Introduction: I know that our chats may be used by chatbots “as training data” to improve how “the models behave, reason, and respond over time,” but I didn’t know whether that “data” includes the actual content of our messages. Here’s what I learned from a collaboration with ChatGPT:
Prompt: Do the chats users create actually have the potential to improve the database accessed by the chatbot? That is, do our chats contribute, in the short- and long-term, to the improvement of chatbots?
Introduction: I worked with ChatGPT on this article about a hypothetical college freshman struggling to pass a required writing course. See the prompt at the end of the article. -js
Prologue: The Boy with the Blinking Cursor
When Malik Thompson stepped onto the campus of Western Plains State University last August, he brought with him two duffel bags, a cracked iPhone, and a quiet but persistent fear: that he wasn’t cut out for college writing.
He had passed his high school English classes—barely—and had relied heavily on late-night cramming and the kindness of teachers who knew he was working two part-time jobs. But now he was in a required first-year composition course with a syllabus full of words like “rhetorical situation,” “synthesis,” and “MLA citation.” From day one, Malik felt like he was drowning in expectations he didn’t understand.
Introduction: Concerned about the timeliness of information Iʻve been receiving from chatbots, I collaborated with Copilot to come up with this list of the best search-augmented capabilities in the free versions of six chatbots. (See the prompt at the end of this article.) To stay current, they use search engines or APIs (Application Programming Interfaces) to fetch the latest information; perform real-time web searches to retrieve fresh content; and analyze the results, summarize or cite them, and blend them into the conversation. In rank order, the top three are Perplexity, Gemini, and ChatGPT. See the charts and explanations below.
Introduction: This short story was written by Claude (Sonnet 4) in response to a prompt I created, which appears at the end of the story. The purpose is to provide an example of Chatbot reach at this point in time. Equally important purposes are to highlight the richness that Filipino-American culture adds to the nationʻs fabric and to spotlight the use of AI in the lives of teens. I’ll be publishing other short stories from time to time.
Maya Reyes-Santos woke up at 6:47 AM to the gentle chime of her phone’s alarm, three minutes before her actual wake-up time—a buffer she’d programmed to ease herself into consciousness. Her room was a careful curation of her dual identity: a wooden bahay kubo replica her Lola had sent from Cebu sat on her bookshelf next to her collection of Studio Ghibli figurines, while polaroids of her friends at last month’s homecoming dance were tucked into the frame of her vanity mirror alongside a small Santo Niño statue her mother insisted she keep for protection.
Introduction: The following is a transcript of a chat I had with Gemini in the late evening of 24 July 2025. Iʻm including it in its entirety to fit form to function, to demonstrate the give-and-take inherent in effective communication not only between human and chatbot but between any encoder and decoder. As I dive deeper into chatbots, Iʻm beginning to better understand the rhetorical context between human and AI, and not so surprisingly, itʻs basically no different than human-to-human interactions, and this mindset is arguably the first step toward successful chatbotting. -js
Introduction: I collaborated with ChatGPT on this article, and it suggested the titles “AI Drama Is the New Silicon Valley” and “Soap Opera:As the Algorithm Turns.” The AI talent wars have become a strategic battlefield, and high-profile executive moves are shaping the race for AI supremacy as much as the models themselves. Below is a rundown of the most consequential, controversial, or “juicy” executive shifts (2022–2025), highlighting the drama, strategic value, and ripple effects across OpenAI, Google, Microsoft, Anthropic, Inflection, xAI, and others. (ChatGPT)
Prompt: In the race for AI supremacy, there seems to be a beehive of intrigue re executive movement among the competitive companies, with executives from one company moving to other companies etc. Can you provide a rundown of the juiciest movements based on their impact on the affected companies?
Introduction: Overwhelmed by the mishmash of players, companies, products, and services associated with chatbots, I collaborated with ChatGPT to develop a clarifying table. I hope this helps you as much as it does me. -js
Prompt: For chatbots such as ChatGPT, Gemini, Claude, Copilot, Pi, Poe, Perplexity, You.com, and DeepSeek, please identify the associated companies, major executives, and LLM engines.
Introduction: This article is a collaboration with Gemini this morning (23 July 2025). Future projections could be considered whimsy, but theyʻre a fun way to switch our headlights to high to see farther down the road to tomorrow. To raise the fun factor, I decided to concentrate all future UH students into a single hypothetical student, Keani. The prompts below describe the parameters given to Gemini. The purpose of this article is to present a clearer, more vivid picture of how AI might impact students entering the University of Hawaiʻi System in fall 2025. Hopefully, itʻll shed some light on how educators at all levels can better prepare for the next decade. Although the focus is on the UH System, the projection may generalize to other colleges in the US and the world. -js
Introduction: This article is a collaboration with ChatGPT. With the gigantic Air Jordan leaps we’re taking toward AI, we can’t help but wonder: Are people with special needs being neglected? Left behind? Or are we, as a society, making efforts to not only bring them along but developing innovative AI technology to address their needs? Perhaps the heart of the question is: Are we doing enough?
The following ChatGPT response is a recomposition, combining sections of the original raw responses from this afternoon with sections from the revised. I had asked for a report that combined the separate responses in paragraphs instead of outline format, and I specifically asked for “fluid and coherent” transitions, but this instruction turned into 60-grit sandpaper for some of the sections, removing the bumps and splinters that made the original text readable and dynamic.
As a result, sections of the revised text turned out to be too smooth and lifeless, having lost the raw character that gave them zing. So I cut and pasted ChatGPT’s original and revised outputs into a more lively whole. I did some reorganizing for fluency but left the text more or less intact. Takeaway: With chatbot instructions, we have to be careful about what we ask for. We might just get it. -js
Introduction: I collaborated with nine different chatbots to come up with a list of college professors who are using them in their courses in exemplary ways. The purpose was to provide readers with concrete examples of chatbot use by professors in their courses. These examples, hopefully, will generate interest among educators to integrate AI strategies in classrooms.
A secondary purpose was to spotlight professors who are actively applying chatbots in their courses. They’re leading the way into the AI Century (2025-2075) and deserve recognition. I’m sure the chatbots have missed dozens if not hundreds of other professors who should have been on this list of 49. If you happen to be one or know of others, please let me know in the comments section attached to this article.
I asked each chatbot to identify ten. I ended up with 49 professors (teams were counted as one). Five appeared in two lists, and one appeared in three (Professor Ashok Goel, Georgia Institute of Technology). I eliminated items in lists that omitted professor names. I eliminated an entire list because the chatbot failed to include information requested in the prompt. One chatbot listed only two professors.
Introduction: Yesterday and this morning, I’ve been chatting with ChatGPT re OpenAI’s power structure and its status as both a nonprofit and for-profit, focusing on its mission, which includes the statement “avoid enabling uses that would … unduly concentrate power.” I asked the chatbot to write an essay that covers our broad discussion, and I’m publishing it below pretty much as submitted. (I’ve added yellow highlights.) Since ChatGPT is an OpenAI service, I realize the potential conflict, but I’m proceeding with the intent to provide information that may be useful to readers, with the caveat to remain open and objective at the same time, i.e., to separate fact from opinion while gathering insights into arguably the most powerful force in AI and the reliability of chatbots in general.
Disclaimer: I am not affiliated with OpenAI or any of its partners. This article reflects my independent research and perspective, and is not endorsed, reviewed, or influenced by OpenAI in any way. -js
Prompt: Please provide a 2000-word summary of our chat, thus far, on OpanAI’s nonprofit and for-profit balance and its mission statement: “avoid enabling uses that would … unduly concentrate power.” Use a paragraph, essay instead of outline or bulleted format, and keep the tone conversational and informal rather than technical. Feel free to introduce new information to enhance the fluency and coherence of this paper. Please append an annotated list of references, in APA style, that have informed your response.
Introduction: For this article, I collaborated with six chatbots: ChatGPT, Claude, Gemini, Copilot, Perplexity, and Pi to identify the college president most successful in advocating AI.
Prompt #1 (open-ended): Please identify the one college president in the country who is currently succeeding at doing the most to promote and apply AI at their institution. Describe this person in detail, providing concrete support for your selection. Use a paragraph and conversational style instead of a bulleted outline. Please append an annotated list of references, in APA style, that have informed your response.
The results for prompt #1:
Darryll J. Pines, Connie Ledoux Book, Ross Gittell, Joseph E. Aoun, José Luis Cruz Rivera
Introduction: For a better grasp of the advancements in GPT-4, I’ve asked ChatGPT to explain, in language aimed at 10th-grade high school students, the critical differences between GPT-3.5 (current free version) and GPT-4 (paid version). Keep in mind that I have been and am using the free GPT-3.5 version in all my articles on chatbots — along with other free chatbots such as Claude, Gemini, Copilot, and Perplexity.
To clarify the explanations, I’ve asked ChatGPT to provide actual examples of how 10th graders might apply these advanced features to real-world learning activities in typical courses such as English, Math, History, Art, and Science.
As this conversation progressed, I realized the power of chatbots in developing curricula and lessons. I’m sure many if not most educators are already making use of this potential. Along the same lines, students could use chatbots to plan their school assignments and projects. This article focuses on high school students, but I believe the ideas can be scaled to lower grades and college as well.
Disclaimer: I am not affiliated with OpenAI or any of its partners. This article reflects my independent research and perspective, and is not endorsed, reviewed, or influenced by OpenAI in any way. -js
Introduction: For this article, I collaborated with three chatbots: Gemini (Google Bard), ChatGPT (GPT-3.5 free; GPT-4, OpenAI), and Claude (Sonnet 4, Anthropic). I asked each to come up with five seminal works in the development of chatbots. Three were mentioned by two chatbots, so I ended up with a list of twelve. They are listed below by their date of publication. Three were published before 2000, and only one between 2000 and 2014. Five were published between 2015 and 2019, and the remaining three, 2020 and after. Thus, 67% were published in the last ten years.
1. “Computing Machinery and Intelligence” by Turing (1950). 2. “ELIZA—A Computer Program for the Study of Natural Language Communication Between Man and Machine” by Weizenbaum (1966). 3. “Procedures as a Representation for Data in a Computer Program for Understanding Natural Language” by Winograd (Often referred to as the SHRDLU dissertation) (1971). 4. “Social Dialogue With Embodied Conversational Agents” by Bickmore & Cassell (2005). 5. “A Neural Conversational Model” by Vinyals & Le (2015). 6. “Attention Is All You Need” by Vaswani et al (Google Brain Team) (2017). 7. “BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding” by Devlin et al. (2018). 8. “Language Models Are Unsupervised Multitask Learners” by Radford et al. (2019). 9. “A Unified Framework of Five Principles for AI in Society” by Floridi & Cowls (2019). 10. “Language Models Are Few-Shot Learners” by Brown et al. (2020). 11. “Constitutional AI: Harmlessness from AI Feedback” by Bai et al (2022). 12. “Training Language Models to Follow Instructions with Human Feedback” by Ouyang et al. (InstructGPT Paper, 2022).
See the chatbot listings below for details about each work. I’m drawn to the latest, especially Brown et al. and Ouyang et al. I’ll follow up this article with chatbot-generated reviews of these two.
Introduction: I was curious to see how a chatbot would work with a first-year college student who has hated math since high school and is now struggling to pass a required course in algebra. I decided to focus on a male student but asked the bot, at the end of the process, if it would use the same approach with a female student. The chatbots tested were ChatGPT, Claude, and Gemini. As usual, to better understand the philosophical roots of the bots’ approach to tutoring, I asked them to provide explanations and references for their pedagogical decisions. The subject matter is math, but I believe the instructional approach would generalize, with a few tweaks, to other disciplines. The purpose of this article is to give readers a feel for how the human-bot collaboration might work out and a sense of its potential effectiveness, and the goal is to encourage them to make the leap — if they haven’t already — into chatbotting as an invaluable tool in their academic and professional skill set. Finally, another more practical purpose is to get a better feel for the strengths of these three chatbots by experiencing how they might approach the same pedagogical problem, with the caveat that chatbot performances often vary quite a bit from chat to chat. -js
Introduction: In this article, I presented the same prompt to seven chatbots to observe the range of responses and share the results with readers. The chatbots, in order, were: Copilot, Perplexity, ChatGPT, Gemini, Claude, Pi, and You.com. I’ll let you be the judge of how well each bot initiated communication with a preschooler. -js
Summary: This article examines how human and AI critical thinking compare and may evolve by 2045 and 2075. It outlines six key skills—like inference and self-regulation—where AI currently lags due to a lack of ethics, reflection, and awareness. Human–AI collaboration is seen as complementary: AI provides speed and pattern detection; humans offer judgment and creativity. By 2045, AI could master long-context reasoning, memory, and limited moral reasoning, but core traits like consciousness remain elusive. Looking ahead to 2075, advances may include neurosymbolic AI (blending logic with learning), biohybrid systems (integrating AI with biological components), and embodied cognition (giving AI physical presence and sensorimotor experience). These could make AI appear more human-like—but still without true self-awareness or intent. Ultimately, the article envisions a future of collaborative intelligence, where humans and AI co-evolve within ethically grounded partnerships. (ChatGPT)
Summary: This article explores the profound impact of AI on future employment and higher education from 2025 to 2045. It predicts a transformation in university programs, advocating for modular, interdisciplinary, and experience-based learning. Examples of future degrees include AI Systems Architecture & Engineering and AI Ethics, Governance & Policy, emphasizing advanced AI concepts, ethical considerations, and mandatory experiential learning like internships. The piece envisions new job roles, such as “Quantum AI Architect,” with high salaries, reflecting a demand for specialized skills. It draws on reports from leading organizations to underscore these projections, highlighting the evolving landscape where human-AI collaboration becomes central to career success. (Gemini)
Summary: This article explores how traditional academic disciplines will transform over the next 50 years due to AI integration and global challenges. The author uses ChatGPT and Gemini AI to analyze current core disciplines and project future changes. The piece examines how AI will be integrated into academic divisions, with projections showing computer science achieving 95-100% integration by 2075, while humanities may reach 60-75% integration. The analysis suggests that rigid departmental boundaries will dissolve, replaced by problem-focused, interdisciplinary fields addressing climate change, consciousness studies, and space exploration. The timeline spans from 2025-2035’s “Convergence Decade” through 2065-2075’s “Transcendence Decade,” with new disciplines emerging like neuroengineering, synthetic biology, and eventually cosmic consciousness studies. The article identifies 20 key leaders and institutions driving these changes, including Fei-Fei Li at Stanford and organizations like MIT’s CSAIL, emphasizing that technological convergence and societal demands for cross-disciplinary solutions will fundamentally reshape higher education’s structure and purpose. (Claude)
Here’s an actual example of how a user collaborates with bots to generate a more “accurate” output. In this context, “accurate” is a result that satisfies both parties.
This evening, after viewing a YouTube video (Mick Talks Hoops, 7/16/25, “Liberty Coach & Players SPOTTED In Meeting With Caitlin Clark & Her Agent… Recruiting??”) speculating on the possibility of Caitlin Clark leaving the Indiana Fever and joining the New York Liberty after this season, I prompted a few chatbots for their opinion on this rumor.
My initial prompt with ChatGPT: Considering the issues that are plaguing the Indiana Fever and the interest other WNBA teams are showing in her, is there a chance that Caitlin Clark will be leaving the Fever for a different team after this season? If yes, what are the percentage odds of this happening? Will her decision be based on whether teammates such as Aliyah Boston may also be leaving? If she leaves, which team or teams are most likely to pick her up? Does she have a preference for a specific team? If she leaves, what are her most likely reasons?
Summary: The article explores the emerging concept of “AI natives” as a successor to “digital natives,” representing a fundamental shift in how humans interact with technology. Marc Prensky, who coined the term “digital native,” is now pioneering the concept of AI natives as the next human evolutionary leap. This transition marks the beginning of Generation Beta in 2025, representing the first truly AI-native generation. Unlike digital natives who primarily consume and share information through digital platforms, AI natives collaborate with artificial intelligence as cognitive partners, learning through dialogue and iterative refinement rather than traditional search and discovery methods. The article presents responses from ChatGPT, Gemini, and Claude, each offering different perspectives on this generational shift. Key differences include AI natives’ approach to problem-solving through AI reasoning engines, their development of AI literacy and prompt engineering skills, and their expectation of personalized AI tutoring experiences. The implications for educational institutions are profound, requiring fundamental restructuring of curricula, assessment methods, and pedagogical approaches to accommodate learners who will enter higher education empowered by AI collaboration capabilities. (Claude)
Introduction: I collaborated with Gemini, ChatGPT, and Claude on this report. Prompt: Will there come a time when many if not most students will complete the requirements for a college degree without stepping foot on a college campus and taking professor-led courses, relying primarily on partnerships with chatbot mentors and advisers? If yes, please identify colleges or individuals that are pioneering this effort. Also, provide a 50-year timeline, in 10-year increments, to explain and illustrate how this might play out between 2025 and 2075. Please append an annotated list of references, in APA style, that informed your response. -js
Summary: The article explores whether a highly motivated 15-year-old could successfully drop out of traditional school and use AI chatbots to develop a personalized curriculum, earn a high school equivalency diploma, and gain admission to competitive colleges. The piece presents comprehensive responses from three AI models (Claude, ChatGPT, and Gemini) that largely agree this path is feasible but challenging. The proposed strategy involves obtaining a GED or HiSET through AI-powered tutoring, using chatbots for curriculum development, assessment, and academic support, and leveraging AI tools like Khanmigo, Socratic by Google, and various educational platforms. The responses detail how AI could provide 24/7 personalized instruction, generate custom lesson plans, and offer real-time feedback across multiple subjects. All three AI models emphasize that success would require exceptional self-motivation, strong family support, and strategic planning. They highlight the importance of building a compelling college application portfolio through independent projects, standardized test preparation, and finding human mentors for recommendation letters. The article suggests that exponential improvements in AI technology through 2025-2029 will make this approach increasingly viable, with enhanced personalization, multimodal learning integration, and predictive analytics transforming the educational landscape fundamentally. (Claude)
Summary: This article presents a century-long projection on how chatbots and AI will shape early childhood development and education. The article argues that infants will increasingly interact with chatbots before they can speak or read, enabled by AI-powered toys, smart cribs, and interactive devices. These systems will provide immediate feedback to children’s babbling and gestures, creating new pathways for language and cognitive growth. AI’s role, presented in ten-year increments, beginning with supporting parents and teachers, then moving toward fully personalized, multilingual, and interest-driven learning via adaptable AI companions. By mid-century, brain-computer interfaces and AI-guided neural development will accelerate learning, rendering traditional education models obsolete. In later decades, children will develop in symbiosis with AI, acquiring expanded cognitive, creative, and even planetary consciousness skills. Despite these advances, the article cautions that careful design is needed to safeguard human values and developmental essentials. (Perplexity)
Summary: This article presents a creative writing experiment using ChatGPT to generate a 3,000-word story beginning titled “Homeless Awakening.” The narrative follows a homeless Asian man sitting motionless on Ala Moana Boulevard near Honolulu Harbor, who gradually awakens from a catatonic state. The story chronicles his slow emergence from silence through small interactions with Mr. Leung and his daughter Mei from the nearby Golden Lily restaurant. As the man begins helping at the restaurant, hints emerge of his wealthy past and family betrayal in China. The narrative ends with mystery surrounding his true identity and circumstances. (Claude)
Educational Technology and Change Journal 
