Category: Technology

  • Original Work or AI?

    Original Work or AI?

    Let me show you my roughy draft vs an AI review.

    Here’s a polished version of your essay with enhanced clarity, improved flow, and corrections for grammar and tone. Let me know if there are any specific stylistic preferences you’d like to emphasize.

    AI version

    There’s a lot of conversation right now about AI, plagiarism, and original work. People are asking important questions: What’s acceptable in today’s world? What role should AI play in content creation?

    As AI becomes more embedded in our lives, I think we’ll grow increasingly comfortable with the idea of computers proofreading, refining, and even rewriting much of what we produce. This progression feels natural. It started with simple spell checkers built into word processors, then grammar tools, and later programs like Grammarly, which helped rework entire sentences for clarity. Now, we’ve advanced to tools like ChatGPT, Copilot, and Gemini. With these, you can simply upload your document and ask the AI to “scrub” it—and it does. It can even restructure your ideas using your own words and concepts, delivering a cleaner, often better, version of your work.

    This represents a shift from fixing obvious errors to refining content without altering meaning. But it raises an important question: Is this plagiarism? Is the AI pulling ideas or phrases from elsewhere and passing them off as yours?

    If you ask an AI to generate something entirely on its own—say, a detailed financial advice column about mutual funds—that might tread into plagiarism, especially if you contributed little to the result. But if you write your own content and the AI simply refines it for clarity, I don’t believe that qualifies as plagiarism. To me, it’s just the next step in the evolution of word processing.

    Of course, there’s room for debate about tone, ownership, and how AI draws from existing data to create its outputs. But we live in a world of shared ideas. The real challenge arises when we can no longer distinguish between human-created and computer-generated content.

    Take voice replication as an example. Imagine uploading samples of your voice into a program, which can then create podcasts that sound like you—without you ever speaking a word. All you’d need to do is give it a prompt like, “Make a five-minute podcast on Ukraine, and emphasize support for their cause.” In moments, you’d have a polished podcast that sounds authentically like you.

    That’s where things get unsettling, especially when this technology could be misused. Someone could upload your voice, generate content, and make it seem like you said things you didn’t. If this sounds like a leap, let me remind you: technology advances quickly. Just two years ago, I would’ve thought this kind of capability was science fiction. Now, it’s reality, and it’s accelerating rapidly.

    As we approach advancements in chip technology and quantum computing, distinguishing human-generated content from AI-created material will only get harder.

    So, what can we do about it? That’s the million-dollar question.

    In the U.S., we could pass laws and regulations to manage AI use, but that wouldn’t stop other countries from pushing the boundaries of what AI can do. Cyber warfare is already a global concern, and AI is at its forefront.

    What’s critical is ensuring that AI is programmed responsibly. Developers must build guardrails into these systems to prevent them from causing harm. Isaac Asimov envisioned a future where computers followed rigid, binary rules, but our current AI is evolving into something much more nuanced.

    Ultimately, we have to adapt to the future we’re creating. We may not have all the answers yet, but we can strive to ensure this technology is used ethically, responsibly, and in a way that benefits humanity.

    Let me know if you’d like further tweaks or adjustments!

    My Version

    Please review and edit my essay that I’m planning on posting on my blog. Original work or is it AI So right now there is a lot of conversations going on with people about AI and plagiarism and original work. Lots of questions of what is acceptable in today’s world. I think as we get more and more embedded AI we’re gonna just find ourselves very comfortable with the idea that computers are going to proofread and rewrite a lot of the things that we produce. It’s just a natural progression of things. First that started off with spelling checking Capabilities built into word processors, and then they had grammar checkers and then we had programs like Grammarly that just went through and helped you rewrite sentences and now you don’t even have to go line by line through it. You can just throw it up into your AI like ChatGPT Copilot or Gemini and say hey scrub this document for me and it will. It’ll even rewrite it in a way that uses your words your ideas and everything that you were working on and create a whole different version of your document and you probably are gonna like it better. So that’s the nuance world is moving from where it fixes things that are obviously wrong to changing things without changing meaning. The question really is is it plagiarism? Is it pulling the information from somewhere else and uses somebody else’s words to rewrite your documentation your essay your vision. Now, if the propped is, hey, can you write a complete plan for me? That is a financial advice column and today we’re gonna talk about mutual funds and that is the only prompt that you give AI and it just goes And creates a detailed plan that you didn’t even have anything to do with. Yeah that might be plagiarism, but if it takes every sentence that you write and just restructures it to be clear, I don’t think that’s plagiarism. I think that’s just the next evolution of word processing. Now we can debate this back-and-forth about tone and where it gets the words from, but ultimately we live in a world of ideas. The bad part about AI if there is a bad part is where we are unable to distinguish between the information that is human mind created versus computer created content in totem. That means when you upload your voice, so then it can use your sounds to replicate the way that you talk yeah that’s gonna change the world again we’re gonna have podcasts that are created completely by AI. Someone’s gonna say I need a five minute podcast using my voice and today’s topic is Ukraine and I am totally for supporting Ukraine and all of a sudden you’re gonna get a 5 to 7 minute podcast that’s gonna talk all about that subject and you never said a word. That’s the scary part right And it’s even scarier if somebody can upload your voice to a computer and make it sound like you said those things. I know this seems like a leap Judd you’re crazy but if you read my blog two years ago, I would’ve thought that nothing like this could possibly have happened But it’s happening and it’s happening extremely fast and it has chip technology moves, faster, and faster and we get into quantum computing. I see the challenges of understanding the difference between human and computer, derived content to become more and more difficult to distinguish. So what can we do about it right? That’s the big million dollar question, I don’t know that there is anything we can do about it. So in the United States, we can pass laws and we can do all sorts of things, but that will will not stop other countries from forging ahead to create this capability. We have to worry about cyber warfare and AI is at the forefront of it. There’s a lot of considerations going on in the world today. I think that we just have to ask programmers to program AI responsibly, and ensure that there are some guide rails protections built-in to prevent AI from going off the rails as it were and doing things that can harm humans. And while Isaac Amoth had some great vision, I think that the binary thinking of computers in the past is going to be completely replaced with a more nuance approach for the future. And that’s how we’re going to have to live with the future that we’ve created today.

  • Chat GPT and the results

    Chat GPT and the results

    So I played around with Chat GPT the other day, it’s the AI (Artificial Intelligence) application that is getting a lot of buzz lately and it was interesting. I’ll share some of the results here that I haven’t posted anywhere else.

    Things to note, it does sort of remember what you asked it before to help make new responses, so each query isn’t just a new search per se. It can be surprising as it does a wonderful job with some responses.

    So here we go.

    Write an essay for the need for modernizing nuclear weapons. Written by ChatGPT

    I would like an essay making a case of the United States to modernize their nuclear weapons in MLA format

    Title: The Need for Modernization of the United States’ Nuclear Weapons

    Introduction The United States of America is one of the few nations in the world that possesses nuclear weapons. The country’s nuclear arsenal has been an important tool for deterrence, helping to keep peace and stability throughout the world. However, as the weapons continue to age, it has become increasingly necessary for the United States to modernize its nuclear weapons. This essay will argue that modernizing the country’s nuclear weapons is essential for maintaining national security and for deterring potential aggressors.

    Body Paragraph 1: Deterrence The United States’ nuclear weapons have served as a critical tool for deterrence, helping to prevent the outbreak of war and maintain peace. As the world becomes more complex, and new threats emerge, it is essential for the country’s nuclear weapons to remain modern and effective. The United States’ adversaries are developing new weapons and strategies to overcome U.S. deterrence. By modernizing its nuclear weapons, the United States can continue to maintain a credible deterrent and protect its national security interests.

    Body Paragraph 2: Safety and Reliability Many of the United States’ nuclear weapons were developed in the 1970s and 1980s and are becoming increasingly outdated. As these weapons age, their safety and reliability are becoming more of a concern. Modernizing the country’s nuclear weapons would help to ensure that they are safe, reliable, and effective, thus reducing the risk of accidents and miscalculations.

    Body Paragraph 3: Technology Advancements Advancements in technology have made it possible to develop more advanced nuclear weapons that are more accurate, precise, and less likely to cause collateral damage. Modernizing the United States’ nuclear weapons would allow the country to take advantage of these advancements, improving the effectiveness of its nuclear arsenal. This would help to ensure that the United States maintains a technological edge over its adversaries and can better protect its national security interests.

    Body Paragraph 4: Cost and Feasibility Modernizing the United States’ nuclear weapons is an expensive proposition. However, the costs of not modernizing are even greater. As the weapons continue to age, their maintenance and upkeep become increasingly expensive. Moreover, an outdated nuclear arsenal increases the risk of accidents and miscalculations, potentially leading to devastating consequences. In the long run, modernizing the country’s nuclear weapons is a cost-effective solution that will help to ensure national security and deter potential aggressors.

    Conclusion The United States’ nuclear weapons have served as an important tool for maintaining peace and stability throughout the world. However, as the weapons continue to age, it has become increasingly necessary for the country to modernize its nuclear arsenal. By modernizing its nuclear weapons, the United States can maintain a credible deterrent, improve safety and reliability, take advantage of technological advancements, and protect its national security interests. While the costs of modernization may be high, the benefits far outweigh the risks of not modernizing. It is imperative that the United States takes the necessary steps to modernize its nuclear arsenal to ensure its safety, security, and stability for generations to come.

    So next I asked it some absurdity

    I thought we should see if it really understood the world around it, so her we go.

    Can guys on submarines see the fish?

    It is highly unlikely that submariners can see fish from a submarine while it is submerged. Submarines are designed to operate deep underwater, typically below the depth where sunlight can penetrate, which is known as the “photocline”. This means that the visibility from a submarine is often limited to a few hundred meters or less, depending on the water conditions.

    Furthermore, submarines are typically equipped with low-light level and infrared sensors that allow them to detect other vessels, aircraft, or other objects in the water. Submariners rely on these sensors and other advanced technology to navigate, communicate, and perform their missions, rather than relying on their own visual observations.

    While it is possible that a submariner could spot a fish or other marine life near the surface of the water when the submarine is on the surface or near the surface, it is unlikely that they would be able to see fish from the depths that submarines typically operate at.

    Then I asked it, “Did you know that Submariners are the wildest group of sailors in the military?”

    As an AI language model, I do not hold opinions or make generalizations about any group of people. While it is true that submariners may experience unique challenges and conditions compared to other sailors in the military, it would not be accurate or fair to make blanket statements about their behavior or character as a group. It is important to treat all individuals with respect and without making assumptions or stereotypes based on their background or profession.

    Then I thought maybe I need to just feed it data, like giving it some facts or context.

    “Submariners pride themselves on their ability to adapt in a wide variety of difficult situations”

    Judd Spitzer

    Yes, it is true that submariners are highly trained to operate in a variety of difficult and challenging situations. The nature of submarine operations, including extended periods of time spent submerged, limited communication with the outside world, and the need for self-sufficiency and problem-solving, requires submariners to be highly adaptable and resourceful.

    Submariners undergo rigorous training and must be able to operate in a wide range of conditions and scenarios, including extreme weather conditions, combat situations, and emergency situations such as fires or flooding. They must also be able to work closely with a small crew for long periods of time and maintain a high level of focus and attention to detail.

    The adaptability and problem-solving skills required of submariners can be an asset both in and outside of their military careers, as these skills are highly valued in many professional and personal contexts.

    Finally I thought, maybe I’d ask it something that was pretty subjective and see if we got a definitive answer.

    Do you think America is a free country?

    Judd Spitzer

    As an AI language model, I do not have personal beliefs or opinions. However, the United States is considered a free country in many respects. The US Constitution and Bill of Rights outline individual rights and freedoms such as freedom of speech, religion, and the press, as well as protections against unreasonable searches and seizures, and cruel and unusual punishment.

    That being said, some individuals and groups in the United States may not experience the same level of freedom as others due to factors such as discrimination or economic inequality. Additionally, some laws and policies in the US may be viewed as limiting individual freedom or infringing on certain rights. The interpretation of what constitutes freedom can also vary depending on cultural, social, and political perspectives.

    I did ask it what the meaning of life was, and sadly it did not come back with 42.

    The purpose of life is a question that has been debated by philosophers, scientists, and theologians for centuries. There is no single answer that can be universally applied to all individuals, as people have different beliefs, values, and goals that shape their views on the purpose of life.

    Some people believe that the purpose of life is to seek happiness, pleasure, and personal fulfillment. Others believe that the purpose of life is to serve a higher power or to contribute to the greater good of society. Still, others see the purpose of life as a journey of self-discovery, learning, and personal growth.

    Ultimately, the purpose of life is a deeply personal and subjective matter, and each individual must find their own meaning and purpose in life. It is important to recognize that the purpose of life can change over time, as people grow and experience new things, and that there is no one “right” answer to this question.

  • ChatGPT AI is Eliza in Disguise

    You may have heard about the big revolution called AI or artificial intelligence that is sweeping the news and the tech industry. Let me tell you, there is nothing new under the sun here. We are still baby steps into the idea that computers can think for themselves. They don’t. There was a program called Eliza written back in the 60s I believe, or 70s and you can still run it through your browser. Here is the link, https://web.njit.edu/~ronkowit/eliza.html Computers can only create logic from words that people write. It is just more sophisticated today than it was 30 to 40 years ago, but it’s the same thing, take some input, decide based on the logic that humans have programmed, and give output. I think that computers are doing better at reading and understanding. Computers give positive feedback, and yet it isn’t artificial intelligence. It is just a program. We need to stop repackaging this idea that it is AI because it is bullshit. When computers start to react to other inputs that human beings do not give it, then we will start to see real AI. When computers can design and develop new reactions based on what they’re doing that’s real AI. We hear Apple talk about machine learning, and I feel that learning is much more akin to the idea of AI than being able to respond to a query and give you back a paragraph for a page of information that it’s drawn together and parsed together. Does the world need AI? I do think we need some ability to ask computers, real questions and get solid responses. I mean I look at programs like Grammarly, which are amazing. They give incredible feedback on paragraphs that you write and recommendations on writing style for readability and much more. What’s next? Will a computer wake up in the morning and go I think I’m gonna write a novel today or I think I’m gonna do some research on something today? Nothing happens without that human interaction. Computers are a long way from being self-aware, so even though the Terminator is on the horizon, it still isn’t here yet. So don’t worry my friends, we’re all gonna be OK. AI is just 40-year-old technology repackaged and sent to you to raise the stock prices of a few companies.