Posts tagged #ai

A recent resignation tweet from an Anthropic researcher has everyone talking about the AI apocalypse again. Among other things, he said: The people building AI…

Their warnings came just days after top OpenAI leaders, including CEO Sam Altman, said AI is speeding into a scary, uncontrollable phase.

Just over a year ago I wrote Why we should anthropomorphize LLMs. Now it’s a hot topic again, driven by Dwarkesh Patel’s description of OpenAI’s recent swarm…

Over the last three months I have written thousands of lines of code and yet I didn't write any of them, claude.ai wrote them. The world of programming, software engineering, computer science, in my opinion has been turned upside down in a matter of months.

How I code today

I've created a process within the world of Claude Code which I use to design, plan and implement code. Many people have created similar ones and so instead of describing mine, in detail, let me explain the general idea. The heart of it is a series of feature specs (in individual numbered markdown files) and for each feature spec a task list (in a different markdown file as well) ...

Jakub Pachocki reflects on increasingly capable AI and the challenge of keeping it aligned. He calls for stronger safeguards and international coordination.

The Coming of Userless Agents

Coding agents as normal technology

Interesting article: GPT in 60 Lines of NumPy | Jay Mody. Implementing a GPT model from scratch in NumPy. This is a detailed article including python source code. Ive skimmed it and it does explain a lot. But even without my being a neural net, ai, machine learning, gpt, chatgpt expert, i think i would learn a lot. Also the article is very well written presented.

Note: written a few years ago before we know how far the "AI thing" was going to go!

I've been dreading this: My robot's rasberry pi 5 was built months ago. I have been using it a lot, installing, updating and changing code and config. All my code is separately in github repos. But there is a lot more state than the github repos. It is in config files, pip installs, and so on. I started to worry about the scenario where the MicroSD on the Pi just bites the dust. This does happen, randomly, and occasionally.

I could make a full and faithful copy of the MicroSD. It's a little time consuming but not hard. I could start over with a clean Ub...

Anthropic is an AI safety and research company that's working to build reliable, interpretable, and steerable AI systems.

Invasive monitoring and a fresh round of layoffs have workers I spoke to on edge. Is this the future of knowledge work?

In higher education, AI should be a human-centered tool inside a curriculum that expects more from students, not less.

🪨 why use many token when few token do trick — Claude Code skill that cuts 65% of tokens by talking like caveman - JuliusBrussee/caveman

I came across a fascinating research paper that puts LLMs head-to-head with human participants in a classic creativity task, and the results are worth sitting with for a moment. Ideas are Dimes a Dozen: Large Language Models for Idea Generation in Innovation examines how well models like GPT perform when asked to generate novel business and product ideas compared to actual humans.

The headline finding is both humbling and thought-provoking: LLMs can produce a higher quantity of ideas that are rated comparably in quality, but the very best human ideas still tend to stand out at the top of the distribution. What I find most inte...

There’s an old concept called Literate Programming. It never stuck or took off, but it came back to me the other day. Here is the Wikipedia explanation. My summary of this approach is to write a program (source code) as if it were a chapter in a book — primarily text and diagrams, interspersed with bodies of code. Hold that thought.

My latest use of Claude Code has yielded some impressive stuff. For the first time I am working with neural networks for image processing and object identification. I am using a camera, the Oak‑D‑Lite, which has a neural‑net processing chip in the camera i...

I don't like this one bit

By Nicolas Hulscher, MPH

All the code for this new version of my web site was written by claude code. Yes, I am not exagerating. Here are some of my take aways.

Amazing Yes, start with admiting that it is amazing. This is a fairly simple application but it has a fair number of moving parts. The code is all open source so you can see for yourself.

Process Basically I have two terminal windows open and vs code on the side. Terminal window 1 is running claude code. What that means is that I cd into the app directory and type claude. That's it (after installing claude of course. It starts up with kind of a "how can I help you" prompt, kind of like a doctor.

From then on you address it like you would a ver...

Amazing ai generated videos

Get up and running with large language models.

The first real AI developer. Contribute to Pythagora-io/gpt-pilot development by creating an account on GitHub.

A framework for rapid development and deployment of production-ready RAG systems - SciPhi-AI/R2R

Syllabi Policies for AI Generative Tools –Syllabi Policies for AI Generative Tools If you would like to submit your course guidelines/policy to be included here, please submit it in this form. Update: If you would like a more searchable version of this document, try out this spreadsheet that allows you to sort by Course, Discipline…

How to Use AI to Do Stuff: An Opinionated Guide –

Whoops. That doesn’t work at all. Let’s ask it to improve it:

Opinion | I’m a Student. You Have No Idea How Much We’re Using ChatGPT. –

GPT Engineer: Build an Entire App with a Single Prompt –Have an idea. Enter a prompt. Get an app.

Are Your Students Ready for AI? –To thrive in a world that now includes generative AI tools like ChatGPT, students need to develop five essential skills, says professor Oguz A. Acar. Use this four-step framework to empower your students and equip them for success in the digital age.

Why AI Will Save the World | Andreessen Horowitz –

llm, ttok and strip-tags—CLI tools for working with ChatGPT and other LLMs –I’ve been building out a small suite of command-line tools for working with ChatGPT, GPT-4 and potentially other language models in the future. The three tools I’ve built so far …

Copilot Leaks: Code I Should Not Have Seen –The number of leaked information from Copilot were surprising. I’ll show you what it his and how to prevent it your code from leaking.

How-To Add Recommendations to a Rails App with pgvector and OpenAI –Creating recommendations with pgvector, ruby-openai, and the neighbor gem.

GigaBrain – Search Reddit and Other Communities for Answers from Real People –GigaBrain finds the most useful discussions on reddit and other communities. We sift through the noise and analyze billions of comments for you. Get real answers from real people.

I have been using ChatGPT on a daily basis. Not for joking or playing but for work. It is like a super search engine combined with a fallible expert. When I tell this story to various people they seem surprised and ask “Give me an example.” I have so many. So this will be a series of daily examples where ChatGPT made a difference.

By the way, people love to say, ChatGPT is nothing more than fancy autocomplete. While they may be technically correct (and I doubt they are) you cannot ignore ChatGPT feats that look nothing like autocomplete.

Tell me how this is simple word completion. Can this be broadened to a more complicated network. Could it be used for some kind of semi-intelligent navigation?

Prompt:

Assume a map that is described in sentences, with node names and costs to go between nodes. Here is map1:

Node A costs 5 to got to Node C and 10 to go to Node D Node C costs 100 to go to node E and 50 to go to node F Node D costs 10 to go to node E and 100 to go to node F

What is the best route to go from Node A to node E? And what is the best route to go from Node A to node F?

Response:

To find the best route from Node A to Node E, we need...

ChatPDF – Chat with any PDF! –ChatPDF is the fast and easy way to chat with any PDF, free and without sign-in. Talk to books, research papers, manuals, essays, legal contracts, whatever you have! The intelligence revolution is here, ChatGPT was just the beginning!

Use ChatGPT From The Command Line With This Wrapper –ChatGPT Wrapper is an unofficial open source command-line interface and Python API for interacting with ChatGPT.

Author says: “Every week now, it seems, events on the ground make a fresh mockery of those who confidently assert what AI will never be able to do, or won’t do for centuries if ever, or is incoherent even …”

Why am I not terrified of AI? –Author says: “Every week now, it seems, events on the ground make a fresh mockery of those who confidently assert what AI will never be able to do, or won’t do for centuries if ever, or is incoherent even …”

CheatGPT –If you teach programming, this is a very insightful article. It doesnt solve anything though. From the author: “This week I stepped out of one world and landed in another. It started while I was marking assignments in a few of my programming courses.The quality of the code I was reading was amazing! My students were doing the sorts of things that my open source peers do.”

Transformer models: an introduction and catalog — 2023 Edition –Author” “I have a terrible memory for names. In the past few years we have seen the meteoric appearance of dozens of models of the Transformer family, all of which have funny, but not self-explanatory, names. The goal of this post is to offer a short and simple catalog and classification of the most popular Transformer models. In other words, I needed a Transformers cheat-sheet and couldn’t find a good enough one online, so I thought I’d write my own. I hope it can be useful to you too”

What Is ChatGPT Doing … and Why Does It Work? –This is a pretty amazing article. Even though it’s “non technical” and I read all of it, I think I only understood about 75%: “Stephen Wolfram explores the broader picture of what’s going on inside ChatGPT and why it produces meaningful text. Discusses models, training neural nets, embeddings, tokens, transformers, language syntax.”

GPT in 60 Lines of NumPy | Jay Mody –Implementing a GPT model from scratch in NumPy. This is a detailed article including python source code. Ive skimmed it and it does explain a lot. But even without my being a neural net, ai, machine learning, gpt, chatgpt expert, i think i would learn a lot. Also the article is very well written presented.

Abstracts written by ChatGPT fool scientists –Researchers cannot always differentiate between AI-generated and original abstracts.

karpathy/nanoGPT: The simplest, fastest repository for training/finetuning medium-sized GPTs. –This looks useful, which is why I am linking to it, I haven’t tried it or anything: “The simplest, fastest repository for training/finetuning medium-sized GPTs. – karpathy/nanoGPT: The simplest, fastest repository for training/finetuning medium-sized GPTs.

Opinion | The Alt-Right Manipulated My Comic. Then A.I. Claimed It. –Another interesting continuation of the debate on the impact of chatGPT and Stable diffusion and the other newest ai based creative genetors. I think these new tools are a big thing, a game changer. Will it bring on a series of new and common ways tomdigitslly sign content produced? This idea sounds ok but there are a lot of issues with bringing it about. Will it bring out a new, singular, ethical concept sitting along side “original work” and “plagiarism”? Here is the author: “My drawings are a reflection of my soul. What happens when artificial intelligence — and anyone with access to it — can replicate them?”

Laurence Tratt: How Might Generative AI Change Programming? –More about a topic that is extremely hot right now

ChatGPT Can’t Kill Anything Worth Preserving –If you, like many, is fascinated and trying to figure out what ChatGPT (and whatever comes next) means, this is a good article: “If an algorithm is the death of high school English, maybe that’s an okay thing.”

ggerganov/whisper.cpp: Port of OpenAI’s Whisper model in C/C++ –Port of OpenAI’s Whisper model in C/C++. Contribute to ggerganov/whisper.cpp development by creating an account on GitHub.

Learning Rust with ChatGPT, Copilot and Advent of Code –Another amazing and somewhat scary example of ChatGPT, the ai tool that everyone is takking about. Heres what the author says: lI’m using this year’s Advent of Code to learn Rust—with the assistance of GitHub Copilot and OpenAI’s new ChatGPT. I think one of the most exciting applications of large language …”

AI-generated answers temporarily banned on coding Q&A site Stack Overflow –I’m interested what’s going to happen as ChatGPT gets better. I hadn’t thought about StackOverflow. But I was thinking about students using it to do homework assignments, like papers and essays. Here’s what the author says: “People have been using OpenAI’s chatbot ChatGPT to flood the site with AI responses, but Stack Overflow’s mods say these ‘have a high rate of being incorrect.’”

For 2022, the performance of this AI is truly amazing. Maybe in a year or five it will seem simple but for now, its crzay good and a bit scary: “A conversational AI system that listens, learns, and challenges. “

ChatGPT –For 2022, the performance of this AI is truly amazing. Maybe in a year or five it will seem simple but for now, its crzay good and a bit scary: “A conversational AI system that listens, learns, and challenges. “

Building A Virtual Machine inside ChatGPT –This is a truly amazing and mind-bending example of something that ChatGPT can do. Wheels within wheels. The Matrix. Check out what the author says: “Unless you have been living under a rock, you have heard of this new ChatGPT assistant made by OpenAI. Did you know, that you can run a whole virtual machine inside of ChatGPT?l

Futurepedia is the largest AI tools directory. Browse 200+ AI tools in various categories like marketing, image generation and video editing. Search and filter the tools by categories, pricing and features.

Futurepedia – The Largest AI Tools Directory | Home –Futurepedia is the largest AI tools directory. Browse 200+ AI tools in various categories like marketing, image generation and video editing. Search and filter the tools by categories, pricing and features.

palette.fm –A new AI colorizer. Colorize anything from old black and white photos 📸, style your artworks 🎨, or give modern images a fresh look 🌶. It’s as simple as instagram, free, and no sign-up required!

From speech to insights: The value of the human voice –Live voice contacts are essential for customer experience. In this McKinsey article, we explore how new speech analytics technology is driving lasting results.

GitHub Copilot investigation · Joseph Saveri Law Firm & Matthew Butterick –Very interesting review of github copilot, open source and the law.

Huskylens is an easy-to-use AI camera | Vision Sensor. DFRobot Huskylens (SEN0305) is equipped with multiple functions, such as face recognition, object tracking, object recognition, line tracking, color recognition, and tag(QR code) recognition.

HuskyLens – An AI Camera: Click, Learn, and Play | Vision Sensor | DFRobot Electronics Huskylens is an easy-to-use AI camera | Vision Sensor. DFRobot Huskylens (SEN0305) is equipped with multiple functions, such as face recognition, object tracking, object recognition, line tracking, color recognition, and tag(QR code) recognition. June 12, 2022 at 03:01AM

PyImageSearch University Courses: You can master Computer Vision, Deep Learning, and OpenCV Computer Vision, Deep Learning, and OpenCV courses from PyImageSearch and Dr. Adrian Rosebrock March 18, 2022 at 02:24PM

You Only Look Once: Unified, Real-Time Object Detection We present YOLO, a new approach to object detection. Prior work on object

detection repurposes classifiers to perform detection. Instead, we frame object

detection as a regression problem to… March 14, 2022 at 01:31AM

What Does It Mean for AI to Understand? | Quanta Magazine It’s simple enough for AI to seem to comprehend data, but devising a true test of a machine’s knowledge has proved difficult. January 2, 2022 at 02:41PM

Predictions Scorecard, 2022 January 01 – Rodney Brooks

January 2, 2022 at 02:50PM

Minimax algorithm and alpha-beta pruning This article will teach you about the minimax algorithm and alpha-beta pruning, from a beginner’s perspective. November 27, 2021 at 07:32PM

Python implementation of the Eliza chatbot. Contribute to wadetb/eliza development by creating an account on GitHub.

wadetb/eliza: Python implementation of the Eliza chatbot Python implementation of the Eliza chatbot. Contribute to wadetb/eliza development by creating an account on GitHub. August 9, 2021 at 02:46PM

Convert a floorplan magically into 3d rendering.

The official project of SimSwap (ACM MM 2020). Contribute to neuralchen/SimSwap development by creating an account on GitHub.

neuralchen/SimSwap The official project of SimSwap (ACM MM 2020). Contribute to neuralchen/SimSwap development by creating an account on GitHub. June 18, 2021 at 03:23PM

Plan2Scene Convert a floorplan magically into 3d rendering. June 18, 2021 at 03:25PM

This is Grammar For Robots. What? Why? 🤖 https://lambdalabs.com/papers📝 The paper “RoboGrammar: Graph Grammar for Terrain-Optimized Robot designer. Very intersting paper ! June 12, 2021 at 07:58PM

Machine Learning: The Great Stagnation The bureaucrats are running the asylum June 6, 2021 at 02:57AM

Neural Magic helps developers in accelerating deep learning performance using automated model sparsification technologies and a CPU inference engine. - Neural Magic

Neural Magic Neural Magic helps developers in accelerating deep learning performance using automated model sparsification technologies and a CPU inference engine. – Neural Magic May 19, 2021 at 04:33PM

Companion site for the book Artificial Intelligence for Computer Games: An Introduction. Includes bibliography, errata, chapter guide, and relevant links.

A Recipe for Training Neural Networks

Very good article for anyone using Neural Networks

tags: neuralnetworks training howto

Posted from Diigo. The rest of my favorite links are here.

Ganbreeder

Pretty interesting example of generative art and AI deepfakes related technology. And pretty pictures too.

tags: deepfakes ai graphics art

Posted from Diigo. The rest of my favorite links are here.

Podcast: Can Machine Learning Solve the Challenges of Cloud Computing? | Graduate School of Arts and Sciences (GSAS) | Brandeis University

Good introduction to Cloud Computing and Machine Learning

tags: elasticity scalability machinelearning

jhspetersson/fselect

Another crazy useful tool for programmers.

tags: utilities ...

Getafix: How Facebook tools learn to fix bugs automatically – Facebook Code

Amazing automated bug discovery and fixing tool

tags: tools AI bugs facebook

Posted from Diigo. The rest of my favorite links are here.

How to Approach Machine Learning Problems

tags: machine learning

Posted from Diigo. The rest of my favorite links are here.

Top 8 Data Structures for Coding Interviews and practice interview questions

tags: . interviews coding

[FoR&AI] Steps Toward Super Intelligence II, Beyond the Turing Test – Rodney Brooks

tags: super intelligence AI ...

Automated Machine Learning on the Cloud in Python – Towards Data Science

tags: google machine learning Python

Posted from Diigo. The rest of my favorite links are here.

Frontier AI: How far are we from artificial “general” intelligence, really?

tags: artificial intelligence

Posted from Diigo. The rest of my favorite links are here.

Random Forest in Python – Towards Data Science

Good overview

tags: statistics machine learning

Posted from Diigo. The rest of my favorite links are here.

JupyterLab is Ready for Users – Jupyter Blog

tags: python Jupiter datascience notebook statistics

...

Build a Text Classification Program: An NLP Tutorial

A little difficult but one can follow and try this out.

tags: classification text tutorial nlp

...

In the world of machine vision, the equivalent goal is to win the ImageNet Large-Scale Visual Recognition Challenge. This is a competition that has run every year since 2010 to evaluate image recognition algorithms. (It is designed to follow-on from a similar project called PASCAL VOC which ran from 2005 until 2012) Click to read the article.

What if when I typed into the search box in Safari (Mac OS X’s web browser) the text is interpreted by Siri? I could type “What’s the weather tomorrow” and it would display a forecast for tomorrow, or type “text my wife that I love her” and it would initiate a text message to my wife. Wouldn’t that be cool?

I just typed into the search box “Is Kashi a company?” because I was in a Siri kind of mood, and of course I didn’t get a good answer. But maybe Siri on my iPhone would have known. (Update: I just tried it, and it didn’t but still, I think it’s a cool idea!)

A lot has been written about Siri (the new voice activated assistant in the iPhone 4s). I have to say that it works surprisingly well. It has been eerily accurate in properly transcribing my voice, although not quite as perfect in understanding what I am saying.

I just had a 3 message dialog with someone over text messaging using Siri. I never ever would have done that with texting previously but here I went speaking in complete sentences over text messaging and with 100% accurate recognition. I don’t know if the other guy was using Siri but if he wasn’t he must have been annoyed by how quickly I responded with a full sentence 🙂

It occurs to me that the average length of text messages might become longer and longer with the advent of services like this. I wonder if that is at least part of the reason that Apple is now bypassing the mobile operators when doing iPhone to iPhone messaging.

A useful article in the New York Times today, just in time for the holidays. Anew site that tries to forecast the best time to buy that item you really want.It seems to draw a graph showing how the price of, for example, a Nikon 50-300mm Zoom Lens has varied over the last 3 months and whether I should buy it today or wait till next month. It’s a very nicely done site and if it works, it could be very useful. The site is decide.com.

“Oren Etzioni writes articles about artificial intelligence for scholarly journals, is a renowned expert on data mining and gained fame when Micro...

An interesting article about how to think about designing Artificial Intelligence for playing poker. It’s high level but gives you a good feel about how this writer went about the design. You also learn the basics of Poker at the same time:

“It is very important in this type of project, for any AI project really, to fully understand the rules of the game and the tactics typically used to win. So when I worked on poker I spent a fair amount of time reading the best books and playing the game, both against existing AI and real humans of various levels of experience.” (from Poker AI: A Starting Point)

And here’s a follow up article on the same topic: “Aces High: Numerical Techniques in Poker AI“

You may not have followed the discussion that ensued when Ray Kurzweil was reported to ‘not understand the brain‘, but it’s pretty fascinating. Here’s what PZ Meyers said: “There he goes again, making up nonsense and making ridiculous claims that have no relationship to reality. Ray Kurzweil must be able to spin out a good line of bafflegab, because he seems to have the tech media convinced that he’s a genius, when he’s actually just another Deepak Chopra for the computer science cognoscenti.” (from PZ eyers)

Ouch!

Here’s a ty...

Just came across this odd link somewhere. He’s a good and clear writer, enjoyable to read whether you agree or not: Why People Think Computers Can’t by Marvin Minsky.

Went to hear Ray Kurzweil yesterday at MIT, at an event jointly sponsored by IEEE and GBC/ACM Robotics Sigs. The blurb said: “[…] With input from people around the world, an international group of leading technological thinkers were asked to identify the Grand Challenges for Engineering in the 21st Century http://www.engineeringchallenges.org/. We’ve invited one of the leading authors of the report, Ray Kurzweil, to present some of its findings, give his impressions of the important technological trends and challenges likely to occur over the next hundred years or so, and challenge you, some of...

I knew about this site but recently came across a link to another quite interesting article, “Why We Can be Confident of Turing Test Capability Within a Quarter of a Century”.

Yes, Kurzweil’s books are too long and repetitive, and he does hold some unusual views, but the guy’s a total genius and always fun to read. I don’t think the article delivers on the subtitle:

“The advent of strong AI (exceeding human intelligence) is the most important transformation this century will see, and it will happen within 25 years, says Ray Kurzweil […]” (from ...

This one was probably one of the most talked about sites. The principle is, Flickr is nice, but no one really wants to take the time to tag. So instead, with Riya, after you upload your images, you can search for them based on who is in them and where they were taken. But the trick is that it uses image recognition to help you locate the pictures. Pretty impressive! Check out Riya.