactive, responsible and trustworthy, or real intelligent AI, its major building blocks should be upgraded as below:
IAI Model = World Knowledge Model + World Embeddings + Vector Knowledge Bases + Causal Algorithms (Fundamental Axioms, Principles, Scientific Laws)
?World Embeddings are the world's data/information/knowledge representation that encapsulates semiotic information crucial for AI to gain real intelligence and learning and understanding to effectively and sustainably interact with the world/environment/context/ physical or virtual.
What goes as AI, including GenAI, has narrow and weak or task-specialized algorithmic intelligence – where a particular system addresses a particular problem. Unlike human intelligence, such narrow AI intelligence is effective?only?in the area in which it has been trained: self-driving, fraud detection, facial recognition or social recommendations. To date, HAGI does not exist.
The key challenge for creating a general AI is to adequately model, map, encode or embed the world with all the entirety of data/information/knowledge, in a consistent and comprehensive and useful manner.
Real/True AI: World Embeddings vs. Word Embeddings
All Machine Learning/AI/DL/Large Language Models work with numerical data, vector spaces, of extremely large training data, a training set, training dataset or learning set, the material used to train algorithms or teach prediction models to predict some specific outcome, as classifying some content or recognizing features, shapes, objects or subjects such as people or animals.
The idea of training data in machine learning programs is foundational to how the computer learns to process information. It mimics the abilities of the human brain to take in diverse inputs/sensory data and analyze/weigh them, in order to produce activations in the brain, in the individual neurons. Artificial neurons replicate these neural processes with machine learning and neural network programs provide detailed models of how human thought processes could work.
Again, before any operation, all text/image/audio/video data has to be digitally encoded or embedded, transformed into numerical representations. Embeddings are data that has been transformed into n-dimensional matrices for deep learning NN computations. Embeddings represent real-world objects, like words, images, or videos to be digitally processed.
Embeddings are foundational for AI, being vectors or arrays of numbers that represent the meaning and the context of tokens processed and generated by the model. Embeddings enable the AI model to handle multimodal tasks, such as image and code generation, by converting different types of data into a common representation.
Embeddings are an essential component of the transformer architecture that GPT-based foundation LLM models use, and they can vary in size and dimension depending on the model and the task. It is used to encode and decode the input and output data, for text classification, summarization, translation, generation, image generation or code generation.
Embeddings created can generalize to other tasks and domains through transfer learning — the ability to switch contexts — the reason of its popularity across AL/ML applications developers [See Popular Embedding Models]
As vector representations of data, embeddings capture meaningful, semantic or syntactic, relationships between entities, which could be thoughts or concepts, words or things, with all possible interactions, interdependencies and connections.
But today's AI embeddings rely on Rote Memory and Pattern Matching, having No Analogy, No Deduction, but simple Mathematical Induction or Statistical Inference from statistical patterns and correlations.
Trans-AI is based on the World Embeddings, the representations or encodings of the world or reality and its content and domains, where its
Entity Embeddings (EEs) are vector representations of categorical or ordinal or interval or ratio or cardinal variables or entities in a dataset. The EEs are learned by training a neural network to capture the relationships between different entities in a high-dimensional space of the encoded world or its domains, systems or processes.
Entity embeddings convert Entity Data into continuous numerical data, enabling the use of AI/ML algorithms that require numerical inputs.
Trans-AI Embeddings are covering the word embeddings, the representations or encodings of tokens, such as sentences, paragraphs, or documents, in a high-dimensional vector space, where each dimension corresponds to a learned feature or attribute of the language, as in LLMs.
The World Embeddings enables the real, true or INTERACTIVE AI/ML/LLM models to effective, efficiently and sustainably interact with the world, understanding semantic and syntactic relationships between the tokens to generate intelligible content.
WHAT IS IAI, OR INTERACTIVE MACHINE INTELLIGENCE AND LEARNING?
Interactive/Interaction/Interacting AI (IAI) as the next big thing of Real/True/Scientific AI is embracing the masses. As to DeepMind’s cofounder: Generative AI is just a phase. What’s next is interactive AI. “This is a profound moment in the history of technology.”
Modern humanity has gotten itself in the data-computing vicious circle: more data demanding more computing producing more data requesting more computing, thus more computing power is needed to power AI.
Next generation AI could soon look planetary-scale computing systems to further fuel AI's computational requirements.
To extrapolate, it could finally look as a cosmic-scale computing system to fully fuel greedy AI’s computational requirements, and all the universe is to become a giant computer simulation.
Here is how to destroy the data-AI vicious circle: Interactive Intelligence and Learning, trademarked and marketed as
IAI (Interactive Artificial Intelligence): Interactive Machine Intelligence and Learning: Interactive DL, Interactive ANNs, Interactive NLU/NLP, Interactive LLMs, or Interacting Generative AI.
Interactive/Interaction/Interacting AI (IAI) is innovated as a machine/robotic/computing/automated/electronic/artificial/technological [hyper]intelligence with the power of autonomous interactions with the world, as in man-machine communication or human-computer interaction, with interactive hardware and software systems (as interactive user interfaces).
The IAI could be implemented as Automated Hyper-Intelligence (AHI) or Hyper-Intelligent AI, embracing all the meaningful approaches, technologies, techniques, methods, models and algorithms of reactive AI:
IAI = Trans-AI = Meta-AI = Hyper-Intelligent AI =
Computerized Hyperintelligence =
Data Computing +
Reactive AI [ANNs + Symbolic AI + Statistical AI + Machine Learning + Deep Learning + NLP + LLMs + Generative AI + Explainable AI + Causal AI + Narrow AI + Internet of Things + Robotics]
General AI +
Super AI +
Quantum AI +...
The IAI could be interpreting as reifying the Aristotelean notion of the active intellect (intellectus agens; agent intellect, active intelligence, active reason, or productive intellect) acting upon the passive intellect to make potential knowledge, as LLMs, into actual knowledge, acting as the Agent Intellect (IAI) providing a unified knowledge of the universe.
Conclusion
The GenAI's companies, products and services are overpriced and overhyped, becoming the most over-hyped technology news of our post-truth, post-modern digital age.
Its fundamental concepts, of "training data" and "word embedding", are to be overruled with the "world knowledge" and "world embeddings", the fundamentals of Interactive AI (IAI) or Agent Intellect, the summit of AI development history:
Cybernetic Systems >
Artificial Neural Networks >
Computing Software/Hardware >
Symbolic/Logical AI >
ML > DL > ANI >
GenAI > LLMs >
AGI > ASI >
Active Intelligence = Trans-AI = Meta-AI.
领英推荐
Resources
SUPPLEMENT 1
Multimodal GenAI models operating with different text, image, audio, and video data prompts and generating different data types:
Popular Real-world Uses for Generative AI
When generative AI is used as a productivity tool, it can be categorized as a type of?augmented artificial intelligence.?
Popular real-world uses for this type of augmented intelligence include:?
Image Generation:?Quickly generate and/or manipulate a series of images?to explore new creative possibilities.?
Text Generation:?Generate news articles?and other types of text formats in different writing styles.?
Data Augmentation:?Generate synthetic data to train machine learning models?when real data is limited or expensive.??
Drug Discovery: Generate virtual molecular structures and chemical compounds to?speed up the discovery of new pharmaceuticals.
Music Composition: Help composers explore new musical ideas by?generating original pieces of music.
Style Transfer:?Apply different artistic styles?to the same piece of content.?
VR/AR Development: Create virtual avatars and environments for video games, augmented reality platforms, and?metaverse gaming.?
Medical Images: Analyze medical images and?issue reports of the analysis.?
Content Recommendation:?Create personalized recommendations?for e-commerce and entertainment platforms.
Language Translation:?Translate text?from one language to another.??
Product Design:?Generate new product designs and concepts virtually?to save time and money.
?Anomaly Detection: Create virtual models of normal data patterns that will make it easier for other AI programs to identify defects in manufactured products or?discover unusual patterns in finance and cybersecurity.
Customer Experience Management: Use generative chatbots to?answer customer questions and respond to customer feedback.
Healthcare: Generate personalized?treatment plans based on multimodal patient data.?
Popular Generative AI Software Apps and Browser Extensions?
Despite concerns about the ethical development, deployment, and use of generative AI technology, genAI software apps, and browser extensions have gained significant attention due to their versatility and usefulness in various applications.?
Popular Tools for Generating Content
ChatGPT: This open-source generative AI model developed by OpenAI is known for its ability to generate realistic and coherent text. ChatGPT is available in both free and paid versions.?
ChatGPT for Google: ChatGPT for Google is a free Chrome extension that allows users to generate text directly from Google Search.?
Jasper: Jasper is a paid generative AI writing assistant for business that is known for helping marketers create high-quality content quickly and easily.?
Grammarly: Grammarly is a writing assistant with generative AI features designed to help users compose, ideate, rewrite, and reply contextually within existing workflows.?
Quillbot: Quillbot is an integrated suite of writing assistant tools that can be accessed through a single executive dashboard.?
Compose AI: Compose AI is a Chrome browser extension known for its AI-powered autocompletion and text generation features.
Popular Generative AI Apps for Art
Art AI generators provide end users a fun way to experiment with artificial intelligence. Popular and free art AI generators include:
DeepDream Generator: DeepDream Generator uses deep learning algorithms to create surrealistic, dream-like images.
Stable Diffusion: Stable Diffusion can be used to edit images and generate new images from text descriptions.
Pikazo: Pikazo uses AI filters to turn digital photos into paintings of various styles.
Artbreeder: Artbreeder uses genetic algorithms and deep learning to create images of imaginary offspring.
Popular Generative AI Apps for Writers
The following platforms provide end users with a good place to experiment with using AI for creative writing and research purposes:
Write With Transformer: Write With Transformer allows end users to use Hugging Face’s transformer ML models to generate text, answer questions and complete sentences.
AI Dungeon: AI Dungeon uses a generative language model to create unique storylines based on player choices.
Writesonic: Writesonic includes search engine optimization (SEO) features and? is a popular choice for ecommerce product description.
Popular Generative AI Apps for Music
Here are some of the best generative AI music apps that can be used with free trial licenses:
Amper Music: Amper Music creates musical tracks from pre-recorded samples.
AIVA: AIVA uses AI algorithms to compose original music in various genres and styles.
Ecrette Music: Ecrette Music uses AI to create royalty free music for both personal and commercial projects.
Musenet: Musenet can produce songs using up to ten different instruments and music in up to 15 different styles.
Popular Generative AI Apps for Video
Generative AI can be used to create video clips through a process known as video synthesis. Popular examples of generative AI apps for video are:
Synthesia: Synthesia allows users to use text prompts to create short videos that appear to be read by AI avatars.
Pictory: Pictory enables content marketers to generate short-form videos from scripts, articles, or existing video footage.
Descript: Descript uses genAI for automatic transcription, text-to-speech, and video summarization.
Runway: Runway allows users to experiment with a variety of generative AI tools that accept text, image and/or video prompts.?
SUPPLEMENT 2
???????????????? ???? ????: ? Reinforcement Learning: AI learns through feedback. ? Computer Vision: Machines interpret visual data. ? Speech Recognition: AI understands human speech. ? Hardware: Specialised processors accelerate computations. ? Neural Networks: Modelled after the human brain for deep learning. ? NLP: Machines understand and generate human language. ? Feature Engineering: Crafting meaningful features enhances performance.
?????? ???????? ??????????????????: ? Data Collection: Gathering diverse datasets for training. ? Data Security: Protecting data against unauthorised access. ? Data Integration: Combining datasets for a comprehensive view. ? Data Preprocessing: Cleaning and organising data for accuracy. ? Data Labelling: Annotating data provides context for learning. ? Data Governance: Policies for ethical data management. ? Data Privacy: Ensuring protection and ethical use of data.
???? ???????????? ?????? ????????????????????: ? AI Governance Model: Frameworks overseeing AI development. ? Ethics Frameworks: Guiding principles for ethical AI. ? Accountability in AI: Holding individuals and organisations responsible. ? Fairness and Bias Mitigation: Minimising biases for fair outcomes. ? Ethical Decision-Making: Incorporating ethics into AI decisions. ? Privacy-Preserving AI: Protecting user privacy while using AI. ? AI Auditing: Assessing AI systems for ethical compliance.
???????????????????? ???? ????: ? Algorithmic Bias: Unintended discrimination in AI models. ? Adversarial Attacks: Deliberate manipulation of data. ? Data Privacy Concerns: Risks associated with unauthorised data use. ? Ethical Dilemmas: Navigating moral choices in AI development. ? Security Risks: Potential vulnerabilities compromising AI systems. ? Resource Intensiveness: High computational requirements. ? Overfitting and Underfitting: Challenges in finding the right model complexity.
?????????????? ???????????????? ????????????: ? Regression Model: Predicting outcomes from input data. ? Ensemble Learning: Combining models for improved accuracy. ? Classification Model: Categorising input data into classes. ? Clustering Models: Grouping data based on patterns. ? Decision Trees: Hierarchical structures for decision-making. ? Random Forests: Ensemble models of decision trees. ? Transfer Learning: Leveraging pre-trained models for new tasks.
Power BI | Tableau | Python | Data Science | AI | Machine Learner | Marketing
10 个月Exciting insights on AI evolution! From narrow AI to interactive AI (IAI), each step forward unlocks new possibilities. World embeddings and interactive machine intelligence promise transformative advancements in human-computer interaction. As a machine learner, I'm thrilled by these prospects.