📌 Start here, a short guide to this blog/website.
The main purpose of this website/blog is to share what I’m up to, thoughts I’m having, media I’m consuming, and designing and redesigning the site ad nauseam.
Two podcasts and why I read/listen to biographies.
While listening to the How to Take Over the World episode of Rasputin, one of the plugs was for another similar podcast that delves into fictional biographies, it’s called Becoming the main character.
Both podcasts, synthesize books on real or fictional characters, to find the hows and whys people became great.
This led me to think about why I read biographies (mainly audiobook format), such as Teddy Roosevelt’s, Churchill’s, Napoleon’s, Hawkings’s, Feynman’s, Von Neumann’s and Gordon Brown’s…
I concluded that I look for alternatives that have not crossed my mind and special ingredients that these people may have that took them to a higher level in history.
We are all made of the same stuff, we all have the same 24 hours/day and roughly the same 30000 days/life, though some people achieve greatness, as per historical standards, and others just whiz by without a trace…
Please listen to these podcasts, I am currently listening to the Paul Atreides episode which is very interesting.
The Physical Internet
Click to open the source.
This visualization focuses on showing the physical structures and the three main objects that make the internet, IRL: data centres, internet exchange points and submarine cables.
It, of course, underlines the wealth and power of countries based on their access and infrastructure of internet services.
Made by Beatrice Bazzan and others.
Types of AI
AI can be broadly categorized into three main types based on their capabilities and functionalities:
1. Narrow or Weak AI (ANI): Narrow AI refers to artificial intelligence systems that are designed and trained for a specific task or set of tasks. These systems excel at performing a particular function within a limited domain. Examples of narrow AI include virtual assistants like Siri and Alexa, recommendation systems, image recognition software, and autonomous vehicles.
2. Artificial General Intelligence (AGI) or Strong AI: AGI, also known as Strong AI, refers to artificial intelligence systems that possess the ability to understand, learn, and apply knowledge across a wide range of tasks and domains. Unlike narrow AI, AGI can generalize its capabilities and adapt to new situations without the need for human intervention. AGI is still largely theoretical and has not yet been achieved, but it represents the goal of creating AI that can match or exceed human intelligence in all cognitive tasks.
3. Artificial Superintelligence (ASI): Artificial Superintelligence is an hypothetical AI system that surpasses human intelligence in all aspects and domains. ASI would possess cognitive abilities far beyond those of the brightest human minds and would be capable of solving complex problems, inventing new technologies, and achieving goals that are currently beyond human comprehension. ASI, if achieved, could potentially have profound implications for humanity and the future of civilization.
Machine Learning
Machine learning is a subset of artificial intelligence (AI) that focuses on the development of algorithms and statistical models that enable computers to learn from and make predictions or decisions based on data, without being explicitly programmed to do so. The primary goal of machine learning is to enable computers to learn from past experiences or data and improve their performance over time without human intervention.
There are several types of machine-learning approaches:
– Supervised Learning: In supervised learning, the algorithm is trained on labelled data, where each input data point is associated with an output label. The algorithm learns to map input data to the correct output by generalizing from the labelled examples it has seen during training. Examples of supervised learning tasks include classification (predicting categories or classes) and regression (predicting continuous values).
– Unsupervised Learning: Unsupervised learning involves training algorithms on unlabeled data, where the goal is to find hidden patterns or structures in the data. The algorithm learns to identify similarities or clusters within the data without explicit guidance. Examples of unsupervised learning tasks include clustering (grouping similar data points) and dimensionality reduction (reducing the number of features while preserving important information).
– Semi-Supervised Learning: Semi-supervised learning combines elements of supervised and unsupervised learning. It involves training models on a combination of labelled and unlabeled data, leveraging the labelled data where available and using the unlabeled data to improve generalization and performance.
– Reinforcement Learning: Reinforcement learning involves training agents to interact with an environment to achieve a specific goal. The agent learns by receiving feedback in the form of rewards or penalties based on its actions. Over time, the agent learns to take actions that maximize cumulative rewards. Reinforcement learning has applications in areas such as robotics, gaming, and autonomous systems.
Machine learning algorithms can vary widely in complexity and application, ranging from simple linear regression models to complex deep neural networks.
Common techniques used in machine learning include decision trees, support vector machines, k-nearest neighbours, neural networks, and ensemble methods.
Machine learning has numerous applications across various domains, including but not limited to:
– Natural language processing
– Computer vision
– Speech recognition
– Healthcare
– Finance
– E-commerce
– Recommender systems
– Autonomous vehicles
Generative AI
Generative AI refers to artificial intelligence systems that can produce new content, such as images, text, music, or even entire scenarios, that mimic or resemble human-created content. These systems use techniques like neural networks to learn patterns from existing data and generate new content that fits those patterns.
Key people on AI
Several key figures have significantly contributed to the field of artificial intelligence. Here are some of them:
1. Alan Turing: Considered the father of computer science and artificial intelligence, Turing developed the concept of the Turing machine, which laid the theoretical foundation for modern computing. He also proposed the famous “Turing Test” to determine a machine’s ability to exhibit intelligent behaviour.
2. John McCarthy: Coined the term “artificial intelligence” in 1956 and organized the Dartmouth Conference, which is considered the birth of AI as a field of study. McCarthy made numerous contributions to AI, including the development of the LISP programming language and the creation of the AI Lab at Stanford University.
3. Marvin Minsky: A pioneer in the field of AI, Minsky co-founded the MIT AI Lab and made significant contributions to the study of neural networks, robotics, and cognitive psychology. He also co-authored the influential book “Perceptrons,” which explored the limitations of early neural network models.
4. Geoffrey Hinton: Known as the “Godfather of Deep Learning,” Hinton is a leading researcher in artificial neural networks and deep learning. His work on backpropagation and the development of convolutional neural networks (CNNs) has had a profound impact on the field of AI, particularly in areas like computer vision and natural language processing.
5. Yoshua Bengio: Another prominent figure in deep learning, Bengio is known for his contributions to the development of deep neural networks and the advancement of unsupervised learning techniques. He co-authored the seminal paper on deep learning with Hinton and Yann LeCun, which helped spark renewed interest in neural networks in the early 2000s.
6. Yann LeCun: Renowned for his work on convolutional neural networks (CNNs), LeCun has made significant contributions to the fields of computer vision, pattern recognition, and machine learning. He is currently the Chief AI Scientist at Facebook and the Silver Professor of Computer Science at NYU.
7. Herbert Simon: A Nobel laureate in economics and a pioneer in the field of AI, Simon conducted groundbreaking research on problem-solving and decision-making processes, laying the foundation for the field of artificial intelligence.
These are just a few of the key figures in the field of artificial intelligence, and many others have made important contributions to its development and advancement.
Nick Bostrom and Andrej Karpathy are also defining currently what AI is or could be.
Deep Learning
Deep learning is a subset of machine learning that focuses on artificial neural networks with multiple layers, also known as deep neural networks. Deep learning algorithms are designed to automatically learn representations of data through the composition of multiple nonlinear transformations. These networks are capable of learning intricate patterns and relationships within large amounts of data, leading to state-of-the-art performance in various tasks, especially in fields such as computer vision, natural language processing, and speech recognition.
Key components and concepts
– Artificial Neural Networks (ANNs): Deep learning is built upon artificial neural networks, which are computational models inspired by the structure and function of biological neurons in the human brain. ANNs consist of interconnected nodes (neurons) organized into layers, including an input layer, one or more hidden layers, and an output layer. Each neuron receives input signals, performs computations, and passes the result to the neurons in the next layer.
– Deep Neural Networks (DNNs): DNNs are neural networks with multiple hidden layers between the input and output layers. These hidden layers allow DNNs to learn complex and hierarchical representations of data, capturing increasingly abstract features at each layer.
– Convolutional Neural Networks (CNNs): CNNs are a type of deep neural network specifically designed for processing structured grid-like data, such as images. They utilize convolutional layers, pooling layers, and fully connected layers to automatically learn hierarchical patterns and features in images. CNNs have achieved remarkable success in image classification, object detection, and image segmentation tasks.
– Recurrent Neural Networks (RNNs): RNNs are a type of neural network designed to handle sequential data, such as time series or natural language. RNNs have recurrent connections that allow them to maintain a memory of past inputs, enabling them to capture temporal dependencies in data. Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) are popular variants of RNNs that address the vanishing gradient problem and facilitate learning long-range dependencies.
– Generative Adversarial Networks (GANs): GANs are a class of deep learning models that consist of two neural networks, a generator and a discriminator, trained simultaneously in a competitive manner. GANs are used to generate synthetic data samples that are similar to real data, and they have applications in image generation, style transfer, and data augmentation.
Deep learning has revolutionized various industries and applications, including:
– Object detection and localization
– Autonomous vehicles and robotics
– Financial modeling and fraud detection
– Computer vision: Image recognition, facial recognition, medical image analysis.
– Natural Language Processing (NLP): Machine translation, sentiment analysis, text summarization, chatbots.
– Speech recognition: Virtual assistants, voice-enabled devices.
– Recommender systems: Personalizing recommendations on streaming services, e-commerce platforms.
– Drug discovery: Accelerating scientific research and development processes.
How it Works
Data is fed into the first layer of the ANN.
Each layer transforms the data, extracting features and identifying patterns.
As data progresses through the layers, it becomes a more abstract representation.
The final layer outputs a prediction based on the learned patterns.
Advantages
Highly effective for complex tasks like image recognition, natural language processing (NLP), and speech recognition.
Learns intricate patterns from large amounts of data, often exceeding human-level accuracy in specific domains.
Can handle unstructured data like images and text data without the need for extensive pre-processing.
Disadvantages
Computationally expensive: Training deep learning models requires significant computing power and large datasets.
Data dependency: Relies heavily on the quality and quantity of data. Biases in the data can lead to biased models.
Explainability: Understanding how a deep learning model arrives at a decision can be challenging, making it a “black box” in some cases.
Deep learning algorithms require large amounts of labeled data for training and significant computational resources for training deep models, often utilizing graphics processing units (GPUs) or specialized hardware accelerators.
Large Language Model
LLM typically stands for “Large Language Model.” It refers to AI models like GPT (Generative Pre-trained Transformer) that are capable of understanding and generating human-like text on a large scale. These models are trained on vast amounts of text data and can perform various natural language processing tasks, including text generation, translation, summarization, and more.
The size of Large Language Models (LLMs) can vary, but they typically have hundreds of millions to billions of parameters. For example, models like GPT-3 have 175 billion parameters, while newer models might have even more. These parameters represent the weights and connections within the neural network that enable the model to process and generate text.
Bookending
I recently read a post by Mandy Brown about a best practice that she incorporated into her routine.
She calls it bookending, as in the bookends that open and close the content of a book.
It is an SOP that defines a beginning and an end to the work day. It consists of the following steps:
1. At the beginning of your work day, start with a blank sheet of paper or notebook and visualise what the day will look like, before getting into the busyness of emails and notifications.
2. At the end of the day, close all the apps, shut down the computer, and review the day.
These simple steps limit and define your workday, especially relevant in a remote work environment.
This concept is something I’ve seen multiple times, though have not yet fully implemented it.
Many Brown -> Bookending
Sahil Bloom -> Power down ritual
Cal Newport -> Work shut down ritual
What is AI
AI, or artificial intelligence, refers to the simulation of human intelligence processes by machines, especially computer systems.
These processes include learning (the acquisition of information and rules for using the information), reasoning (using rules to reach approximate or definite conclusions), and self-correction.
AI technologies encompass a wide range of applications, from robotics to natural language processing, computer vision, machine learning, and more.
The goal of AI is to create systems that can perform tasks that would typically require human intelligence. These tasks can include problem-solving, decision-making, understanding natural language, recognizing patterns, and adapting to new situations.
There are different approaches to AI, but a common thread is the use of algorithms that learn from data or experience. This enables AI systems to improve their performance over time.
Artificial Intelligence (AI) also refers to computer systems capable of performing tasks that historically required human intelligence, such as recognizing speech, making decisions, and identifying patterns. It encompasses a wide variety of technologies, including machine learning, deep learning, and natural language processing (NLP).
Here are some examples of AI applications you might have encountered:
- Recommendation Systems: These use machine learning algorithms to suggest movies, songs, or products based on your preferences.
- Navigation Apps: They find the fastest route to your destination by analyzing traffic data.
- Language Translation: AI-powered tools translate text from one language to another.
- Chatbots: These provide real-time customer support and engage in conversations.
- Computer Vision: AI algorithms can analyze images and videos, enabling tasks like facial recognition or object detection.
Although there are philosophical disagreements about whether “true” intelligent machines exist, when most people use the term AI today, they’re referring to a suite of machine learning-powered technologies that enable machines to perform tasks previously achievable only by humans, such as generating written content, steering a car, or analyzing data.
AI is having a profound impact on our world, and its applications are becoming increasingly widespread.
How to learn AI
Learning AI involves a multidisciplinary approach that includes understanding concepts from mathematics, computer science, and domain-specific knowledge. Here’s a suggested path to learn AI:
1. Mathematics Fundamentals
– Linear Algebra Understand vectors, matrices, transformations, and eigenvalues.
– Calculus Learn differential and integral calculus.
– Probability and Statistics Study probability theory, random variables, distributions, and statistical inference.
2. Programming Skills
– Learn a programming language commonly used in AI, such as Python or R.
– Familiarize yourself with libraries and frameworks used in AI, such as TensorFlow, PyTorch, or scikit-learn.
3. Machine Learning Basics
– Understand foundational concepts such as supervised learning, unsupervised learning, and reinforcement learning.
– Learn about common machine learning algorithms like linear regression, logistic regression, decision trees, k-nearest neighbors, support vector machines, clustering algorithms, etc.
4. Deep Learning
– Dive deeper into neural networks, including concepts like feedforward neural networks, convolutional neural networks (CNNs), recurrent neural networks (RNNs), and transformers.
– Learn about deep learning frameworks like TensorFlow and PyTorch.
– Understand how to train and optimize deep learning models.
5. Natural Language Processing (NLP)
– Study techniques for processing and understanding human language, such as tokenization, word embeddings, sequence-to-sequence models, and transformers.
– Learn about NLP libraries like NLTK, SpaCy, and Hugging Face Transformers.
6. Computer Vision
– Explore techniques for processing and understanding images and videos, including convolutional neural networks (CNNs), object detection, image segmentation, and image classification.
– Familiarize yourself with computer vision libraries like OpenCV and deep learning frameworks for vision tasks.
7. Reinforcement Learning
– Understand the principles of reinforcement learning, including Markov decision processes, policy gradients, Q-learning, and deep Q-networks (DQNs).
– Experiment with reinforcement learning algorithms and environments.
8. Projects and Practice
– Apply your knowledge by working on AI projects. Start with simple projects and gradually increase complexity as you gain proficiency.
– Participate in AI competitions or contribute to open-source AI projects.
– Continuously practice coding and experimenting with different algorithms and techniques.
9. Stay Updated
– AI is a rapidly evolving field, so it’s essential to stay updated with the latest research papers, conferences, and advancements.
– Follow AI experts, join online communities, and read blogs and forums to stay informed about emerging trends and techniques.
10. Specialize
– Once you have a solid foundation, consider specializing in a specific area of AI that aligns with your interests or career goals, such as computer vision, NLP, robotics, or healthcare AI.
Remember that learning AI is a continuous journey, and practical hands-on experience is crucial for mastering the concepts effectively. Keep practicing, experimenting, and exploring new ideas to become proficient in AI.
Colophon
Built with WordPress.
The font is based on a system font stack.
I designed and coded the website myself. (any tips?)
This website got an A+ on the Websitecarbon score.
The color palette is:
#233142;
#b0bec5;
#0277bd;
#FFD800;
How I work
1. Email is the best way to contact me. Chat, slack… just get lost.
2. I work M-F, 09.30-17.30. (Some time more, if needed).
3. If you need anything from me, please clearly let me know what is needed and when.
4. I prefer meetings after 10 am, and before 16 pm.
5. My time zone is GMT+2.
* Inspired by Jamie Tana and Melanie Richards
Pancho
This is Pancho, our dog. He’s a Spanish waterdog.
Pocketmoney
I read this post from Michael Karnjanaprakorn about he handles his girl’s allowance.
I think there are some great ideas, that I am taking for Lucas’ and Nico’s pocket money in a few years:
– How to set the allowance: weekly, 1€/year. i.e: 10 years, 10euros/week.
– Compounding: set an interest rate on savings, i=10%/year. This teaches them about compunding.
– Doubling savings: every year if savings are over 50€, double it to reinforce patience and savings.
This will need a spreadsheet, another great skill to teach.
A thoughtful and great approach on how to handle finance skills and allowance.
Dune Part 2
I enjoyed part one of the recent films of DUNE, honestly, I have not seen the old ones, so I can’t compare. I don’t like how old films represented the future (except Star Wars) though they made the best out of the available tools.
In the first film, I was amazed at how they managed to create such incredible recreations of the different great houses, how well it was filmed, how great the technologies (ornithopters, protective shields…) are, the story…
Of course, the second film still is great in all those regards, but it seems they tried cramming too much of the story in a single film (which is understandable) and there’s no “wow factor”, but that may just be me. I was getting used to amazing things and they no longer have such the impact that they used to have.
All in all, I give it a 4/5.