Ten Reflections on the Future of AI
From ten different authors.
The Intelligent Friend - The newsletter that explores how AI changes the way we think and behave, only through scientific papers.
Intro 🖍️
Dear readers of The Intelligent Friend, how are you? I hope you had a wonderful week and a wonderful Thanksgiving.
I am really excited to show you this week's issue. In these weeks I have worked with ten authors on a large collaborative issue of “mini-perspectives” that could provide useful ideas to reflect on the future and present of AI from many points of view: on its impact on writing, on education, on skills, on daily life and much more. The result is what you can read today.
As you know, in this newsletter we delve into the topics of AI and AI-human interactions through scientific papers. I thought it would be interesting, in an issue with so many points of view, to also bring the perspectives of professionals who deal with AI or people who write or constantly use tools based on this technology. They are also, as I said, all people who write about AI, and therefore who cover these topics and constantly inform themselves about them. I am really happy with the final result and to be able to bring you this issue as well.
There are several paragraphs by exceptional authors here on Substack, on the most disparate topics related to AI. I think that, however, despite their diversity, they can really stimulate, all together, a transversal reflection and give you many ideas and suggestions. I would like to sincerely thank each author who contributed to this issue. Of course I will include their references as you read. If you like this format, I would like to propose it again on other topics and with other authors. So, if you write about AI or related topics, let's get in touch!
Now enjoy this issue, I'm sure you will appreciate it.
AI Trends 2025 -
I’m Mike Spencer, of the Newsletter AI Supremacy. You can find me on Bluesky here. In 2025 there are a few major things I’m watching for the continued development of Generative AI. The massive capex of BigTech into AI infrastructure will become much more competitive with fewer corporations, countries and frontier model builders able to compete. Meanwhile, the scaling laws of the models begin to fundamentally slow down impacting a lack of new breakthrough and emergent capabilities.
More datacenter projects and more mega hyper-compute AI supercomputers mean more demand on Energy grids, real-estate, water and AI pushing environmental concerns. BigTech is likely to spend over $300 Billion in capex, a significant ramp up from 2024 levels. Meanwhile Nvidia’s AI chips will continue to get better, pushing Taiwan’s semiconductor industry to new limits of demand centered around TSMC. TSMC has gotten Government and state funding from Japan, the U.S. and Germany to build in those countries too.
The great datacenter boom and new bottlenecks in the upper capabilities of LLMs, means pressure to rethink the value and ROI of Generative AI with a continued transition to a focus on Enterprise AI, product and AI agent frameworks and task-automation goals. At the end of 2024, China’s Open-source LLM innovation in Deepseek and Qwen are reaching new levels and are no longer 9-months or more behind the U.S. The geopolitical rivalry, trade war and cold-tech war between the U.S. and China could escalate significantly due to China’s technological leverage for the future. The Trump administration’s impact on AI, in particular around policy and regulation could influence the outcome of “AI in the wild”, leading to unintended consequences.
Things to watch: Datacenter growth, Energy demands, semiconductor supply chains, Nvidia, BigTech capital expenditures, focus on AI product and Enterprise AI, AI Agent ecosystem and frameworks, Open-source LLMs, China, TSMC, Trump’s impact on the stock market and AI policy. Diminishing returns and mean reversion of exaggerated impact of Generative AI on society, innovation and business. AI risks around geopolitical conflict, national rivalry, AI arms race like dynamics, and escalation of regional unrest.
New Impact for New Interfaces - Josh Brake
✍️ Author of The Absent-Minded Professor
We’re living in the command line era of AI. The past two years of development since the public release of ChatGPT have been largely defined by a single interface: the chatbot. The hunt is on for the graphical user interface for generative AI. As we look toward the next two years, I suspect we’ll see lots of experimentation with new interfaces that explore new ways of interacting with our computing paradigms without screens and keyboards.
Generative AI provides opportunities to leverage modalities such as voice and sight in ways that have not been possible until now. OpenAI’s Advanced Voice mode is one example of this, but there are many more on the horizon. As new interfaces for generative AI emerge we’ll need to be increasingly aware of how these interfaces are forming or deforming us as we interact with them.
Marshall McLuhan’s phrase “the medium is the message” will be increasingly important as a lens through which to see the tools we use. New interfaces will provide opportunities to break us free of the worst aspects of our current screen-based technology allowing us to more effectively and naturally interact with computational resources.
And yet, these new interfaces will also present new challenges. We’ve already seen how human-sounding chatbots can deceive us, sometimes with tragic consequences. As we continue to grapple with generative AI we’ll need to focus not only on its technical capabilities, but even more importantly on the impacts these new tools are having on us as individuals and communities.
Reflecting on the Near Future of AI in Education - Nick Potkalitsky
As an educator, writer, and AI literary consultant, I feel that the future of AI is very bright indeed. That said, I have had to work exceptionally hard to reach this patch of pragmatic optimism. My colleague Mike Kentz characterized most educators' experience of AI disruption through the lens of the five stages of grief. Many of us in the teaching field have worked our way through denial, anger, bargaining, depression, and finally acceptance—a journey facilitated by reading, experimentation, collaboration, and innovation. While some of us have received support from our schools and institutions, many have had to build our own literacy programs, discovering something both challenging and transformative in these new technologies.
In my early thinking, I viewed AI primarily as a magnifier of existing challenges in today's schools. We have long struggled with how to assess students fairly, accurately, and in ways that foster engagement. AI's arrival deeply challenged our traditional methods rooted in long-form essays, fact memorization, and high-stakes testing. However, I now find this initial perspective quite limiting.
My recent work with students has revealed that AI, when implemented through secure tools and skillful approaches, can inspire students to think generatively. By this, I mean that AI prompts students into open-ended, multi-perspectival, process-oriented, and cognitively nuanced work cycles. These cycles keep students engaged longer and more meaningfully as they build essential 21st-century skills. While developing instruction and curriculum to scaffold these work cycles around tools not necessarily designed for educational purposes is challenging, it has become the most exciting work of my educational career.
For those interested in learning more about AI literacy and these classroom work cycles, I encourage you to explore my new book, co-authored with Mike Kentz, AI in Education: A Roadmap to a Teacher-led Transformation. I have also developed two online courses to help teachers quickly develop their AI skills. The first course focuses on developing effective AI prompting strategies, while the second explores using AI for instructional differentiation.
The Compressed 21st Century: How AI Could Revolutionize Humanity's Future - Matthew Harris
As artificial intelligence continues to advance at an unprecedented pace, its potential to reshape various facets of human life becomes increasingly evident.
From eradicating diseases to enhancing economic development, AI stands on the brink of ushering in a new era of human flourishing.
A concept I and others refer to as the "compressed 21st century" .
The concept of the compressed 21st century envisions AI achieving 50-100 years' worth of biological and medical advancements in just 5-10 years.
This acceleration hinges on AI systems acting as virtual scientists, designing and conducting experiments, inventing new techniques, and rapidly iterating on discoveries.
AI could drastically amplify the rate of key scientific discoveries by simulating and analyzing complex biological processes and identifying research avenues that might elude human colleagues.
This could lead to a tenfold increase in critical breakthroughs like CRISPR gene-editing technology, advanced microscopy, and mRNA vaccines.
For example, clinical trials could become significantly more efficient with AI. By identifying drugs with larger effect sizes and predicting outcomes more accurately, AI can help bring effective treatments to market faster than ever before.
But it's not just biological advancements which could accelerate. We’re also talking about economic development and the globalized distribution of benefits to all people.
Ensuring equitable access to AI-driven advancements is a moral imperative and as such, AI's transformative potential must extend beyond developed nations to have a truly global impact.
This includes:
Efficient Health Intervention Distribution: AI-powered epidemiological modeling and logistics can make life-saving treatments widely accessible.
Unprecedented Economic Growth: Optimized economic planning and AI-enabled technologies could sustain high growth rates in developing countries.
Food Security: AI-driven agricultural improvements and optimized food distribution could alleviate hunger and malnutrition.
Climate Change Mitigation: Innovations in clean energy and sustainable practices could disproportionately benefit regions most affected by climate change.
The potential of AI to revolutionize our world is both immense and imminent.
By responsibly developing and deploying powerful AI, humanity stands on the cusp of unprecedented advancements across health, neuroscience, economics, governance, and personal fulfillment.
Achieving this vision is not guaranteed—it requires concerted efforts from developers, policymakers, and individuals to mitigate risks and ensure that the transformative potential of AI benefits all of humanity.
We live in the best of all possible worlds,” said Doctor Pangloss.
“Yes” agreed Candide, “but we must cultivate our own gardens.”
Candide by Voltaire, 1759.
What’s happening to creativity? A reflection - Andrew Smith
What is creativity? We humans have been wondering this for thousands of years. During much of our past, creativity was a gift from the gods. Our earliest and well-known myths have reinforced this over the generations, and these myths laid the foundation for our view on creativity today. Take a gander at the Greeks, who came up with the Nine Muses, explaining why some human beings could come up with novel ideas, while others could not. Likewise, Abrahamic religions insisted that God created everything ex nihilo. Creativity was clearly a divine spark that took hold of a person’s mind, inspiring them to… well, create.
How did this divine spark first emerge? I imagine it was out of necessity, the mother of invention. We humans didn’t have everything we needed to survive and thrive, so we invented new ways of solving problems. This very practical and necessary step arose so that we could survive and become human, for creativity and humanity are joined at the hip. Or so we’ve been led to believe our whole lives. There’s no doubt that we are incredibly creative. We needed to create stones that were sharper than those found in nature, and so we did. We needed a way to stay warm and to cook our food, so we figured out how to keep fire under control. We needed a better way to communicate, so we made up language. One day, we made something that was capable of creating new things all on its own. Suddenly, people can generate ideas that nobody before them has ever come up with, and they can do this in an instant. Why is everyone so reluctant to call this creativity, when we’ve been the only ones we considered creative for as long as we can remember? I think the question answers itself.
The “minimum viable prompt” - Daniel Nest
AI chatbots are fantastic at helping you brainstorm ideas and gain insights, regardless of your background and the task. But people still often hesitate to use LLMs because they don’t know where to start. My advice?
Start with the “minimum viable prompt”: the most basic description of your task. In that prompt, tell the chatbot to ask you questions before it responds to understand your needs better. Those two steps are super simple to apply and will help you get much more out of any interaction with AI.
At the same time, I’m a strong proponent of never outsourcing your critical thinking to AI. Use AI as your sparring partner, beta reader, and advisor—not your replacement. With the emergence of AI-powered web search, we now risk a situation where we might start deferring to chatbots for our research. But remember that LLMs can hallucinate, and always double-check any critical information and facts.
The Seduction of Outsourced Thinking - Jeremy Caplan
A blank page is uncomfortable. An uncertain mind is stressful. I could summon an AI to fill this space. Outsourcing the thinking behind these words would shield me from the discomfort of not knowing precisely what I want to say.
Those of us who write regularly are intimately acquainted with the friction we confront when we attempt to write something new. It’s especially challenging to put into words an idea we haven’t fully figured out yet.
It’s more than writing that we’re at risk of outsourcing. The temptation to outsource our thinking is substantial. Just as frozen dinners, CliffsNotes, cake mixes, and store-bought greeting cards spare us the mental and physical exertions associated with deciding, measuring, and wrestling with details, so does AI offer us an easier path.
The lazy brain’s temptations will grow ever more alluring. Once we relish the ease with which we can evade troubling our own minds with tricky knots of ethics, principles, values, feelings, complexities, and nuances, it’ll be hard to turn back.
When we outsource thinking, we sacrifice not just words—we surrender intellectual autonomy. When we let an AI untangle an ethical dilemma or resolve an emotional issue, we atrophy our own capacity for independent sensitivity and deep consideration. We will increasingly rely on — crave — the convenience of automated insight.
The Responsibility Role in AI - Jurgen Gravestein
✍️ Author of Teaching computers how to talk
Bad people will use technology to do bad things. That is as true for AI as it is for any technology that came before it. The problem with AI, however, is we don’t need bad actors to do harm. All we need a healthy dose of ignorance and overreliance on systems that cannot be trusted. Whether it is in healthcare, education, government or the military — AI in its current form has no business in making decisions for us. To rely on it is to outsource agency, but guess what you cannot outsource to AI? Responsibility. Our ignorance cannot free us from accountability for our own actions. If you kill someone with a gun, the gun isn’t found guilty, you are. Don’t become complacent. Happy holidays and a magnificient 2025.
A scenario to reflect - Terry Underwood
✍️ Author of Learning to Read, Reading to Learn
By 2039, education has transformed radically in response to the "Developmental Autonomy Paradox": Although AI tutoring eliminated achievement gaps by 2030, students became unable to learn without digital help.
The solution? "Conscious AI Integration" (CAI), featuring mandatory "AI-free zones" where students spend 30% of their time in analog "Pure Human Learning." Here they develop "Human Core Competencies" teaching skills like empathy, emotional intelligence, and joy that remain uniquely human despite AI advancement. Students now take "Neural Autonomy Training," using neurofeedback technology to switch between AI-augmented and independent thinking.
The goal is "cognitive sovereignty" which is bringing its own challenge: "Reality Synthesis Disorder." Students report living in parallel realities—one AI-optimized, one chaotically human—leading to "authenticity anxiety" as they question whether their achievements and thoughts are truly their own.
The core question of education in 2039: How do we teach students not just how to learn with AI, but when to think without it?
Digital Twins A Practical AI Tool for Time-Strapped Founders - Martin O'Leary
An AI version of yourself? Sounds a bit sci-fi and maybe even impersonal.
But stay with me—AI-powered digital twins are becoming a practical solution for founders juggling a million things. Tools like HeyGen let marketing teams create personalized welcome messages and social content that feel like the founder, without blocking calendars for endless video shoots.
Let's be real: startups need both personal connection and scalability to compete.
Traditional video content hits a wall—your founder can't be everywhere. A digital twin bridges this gap. Your marketing team can create and test different messages using the founder's digital twin, optimize what works, and scale it across thousands of customer touchpoints. It's like having your founder's best self available on demand.
The key? Be upfront about using AI and focus on the value it brings.
Don't fake it—customers can tell. Instead, use it to offer faster responses and smoother onboarding. Your marketing team can continuously iterate on content without disrupting the founder's schedule. If it's transparent and helpful, people get it. Sure, the tech isn't perfect yet. But used thoughtfully, it could free up your founder while still delivering that personal touch customers love. Win-win, right?
A Final Note
I would like to thank once again the authors who participated in this collaborative issue.
Theirs are valuable reflections of various nature that can stimulate ideas and discussions. Furthermore, I am enthusiastic about having been able to have perspectives that also derive from data and experiences of different types.
See you soon!
Riccardo
Thanks for the opportunity to share my thoughts!
PS. it looks like my piece cuts off mid-sentence… Can you fix that?
Such a great group of incredibly sharp people giving their nuanced perspective on every facet of society.
Great job pulling this together!