In the beginning...
Everything has a beginning, today it's the publishing of this newsletter!
For those of you that know me…
Your support at this time of change and opportunity, is greatly appreciated.
Thank you for your support and encouragement!
For those of you that will get to know me…
I’m overjoyed that people I may not even know are excited enough to take an interest in what I have to share.
The road ahead is totally unknown, but with good friends, clear thinking, and a great deal of passion about technology and making things works, I know this is going to be an incredible experience.
I’m launching this newsletter because I’ve spent the last several decades learning about technology and there is now so much more to learn. In fact, that’s exactly why I love working in technology. There is a never ending flow of new things to experience, play with, learn about and share with others.
Starting this newsletter has a central role in the opportunity for shared experiences and learning together. I have been increasingly feeling that I have much more to share, and what better way than to create a system that solves this problem. Creating this newsletter will be a great opportunity to share my experiences and to connect with others all over the world that share my passion. I am genuinely excited about using technology to improve our daily lives, and this is where the rubber meets to road for me.
I’ve started this and will let it take me where it may go. This could be my next great big deal or just wind up as a great way to learn and share in a more open way, so I’m excited that you have decided to join me.
I have big plans ahead to build value and in return I’d like to hear from you when you see something great AND when you see something that’s not great. My goal is to provide value and maximize value I want to ask for your feedback and participation. We all have something of value to share. Let’s find it together!
Let’s have some fun and learn something!
There is so much happening today. Each day brings a new discovery, tool, insight, benchmark or application. There are plenty of newsletters out there that help identify the latest news. I’ll be including news items that I see are particularly relevant, but intend to take a more intentional approach. You can expect this newsletter to happen on a monthly basis. I suppose it’s possible that I will post more frequently, but I intend to stay above the noise and provide something beyond the rapid fire headlines in the newsletter.
You can expect the content to include: news, prediction and analysis, educational content, a bit of code and some nice relevant bonus content in each edition.
Substack has many features beyond publishing the newsletter, so you might consider popping into the site to see what else is available.
What to Expect
I also want to be clear that I have turned on the subscription option, and I would love for you to support me, but I will always have something valuable in the free content. If you are ready to sign-up for the paid subscription now, there’s a great surprise at the end of this very first newsletter!
I will be publishing the newsletter at least once a month. This means there will be something I share every month, but it won’t necessarily be on certain day of the month. If an opportunity presents and the words start to flow, I will release more than once a month to the paid subscribers. The monthly cadence is to ensure I am leaving space to get above the day-to-day frantic news cycle that everyone is experiencing.
This is a marathon, not a sprint. I want to bring clarity of thought and be thoughtful in how we are understanding this constantly evolving opportunity. I will be constantly experimenting and making mistakes all the time. So, if you sign-up for a paid subscription, you will get even more value with paid only editions and it will continue to grow as I work hard to make this well worth your investment.
What about the content?
Each edition will include headlines, history, code and architecture or examples. As time passes I’m sure this will be refined and perhaps even expanded. I have both interest and experience that touches each of these content areas and these are well within my skillset to write about and discuss with all of you.
Feedback
Clarity of purpose is essential to success. I want to help you and I want to build something that creates value. Let me know what you are thinking as you read and let me know if you seen an opportunity for me to improve or provide more value.
Thanks for Subscribing
Thank you seeing the value and for subscribing. You can help me by sharing it with your friends and colleagues.
Now let’s get to the why you’re actually here!
In this edition:
Notable headlines
The Turing Test (a useful mistake)
What’s happening in hardware
Code and installations to start working
Architecture and how to start building AI with your own data
Notable headlines
KPMG: US executives unprepared for immediate adoption of AI
Almost 2/3 US execs surveyed shared that Generative AI will have a major impact on their organization and almost the same amount say they are up to 2 years away from implementing their first generative AI solution.
New MIT/Stanford Study Shows Generative AI Boosts worker productivity
Generative AI tools have the potential to improve productivity and happiness of workers and could also improve customer sentient at the same time.
Generative AI is changing your technology career path
A rise is automation and applications of generative AI will have an impact on many jobs. The upshot of the growing use of AI is on the horizon for all IT professionals. Every developer, product manager, and designer should be thinking about how to incorporate AI technology into their products.
The Turing Test (a useful mistake)
Alan Turing is widely regarded as the father of theoretical computer science and artificial intelligence. Although his work laid the foundation for the development of computing machines, he did make some mistakes along the way. I had the opportunity to take in his biography on Audible and if you’re into math and the early days of computing, I would wholeheartedly recommend. You’ll definitely learn something!
Turing imagined a test as a definitive test for artificial intelligence. The Turing Test, proposed by Turing in his 1950 paper "Computing Machinery and Intelligence," is a test of a machine's ability to exhibit intelligent behavior indistinguishable from that of a human.
The test involves an evaluator who interacts with two entities—a human and a machine—through a text-based interface, without knowing which entity is the human and which is the machine. If the evaluator is unable to reliably tell the difference between the human and the machine, the machine is said to have passed the Turing Test.
While the Turing Test has historical significance and sparked early discussions around artificial intelligence, it has since been criticized for several reasons:
Oversimplification: The Turing Test reduces the complex concept of intelligence to a single criterion—imitating human-like conversation. Intelligence, however, is multifaceted, encompassing problem-solving, creativity, emotional intelligence, and more. The test does not account for these other aspects of intelligence.
Deception: A machine could be designed to deceive the evaluator by imitating human-like errors or purposely making mistakes, without necessarily demonstrating true intelligence. Conversely, an intelligent machine might fail the test simply because it does not perfectly mimic human behavior.
Anthropocentrism: The Turing Test assumes that human-like behavior is the ultimate benchmark for intelligence, which may not be accurate. There could be forms of intelligence that do not resemble human intelligence at all, and focusing solely on human-like behavior could be limiting.
These criticisms are driven out of the increasingly complex systems that we have been able to build since the time he imagined a way to conduct a meaningful or useful test of a computer’s ability to communicate in a meaningful way. The truth is, that we needed a useful measure and this test has provided that metric for over 5 decades. It encouraged researchers to consider the possibility of machine intelligence and stimulated philosophical debates about the nature of intelligence, consciousness, and the ethical implications of creating intelligent machines.
However, it is important to recognize that the Turing Test is not a definitive or comprehensive test for artificial intelligence, and the field has since evolved to consider a wider range of factors in evaluating machine intelligence. We have devised more critical and detailed ways to examine machine intelligence, and we can now even use tests that were devised to evaluate human intelligence and mastery for determining the aptitude of AI models.
What’s happening in hardware
This space continues to develop and great articles are pushed out and then become outdated because of new announcements. This page provides some great details on AI chips.
Top 10 AI Chip Makers of 2023: In-depth Guide
Since this guide was produced, NVIDIA made some announcements at GTC (GPU Technology Conference) in March. NVIDIA announced 3 AI relevant developments:
DGX Cloud: a service that allows enterprises to access Nvidia’s AI supercomputing infrastructure and software through a web browser
AI Foundations: provides three cloud services: Nvidia NeMo for large language models (LLMs), Nvidia Picasso for image, video and 3D applications, and BioNeMO to generate scientific texts based on biological data
Four inference platforms: Nvidia L4 for producing AI video; Nvidia L40 for 2D/3D image generation; Nvidia H100 NVL for deploying large language models; and Nvidia Grace Hopper for recommendation systems built on giant datasets
Microsoft, Google, Apple, Baidu, Tesla and Meta all have their own AI hardware development efforts, and this is not a secret. They've been working on this for years. AMD is also rumored to be working on updates that incorporate AI but for now, NVIDIA has the hardware and they are running workloads in all of the hyperscale providers, so right now it's their world that everyone is playing in.
Experiment with Code!
Auto-GPT is an experimental open-source application showcasing the capabilities of the GPT-4 language model. This program, driven by GPT-4, chains together LLM "thoughts", to autonomously achieve whatever goal you set. As one of the first examples of GPT-4 running fully autonomously, Auto-GPT pushes the boundaries of what is possible with AI.
This Python script is an example of an AI-powered task management system. The system uses OpenAI and Chroma to create, prioritize, and execute tasks. The main idea behind this system is that it creates tasks based on the result of previous tasks and a predefined objective. The script then uses OpenAI's natural language processing (NLP) capabilities to create new tasks based on the objective, and Chroma to store and retrieve task results for context. This is a pared-down version of the original Task-Driven Autonomous Agent (Mar 28, 2023).
This repository is intended as a minimal, hackable and readable example to load LLaMA (arXiv) models and run inference. LLaMA was released by Meta. The below repos are connected with the development of LLaMA and because it was open sourced it is the roots of many LLMs that can run on a PC.
Demo, data, and code to train open-source assistant-style large language model based on GPT-J and LLaMa. GPT4all is a promising open-source project that has been trained on a massive dataset of text, including data distilled from GPT-3.5-Turbo.
LlamaIndex (GPT Index) is a project that provides a central interface to connect your LLM's with external data. This code allows you to begin considering how to start building AI with your own data.
Architecture
So you want to start thinking about how to build your own datasets and using AI with them? Matt Boegner has some great things to share, he’s mapped out some architecture to help us understand how to put the pieces together.
https://mattboegner.com/knowledge-retrieval-architecture-for-llms/
Keep reading with a 7-day free trial
Subscribe to AI MISTAKES to keep reading this post and get 7 days of free access to the full post archives.