Welcome to Don's Thoughts: Using Artificial Intelligence Tools

As part of my Family Tree research in 2025 I created this page to share today's capabilities of the Global Artificial Intelligence (AI), which is accelerating at an exponential pace…

Questions, thoughts & ideas are shared here; consider viewing this YouTube commentary about AI for “The Painful Truth about AI & Robotics” sharing some background, as well as this discussion with the “Godfather of AI,” Geoffrey Hinton, who shares why he thinks AI will cause the fact that "Humans will no longer be needed” in the very near future…

Moving back a few years from today, late 1951 brought an early AI-equivalent computing machine, the UNIVAC to us ( just ~3½ years before my birth )…

Named after the concept of a “UNIVersal Automatic Computer,” it began crunching AI-related data, directed by sets of punched cards—it performed quite slowly at just 1,905 operations per second vs. today's (as of this writing)
"El Capitan" supercomputer at the Lawrence Livermore National Laboratory that has a peak performance of 2.746 exaflops per second

There were 6,103 vacuum tubes stacked in the UNIVAC, a heavily-weighted structure of 8.3 tons (16,686 pounds), running on a 2.25 MHz clock—fast for that time! The Central Complex (i.e. the processor & it's memory unit) was 4.3 meters deep by 2.4m wide and was 2.6m high; the complete system occupied more than 35.5 square meters of floor space

Earlier computer systems such as the ENIAC did not yet do anything related to AI, just processed data

Today's AI falls into 3 categories: Generative, Predictive & AGI…

Generative tools, the“chatbots,” have become more common since the AI boom in the early 2020's.
These (and the Predictive Systems, below) process specific tasks, working for narrowly-defined domains & excelling at tasks within their trained domains… they do however appear to lack a “true understanding” of how to generalize new situations—examples include self-driving cars, predictive modeling, & factory AI tools

Many improvements to the transformer-based deep-neural-networks, particularly with Large Language Models (LLMs) allows them to be extremely efficient in their responses; view a few below?
Predictive systems today are not running well at all for our future… view an AI commentary that shares the ideas of “AI Snake Oil: What AI Can & Can’t Do, & How to Tell the Difference” (drag video back to start?)
An AGI intelligence will begin to appear (IMHO) within 1,000 days of this writing… It will perform **any** intellectual task, possessing human-like cognitive abilities such as reasoning, learning, understanding, & problem-solving across various domains, bringing each of us to one of these 3 possible paths:
      1st A Good Path, which I do **surely** hope will be the case,
      2nd Indifference (e.g. nothing new), which is possible yet really less likely, or
      3rd The Very Bad Path which I sadly anticipate…

I asked several general AI apps to generate a 2-page report based on my Profile page with this question:
  “Can you please write a publishable 2-page essay from this website?“

Some did not respond very well at all but these did really well!
Gemini The best report created!
View the converted Google response & the report ?
ChatGPT 2nd is another good one… view the PDF here & the report ?
Groc Pretty Good! too View the PDF & the report ?
DeepSeek The most creative Chinese report, with interesting additions!
View it here & the report ?
coPilot Finally, another good report, albeit a tad tweaky! View the report—I didn’t create a PDF here…

The classic 1984 movie, The Terminator shows how disastrous a mechanical AGI device would be—today's 8+billion people wish for this to not happen, yet some apparently do, very sadly…

Many AI-enhanced devices exist today to help our everyday lives, such as Elon Musk's Tesla cars, trucks & his human-like robots, the multitude of general manufacturing devices, Amazon's Alexa devices (where I own 3 units for news, music, alerts, etc) & the upcoming hospital surgery devices with many, many more

As a brief look-back into their history, mechanical machines have existed for many centuries without computing power, etc such as Archimedes' water control screw from around 300 BC & one of the many things created by Leonardo da Vinci, the anamorphosis tool around 1530 for his "angle of contingence" study, which refers to the angle at which a ray of light reflects off a cylindrical curved mirror, a technique where a distorted image appears normal when viewed from a specific angle

The concepts of AI have been around for a while, starting out with Alan Turing publishing Computer Machinery & Intelligence in 1950 at the Manchester University in England that describes the following thoughts…
    1) To pass the Turing Test, a machine has to be able to hold a natural language dialogue with a person, & know
         enough to not make a fool of itself while doing so! Also, view the test on the WiKi page?

    2) For it to pass as a human it has to exhibit dozens of incorrect replies that are predictable reasoning,
         what is known as Translogical Reasoning

    3) Yet, is it desirable to build such foibles into our programs?
         In short, we need to unravel several issues that are often tangled up together…
                a) How could we get a machine to pass the Turing Test?
                b) What should we get the machine to do (or not do)?
                c) And others like that…

I caught a YouTube discussion from April, 2024 a few months back with Sir Demis Hassabis, the CEO of Google's DeepMind Technologies describing how fast AGI has evolved over the last 20 years…

IBM's Deep Blue computing code was developed in the early 1990's & did something that no machine had done before… in February, 1996 it took on the reigning global Chess champion, Garry Kasparov (who is a Russian) in Philadelphia, USA. Deep Blue won the first game, which marked the first victory by a computer against a reigning world champion under regular time controls. But Kasparov recovered & won the match 4–2

With the eyes of the world media upon them, an updated Deep Blue program again squared off against Kasparov in May of 1997, this time at the Equitable Center in New York. The rematch was widely seen as a symbolic test of whether supercomputers were catching up to human intelligence… Kasparov won the first game, Deep Blue took the next. The following three games were played to a draw but Deep Blue prevailed in the tension-filled Game 6, thereby achieving a resounding victory, 3.5–2.5, in the rematch & becoming the first computer system to defeat a reigning world champion under standard tournament time controls! I do seem to remember watching that match on TV, with the younger Demis cheering loudly… that may not be the case, as I failed to find a link to share !!

Sir Demis worked in the computer development world of Chess (as did Alan Turing) & he went on to lead the deployment of the **stunning** AI success in 2016 of winning at the game Go, as well as releasing to the world in 2024 the results of his team of scientists who predicted (with AI) the structures of almost all known proteins after creating the AlphaFold2 tool; this is also shown on WiKi

My curiosity for this techno-world has taken me to several apps on my PC (& on my iPhone/iPad) that I've broken down above… While none of these apps are particularly expensive, OpenAI is the one I've preferred & am using today with the free version. A page sharing some of these links can also be viewed here: AI web pages?

Here's a breakdown of each key area of AI with resources & explanations from OpenAI so you can dive deeper into AI devices like ChatGPT & SearchGPT; some links are sadly stale & will open in error…

1. AI Development: Understanding how models like ChatGPT are built, trained, & improved
Key concepts to learn (copy to your browser to view?)
    • Transformer Architecture—The backbone of modern AI (like GPT-4)
    • Pre-training vs. Fine-tuning—How AI learns from massive datasets & adapts to tasks
    • Neural Networks & Deep Learning—The fundamentals of AI model training
    • Reinforcement Learning from Human Feedback (RLHF)—How models like ChatGPT improve responses
Resources
> Read
    • "Attention Is All You Need" - A Transformer Paper
    • OpenAI's GPT-4 Technical Report
    • Hugging Face Course - Deep Learning Basics
> Watch
    • DeepLearning.AI YouTube - An Intro to AI & LLMs
    • Jay Alammar's Visual Guide to Transformers
> Try
    • Google Colab—Free AI Model Testing
    • OpenAI API—Experiment with AI Models
    • Hugging Face—Play with Open-Source Models

2. Practical Applications: Exploring how AI like ChatGPT & SearchGPT are used in real-world scenarios
Common Applications (copy to your browser to view?)
    • Search & Information Retrieval—SearchGPT enhances web browsing by integrating AI
    • Chatbots & Virtual Assistants—AI helps in customer service (e.g., ChatGPT, Google Gemini)
    • Content Creation—AI writes articles, generates code, & creates images (DALL·E, Midjourney)
    • Automation & Productivity—AI automates emails, summarization, & data analysis
Resources
    • How AI Powers Google Search - Using BERT & MUM
    • Microsoft's Bing AI Overview
    • AI for Business—Harvard Business Review
> Watch
    • AI in Search (Google DeepMind)
    • How Chatbots Work (Computerphile)
> Try
    • ChatGPT Plugins & SearchGPT
    • Google Bard (AI-Powered Search)
    • AI Tools for Writing & Coding

3. Ethical Considerations: AI comes with challenges related to fairness, bias, misinformation, & responsible use
Key Issues (copy to your browser to view?)
    • Bias & Fairness—AI can reinforce biases from training data
    • Misinformation—ChatGPT can generate inaccurate or misleading content
    • Privacy & Security—AI must protect user data & prevent misuse
    • Job Automation—AI is changing industries, raising concerns about job displacement
Resources
> Read
    • AI Ethics Guidelines—OpenAI
    • The Alignment Problem—Brian Christian (Book)
    • AI Bias in Large Models—MIT Technology Review
> Watch
    • AI Bias & Ethics Explained (Veritasium)
    • The Risks of AI (Lex Fridman & Sam Altman)
> Explore
    • AI Fairness Toolkit (Google)
    • AI Alignment Research (OpenAI)

Here's another list to help deepen your understanding of AI devices
1. Beginner-Friendly AI Overviews
YouTube Channels:
    • Two Minute Papers—Short AI research summaries
    • ColdFusion—AI history & impact
    • Fireship—Fast, engaging AI explanations
Articles & Blogs:
    • OpenAI Blog—Research updates & new AI advancements
    • DeepLearning.AI—Weekly AI newsletter
    • Google AI Blog—Insights from Google's AI team

2. Understanding How AI Works
Courses:
    • Elements of AI—Free introduction to AI concepts
    • Fast.ai—Practical deep learning with real-world examples
    • Coursera—AI For Everyone—Simplified AI concepts by Andrew Ng
Books: (copy to your browser to view?)
    • Artificial Intelligence: A Guide for Thinking Humans—Melanie Mitchell (Great for non-experts)
    • You Look Like a Thing & I Love You—Janelle Shane (Fun, humorous AI explanations)
    • The Alignment Problem—Brian Christian (Ethical challenges in AI)

3. Experimenting with AI
Try Different AI Models:
    • OpenAI's ChatGPT
    • Google's Gemini
    • Anthropic's Claude
Hands-on AI Tools:
    • OpenAI API—Experiment with AI-generated text
    • Hugging Face—Try AI models for different tasks
    • Kaggle—Free datasets & coding exercises

4. Staying Up-to-Date
Podcasts:
    • The AI Alignment Podcast—Discusses AI's future & safety
    • Hard Fork—Covers AI's impact on society
    • Lex Fridman—AI discussions with top researchers

Finally, another list sharing the best AI apps that depends on what you want to do, such as writing, creating images, or improving your productivity [1]

Here are some of the top AI apps (copy to your browser to view?)…
  ChatGPT: A popular chatbot that can help with writing, coding, & customer care [2, 3]
  Grammarly: Improves writing by checking grammar, suggesting style, & adjusting tone [2]
  starryai: Generates art from text descriptions [2]
  ELSA Speak: Helps with language learning & pronunciation [2]
  Google Assistant: A virtual assistant that can perform internet searches & control other devices [4]
  Runway: An AI video generator that creates videos based on prompts [5]
  ElevenLabs: An AI voice generator that offers many voice options & control over voice style & tone [5]
  Rytr: An AI writing generator that offers affordable upgrades for AI content [5]
  Arc Search: An AI search tool that summarizes source content into structured answers [6]
  Databutton: An AI app builder that offers transparent AI thought processes & the ability to create multiple conversation threads [7]

Generative AI is experimental, links are [1] the 50 best AI tools / [2] 7 of the best AI apps / [3] top 10 AI apps / [4] 33 AI apps / [5] best free AI tools / [6] best AI productivity tools / [7] 6 best AI app builders

Other AI apps include:
  Synthesia: An AI tool for video generation & editing
  Fathom: A notetaker & meeting assistant
  Deep Research: An AI tool for research
  Midjourney: An AI tool for image generation
  Canva Magic Studio: An AI tool for graphic design

As a close for my AI page, a guy named Doug Lenat (who died in 2023, R.I.P.) was one the leading computer science & AI experts in the world at that time who created the CYC common sense knowledge base & inference system

He was working before his death with Bill Gates to establish the Microsoft Research Labs & was the only person to have served on the technical advisory boards of both Apple & Microsoft. He also received the bi-annual IJCAI Computers & Thought Award (scroll down for details?) in 1997—the highest honor in Artificial Intelligence & was named the First Fellow of the American Association of Artificial Intelligence (AAAI) & named as a Fellow of the American Association for the Advancement of Science (AAAS)

Feel free to share your thoughts, comments or criticisms etc via eMail?
Also feel free to visit my Website structure page to view the many pages I created?
Cheers!
Regards, Don Shave

Some relevant advisories:
    (1) My data is All Rights Reserved © April/1988 – date Don Shave
    (2) My Copyrighted (©) web pages do not hold any Fake data, nor are they using any AI enhancements
    (3) As legal issues can surround such things as my published advice & opinions,
          my Privacy Policy & Disclaimers page is a mandatory addition… please review?

Created in March, 2025 & last updated in June, 2025
This website operates best under the Chrome browser; other browsers & devices fail to display correctly