Real AI 101: AI > ML > DL > Generative AI > Causal AI > Interactive AI
“To build truly intelligent machines, teach them to learn about reality, its entities and interactions, causes, mechanisms and effects, to sustainably interact with the world and its data universe” - ASHA
The idea of true real and interactive AI is to integrate causal learning with machine learning
True Real Interactive AI (TRIAI) = AI + ML + DL + ANNs + GAI + CAI + Interaction Networks Model Engine = Trans-AI = Real/True AI
The TRAI paradigm [AI > ML > DL > Generative AI > Causal AI > Interactive AI] is overruling the traditional correlational AI paradigm [Artificial Narrow Intelligence (ML; DL; NLP/NLG; Robotics) > AGI > ASI]
What is Today's AI/ML/DL?
There are many different definitions and versions and aspects and applications of Artificial Intelligence, Machine Learning and Deep Learning, while there is no definition, WHICH IS GENUINE, TRUE AND REAL.
Artificial intelligence (AI) systems are software (and possibly also hardware) systems designed by humans that, given a complex goal, act in the physical or digital dimension by perceiving their environment through data acquisition, interpreting the collected structured or unstructured data, reasoning on the knowledge, or processing the information, derived from this data and deciding the best action(s) to take to achieve the given goal.
AI systems can either use symbolic rules or learn a numeric model, and they can also adapt their behaviour by analysing how the environment is affected by their previous actions.
As a scientific discipline, AI includes several approaches and techniques, such as machine learning (of which deep learning and reinforcement learning
"Machine Learning?is an AI technique that teaches computers to learn from experience. Machine learning algorithms use computational methods to “learn” information directly from data without relying on a predetermined equation as a model. The algorithms adaptively improve their performance as the number of samples available for learning increases. Deep learning is a specialized form of machine learning".
What is Generative AI (GAI)?
Generative AI describes any type of artificial intelligence that can be used to create new contents, text, images, video, audio, code or?synthetic data.
Generally, it could embrace machine learning and deep learning, being associated with deepfakes and ChatGPT.
Sometimes, there is a narrow interpretation. "Generative AI is a subfield of machine learning that involves training artificial intelligence models on large volumes of real-world data to generate new contents (text, image, code,…) that is comparable to what humans would create. This is achieved by training algorithms on large datasets to identify patterns and learn from them. Once the neural network has learned these patterns, it can generate new data that adheres to the same patterns. However, this process is computationally intensive.
Different model architectures, such as diffusion models and Transformer-based large language models (LLMs), can be employed for generative tasks such as image and language generation.
Diffusion models are a type of generative AI model that can be used for a variety of tasks, including image generation, image denoising, and inpainting. Similarly, the Transformer architecture revolutionized the language domain. The new era of language models are Transformer-based, which is a type of deep learning architecture for natural language processing (NLP) tasks. They utilize a self-attention mechanism to transform the input sequence into a set of context-aware high dimensional vectors (also known as embeddings) that can be used for a variety of NLP tasks, including language generation, machine translation, and text classification.
The most well-known transformer-based LLMs are the GPT family, developed by OpenAI. The primary advantage of transformer-based LLMs over traditional NLP models is that they are highly parallelizable and can handle long-range dependencies between words in a sentence more effectively. This makes them more suitable for tasks that require a deeper understanding of the context, such as text summarization or generating a coherent and fluent text.
The breakthroughs in Generative AI have left us with an extremely active and dynamic landscape of players. This consists of 1) AI hardware manufacturers such as Nvidia and Google, 2) AI cloud platforms such as Azure, AWS, Nvidia and Google, 3) open source platforms for accessing the full models, such as Hugging Face, 4) access to LLM models via API such as OpenAI, Cohere and Anthropic and 5) access to LLMs via consumer products such as ChatGPT and Bing. Additionally, there are many more breakthroughs happening each week in this universe such as the release of multi modal models (that can understand both text and image), new model architectures (such as Mixture of Experts), Agent Models (models that can set tasks and interact with each other and other tolls).
By some measures, consumer facing Generative AI has become the fastest growing technology trend of all time, with various models emerging for image, text, and code generation.
For example, MidJourney’s Discord has attracted around 13 million members for Image Generation, while ChatGPT has reportedly gained over 100 million users within a few months of release.
Software development use cases have also seen a significant rise with over 1.2 million developers using GitHub Copilot’s technical preview as of September. [The Generative AI Revolution: Exploring the Current Landscape]
GAI involves ML based on correlational pattern recognition what is insufficient for robust predictions and reliable decision-making. Causal AI combining machine learning with causal inference methods provide a promising path forward.?
What is Causal AI (CAI)?
Generative AI and causal AI are related in that both can be used for generating new data or making predictions, but?generative AI generates new data based on existing data patterns, while causal AI focuses on understanding the relationships that influence the data being analyzed.
Causal AI?is an artificial intelligence system that can explain cause and effect, quantifying the impact of different interventions, policy decisions or performing scenario planning.
Causal AI reasons and makes decisions like humans. By analyzing data and identifying the cause-and-effect relationships between variables, Causal AI will help you find precise and actionable strategies to tackle extremely complex scenarios. But in a dynamic, transparent, and bias-free way.
"Causal AI?teaches you how to build machine learning and deep learning models that implement causal reasoning".
领英推荐
The key idea behind Causal AI is to learn cause and effect relationships within data and use this to inform the output of AI models. This is vastly different from the approach that current state-of-the-art ML models, such as LLMs take, deep neural networks using billions of parameters and trained with petabytes of biased data: consume a lot of data, learn the correlative patterns and predict the next stochastic pattern.
Causal AI is the only technology that can reason and make choices like humans do. It utilizes causality to go beyond narrow machine learning predictions and can be directly integrated into human decision-making. It is the only AI system organizations can trust with their biggest challenges – a revolution in enterprise AI.
Causal discovery?is the process of combining algorithms and domain expertise to discover a causal graph from observational data. Causal graphs attempt to model the underlying data generating process rather than simple associations between variables.
CAI is driven by the Causality Inference Engine relying on Causal [Graph] Model. Graphical causal models are applied in statistics, econometrics, epidemiology, genetics and related disciplines,?also known as?path diagrams, causal?Bayesian/belief/decision networks?or?DAGs, as?probabilistic graphical models, all via a DAG (directed acyclic graph).
There is "Causality Link, a CAI cloud-based platform that can read and understand millions of words each day from thousands of sources worldwide,... with automated causation identification and analysis to maximize your understanding of the future".
The idea is to integrate causal learning with machine learning, deep learning and artificial neural networks, combining statistical models, data analytics techniques, narrow AI models, ML mechanisms, DL algorithms and integrative ontological [causal] frameworks.
What is Interactive AI?
Interactive AI (IAI) = AI + ML + DL + ANNs + GAI + LLMs + CAI + World Model [Learning, Inference and Interaction] Engine = Transdisciplinary, Transformative and Translational AI = Trans/Meta/Hyper-AI = Trans-AI
The domineering assumption of AI as emulating, mimicking, simulating, or replicating human body/intelligence/brains/mind/behavior is scientifically unjustified and ethically harmful and existentially risky and should be discarded in the favor of causative machine intelligence and learning as an alternative and augmenting intelligence (AAI).
Being automatic and autonomous, adaptive and reactive, transformative and translational, transdisciplinary and integrative, mathematical and statistical, computational and algorithmic, digital and numeric, proactive and interactive, AI systems are complementary with humans, our bodies and brain, brains, behavior and business.
IAI is?a range of algorithms and models, techniques and technologies, enabling computing machines to effectively and sustainably interact with the world, transforming its data into information, actions, and reactions.
Then IAI algorithms find natural or causative patterns in data that generate causal insight to provide critical decisions and realistic predictions in all parts of human life, be it scientific research or engineering design, medical diagnosis, stock trading, energy load forecasting, space exploration and more.
Again, the full causal graph networks of complex interactions, or Interaction Networks (INs), are the best intelligent machine organization of data, structured or unstructured. Interaction networks?deal with things or entities or objects, substances, states, or changes, and the interactions among or between them.
Examples of such causal hypergraphs are?all the complex real-world systems and digital networks, molecular networks, neural networks, food webs, signal transfer path- ways, the bitcoin network, social networks, traffic networks, etc.
The IAI's interaction networks can capture the flow of matter and energy, data and information and knowledge, good and services, ideas and humans, transferred between their nodes or vertices, with all possible combinations and causal topologies, lines and chains, triangles or cycles.
Again, data, unstructured, semi-structured or structured, as SQL tables, spreadsheets, etc., tweets or text, could be transformed from its original format using an Extract-Transform-Load (ETL) or Extract-Load-Transform (ELT) into a Causal Graph format.
One could prompt an LLM with an ontology/schema/world model to drive Causal Knowledge Graph extraction from unstructured or structured documents.
DATA [unstructured or structured] > GLOBAL ONTOLOGY PROMPT > LLM [ChatGPT or Bard] > EXTRACTED CAUSAL KNOWLEDGE GRAPH [scientific knowledge, common sense knowledge, machine intelligence]
ANY COMPLEX SYSTEMS, BE IT THE WHOLE UNIVERSE OR QUANTUM SYSTEMS, IS THE CAUSAL NETWORK OF THINGS, FORMALIZED AS THE FULL CAUSAL GRAPH NETWORKS OF ENTITIES AND THEIR INTERACTIONS.
ANY INTELLIGENCE, BE IT GOD or HUMANS, ANIMALS or MACHINES, follows the World (natural environment, humans, information processing systems) - Intelligence/Knowledge/Information/Data System Interaction Algorithms, as
The World of Realities [physical, mental, social, digital, virtual, cyber-physical, etc.] - Causal Input (Matter, Energy, Information) - Transformation/Translation Mechanism/Function - Causal Output - Causal Feedback - Reinforcing or Balancing Interactive Loop [with all the transformation and translation, explanation and transparency, inference and prediction, actions and reactions] - The World of Realities
It is true for AI/LLMs, where the Causal Graph of Intelligence follows the key causation models,?simple linear models, complex linear models and complex non-linear models, as
possible prompts, from a simple linear prompt to the complex Graph of Thought, topped by the causal loop diagrams, with ''the Human-in-the-Loop (HITL)?enabling human verification and corrections to ensure accuracy of data extracted by Document AI processors before it is used in critical business applications":
Crucial, such an IAI can’t be over-attributed human values and attitudes, desires and beliefs, emotions and feelings, cognition and decisions.
It naturally aligns very well with the ethical and moral principles, as below:
1. Safety, security and robustness
2. Transparency and explainability, an understandable and explainable AI
3. Fairness, an AI system should not perpetuate or exacerbate existing biases or discrimination, designed to treat all individuals and demographic groups fairly
4. Accountability and governance, Human oversight, Every AI system should be designed to enable human oversight and intervention when necessary
5. Non-maleficence, AI systems should be designed and used in a way that does not cause harm.
Today's AI is with the inherent negative financial, reputational and ethical risks that?black box AI and its machine bias?can introduce.
It is hardly any responsible AI, as?illegally scraping the web/internet data?to get the large data sets required to train their generative models.
Governments, companies and organizations develop or using a non-interactive acausal AI have a responsibility to govern the technology by establishing their own policies, guidelines, best practices and maturity models for RAI.
There are a lot of additional efforts for creating the Toolkits for Responsible AI and New Legislation for Responsible AI, such as the General Data Protection Regulation (GDPR), Liability Rules for Artificial Intelligence, and EU AI Act in the European Union.
The IAI embraces both RAI, XAI or AI for good, the concept of using AI to address a social or environmental challenge, to help solve the world’s problems such as poverty, hunger and climate change.
Conclusion
AI must be upgraded as?the transdisciplinary science and engineering of making intelligent machines, as complementing and augmenting human intelligence, individual and collective.
To build truly intelligent machines, we need to teach them to learn the world's interaction networks, with all possible structures and patterns and probable interrelationships, uniformly presenting any complex systems, physical, chemical, biological or mental, social, informational or technological, virtual, digital or cybernetic.
The Interactive AI paradigm [AI > ML > DL > Generative AI > Causal AI > Interactive AI] is embracing the traditional Non-Interactive AI paradigm [Artificial Narrow Intelligence (ML; DL; NLP/NLG; Robotics) > AGI > ASI]