Michael Martinez

Continuing on the theme of LLMs enhancing your own work rather than replacing it. I had several platforms sift through and critique my writings on my posts and blogs. Some of the results were interesting, and unexpected to say the least. For this test I prompted Gemini 2.5 and Claude Sonnet 4 the following: “Critique the writings of michaelmartinez.co.”

This is yet another tool in my workflow. I could take or leave the suggestions, but as always, it’s an option. After reading through the “experiment” I want to know what you – the reader would like to see more of from me in future posts and writings.

Gemini 2.5 talked of the following strengths:

Breadth of Knowledge: Martinez exhibits a wide array of interests, seamlessly transitioning between discussions on cybersecurity, the nuances of digital advertising antitrust cases, and historical analyses of xenophobia. This breadth enriches his perspectives and allows for interdisciplinary connections.

Topical Relevance: His chosen subjects are highly relevant to current global discourse, addressing pressing concerns in the tech industry (LLMs, AI, antitrust) and societal issues (xenophobia, foreign policy).

Clear and Concise Communication: The articles are structured well, with clear headings and a direct writing style, making complex topics accessible to a broad audience. He effectively summarizes key points and arguments.

Thought-Provoking Insights: Martinez often delves beyond surface-level observations, offering insightful critiques and raising pertinent questions, such as the effectiveness of breaking up large tech companies and the future impact of AI on advertising. His willingness to express concern about negative rhetoric and connect it to historical patterns of xenophobia adds a moral and ethical dimension to his foreign policy discussions.

Personal and Reflective Tone: Elements like his “Reading Updates” and experiences with Micro.blog add a personal touch, making his content more relatable and authentic. His emphasis on continuous learning and reading widely underscores a commitment to self-improvement and intellectual growth.

Use of External References: He frequently references external sources, such as TechCrunch’s “Cyber Glossary,” Jesen Huang’s views on AI, Mark Gurman’s criticisms of Apple, and Clayton Christensen’s “The Innovator’s Dilemma.” This demonstrates research and supports his arguments with credible insights.

The second part of the output were the “Areas for Potential Enhancement”.

Deeper Dive into Specific Arguments: While he touches on various aspects of each topic, providing more in-depth analysis or expanding on specific arguments with additional evidence or case studies could further strengthen his positions.

Original Research or Unique Perspectives: While he synthesizes information effectively, incorporating more original research, unique data points, or highly distinctive viewpoints could elevate his critiques to a more authoritative level.

Engagement with Counterarguments: Presenting and directly addressing potential counterarguments to his positions could lead to more robust and nuanced discussions.

Claude provided me with almost the same strengths and weaknesses as the Gemini prompt but went further in two ways. First, is a “Content Assessment”.

Technology Commentary: His AI/LLM analysis is generally sound but not groundbreaking. He correctly identifies key limitations but doesn’t offer particularly novel solutions or frameworks for thinking about these issues.

Political Commentary: His piece on American xenophobia attempts to draw historical parallels between treatment of Japanese Americans, Muslims post-9/11, Russians during the Ukraine conflict, and current Latino deportations. While the historical connections are valid, the analysis lacks nuance and depth that such complex topics deserve.

Educational Insights: His observations about AI in education are practical and well-reasoned, particularly his emphasis on teaching critical thinking alongside AI literacy.

Lastly, Claude provided an “Overall Assessment”.

Martinez’s blog represents competent but not exceptional commentary on technology and society. His strength lies in providing accessible, practical perspectives on complex topics, particularly around AI adoption and digital literacy. However, the writing lacks the depth of analysis or original insight that would distinguish it in a crowded field of tech commentary.

The blog serves its apparent purpose as a personal platform for sharing thoughts and maintaining an online presence, but it doesn’t rise to the level of must-read commentary on the topics it covers. For readers seeking practical, moderate perspectives on AI and technology adoption, it offers value. For those looking for cutting-edge analysis or original research, they would likely need to look elsewhere.>The author’s commitment to continuous learning and reading is admirable and likely keeps his commentary reasonably well-informed but translating that input into more distinctive analytical output would strengthen the overall impact of his writing.

I will keep retooling and refining my methods in the future based on this feedback. What do you, a human, think of the assessments from two of the biggest GenAI tools out there?

Just like with all industries and applications, the use of LLMs is literally rewriting all of the best practices with education currently being the largest social sector being disrupted at the moment. Rather than using LLMs to ‘cheat’, educators and public administrators must teach students how to use and coexist with them rather than relying solely on them.

A lot of students have abandoned critical thinking all together and resulted in outsourcing all of their knowledge to these LLMs just to make answers and save time without learning or even conducting reasoning of their own to consider what the prompts are or what they’re telling them. As a result, teachers, administrators, and parents are beyond frustrated and returning to the good old blue books as cited by this Gizmodo article.

Students need to learn how to properly research while understanding proper time management. I won’t say laziness is at work on the part of all parties involved, but the quickest way to get from A to B is often using tools at your disposal – this is true in education or eventually the workplace, however, a lot is lost in this process including how to critically think about what output LLMs are displaying or what it might mean for the overall context of the subject.

Education has always been a lagging indicator of technological trends, and this is no different. LLMs and other types of GenAI are tools, not the end all be all solution in the classroom. Using it as a partner in research yet taking a critical view of what it’s telling you is paramount to making research easier for all, without compromising the time-honored tradition of writing research papers and a child’s knowledge retention.

A full education, as always, should concentrate on a child’s soft-skills – learning how to critically think, put together proper research, how to write for life, and home in on communication skills to make them successes in their careers and lives. Tools can do that, pouring prompts into an unchecked LLM cannot.

Believing AI is sentient leads to false expectations—some may trust AI’s recommendations without skepticism, assuming they are the result of independent reasoning rather than probabilistic predictions. In policy discussions, regulators struggle with defining AI responsibility, misplacing ethical accountability onto models instead of the humans who deploy them. The danger is not AI itself but how we perceive and integrate it into decision-making, security, and governance.

‼️ Don’t Be Like Everyone Else:

🧠 Understand AI’s Limitations – It predicts and mimics, but it does not think. Recognizing this distinction helps avoid misinterpretation.

📖 Stay Informed on AI Ethics and Policy – Governments are trying to regulate AI’s role, but misconceptions could shape flawed laws.

🔨 Use AI as a Tool, Not a Decision-Maker – It can enhance productivity, but critical thinking must always come first.

The future belongs to those who understand AI for what it is—not what sci-fi wants it to be. Whether you’re a student, a professional, or a policymaker, mastering AI’s true capabilities is no longer optional—it’s a necessity.

This was originally posted on LinkedIn on May 30, 2025.

After reading this piece on CNBC’s website regarding how students should be using AI (I still prefer the term LLMs for purposes of nobody can identify what AI is and isn’t), it’s making me consider what today’s youth is learning with respect to preferential employment skills moving forward.

First consider the interviewer. Jesen Huang’s goal is to sell as many customers on GPU compute and resources as humanly possible given his field so we must take that with a grain of salt. Secondly, the tools of LLMs are a must have in today’s educational and workforce. If you aren’t utilizing prompting to the fullest extent; you are already falling behind. Take the time to find an online course through EdX or Coursera, for example, to home in these skills.

Huang stated:

Learning how to interact with AI is not unlike being someone who’s really good at asking questions,” he added. “Prompting AI is very similar. You can’t just randomly ask a bunch of questions. Asking AI to be an assistant to you requires some expertise and artistry of how to prompt it.

Like all tools, they must not only be learned but practically used. Using a tool for the sake of it, creates more problems and diminishing results will follow.

We also must question whether the idea of learning how to code is vital for today’s computer science programs. I still argue yes! If we blindly follow output from a chatbot, we lack the ability to understand what the code means, if it works, or how useful it is to the original prompt. “Is this what the client actually asked for?” or “How can we implement this?” are two major questions that will never go away. Just like Wikipedia was and is a tool as a basis for research and learning, chatbots, and LLM products should also provide this starting point. As always, check the source material as like Wikipedia, bias and humans still intervene in the research and results that an LLM product provides.

The part that Huang really gets correct is as follows:

Perfecting AI prompts — and asking better questions in general — is a skill that will remain relevant for years to come, so students should take the time to develop it, no matter what career field they see themselves in.

A large amount of Academica is yet to be versed on Large Language Models as a whole and prompting is still new to those set in their ways. Aiding in research is paramount to have a companion in the room that makes learning easier and the consumption of knowledge more streamlined.

We have a long way to go as a society between those who are skeptical of LLMs at all costs and those who talk to it and treat it as it’s a human being with real thoughts and valid feelings. The battle between skeptics vs. accelerationists is not one we should be having; but a moderating position to accept them as tools that we all must learn to be successful in any career or academic endeavor; no matter what areas of study we choose to pursue.