B21 notes: 2025-01-14 to 2025-02-25 (14 posts)#

Reading notes on Andrew Maynard’s Substack posts in batch B21. Only his own prose is treated as evidence. Quotes are exact, including his original typos and curly punctuation; asterisks mark his italics.

Batch context: six weeks at the start of 2025. The posts themselves mention the $500B Stargate announcement, the rescinding of President Biden’s Executive Order on AI, the release of DeepSeek’s R1 reasoning model, the International AI Safety Report and the Vatican’s Antiqua et Nova (both out in the same week), OpenAI’s launch of Deep Research, the Arc Institute’s Evo 2 genome model, and the 50th-anniversary “Spirit of Asilomar” meeting. Maynard is teaching “Pizza and a Slice of Future” at ASU and co-hosting the weekly Modem Futura podcast with Sean Leahy.

Six of the fourteen posts are Modem Futura episode promotions. Following the user’s instruction not to spend time on the Modem Futura podcasts, they get one line each below, whatever their topic. Two other posts contain a lot of text Maynard did not write: “AI at a Crossroads” is a ChatGPT-written article with his postscript, and “The Artisanal Intellectual in the Age of AI” is built around an article that Deep Research drafted and he then line-edited. In both, only his framing prose counts as evidence.

The centre of the batch is a run of four posts (2025-02-04 to 2025-02-16) in which he tests OpenAI’s Deep Research on scholarship: a framing paper, a full “dissertation”, and a co-written article. From these tests he develops the idea of the artisanal intellectual.

Relevance summary:

HIGH#

2025-02-23 — evo-2-dna-ai — “An AI model that can decode and design living organisms”#

Provenance. His own prose throughout. Short quotations come from the Arc Institute paper. This is the richest risk and governance post in the batch.

Argument in his terms. - DNA as another language. He treats the arrival of generative models that “speak” DNA as close to inevitable, because “DNA as it appears in biological organisms is just another type of language that connects a sequence of symbols with functional outcomes.” Evo 2 is “fundamentally a generative model trained to predict the next base pair” (the paper’s words). It does for DNA what ChatGPT does for words, trained on 9.3 trillion base pairs from more than 128,000 genomes. - What kind of thing it is. “It is, in effect, a DNA-based stochastic parrot — except that it’s ability to “parrot” biology far exceeds anything humans are capable of on their own.” He stresses that it has no training on how genes relate to phenotypes, which he compares directly to text AI that “can generate beautiful text that emulates understanding, but which isn’t actually grounded in understanding”. So here, with no qualification, he describes current language models as emulating understanding rather than having it. - Speculative applications, which he labels as speculative. He imagines novel precision gene editing, new molecular machines, bio–machine interfaces (brain–computer interfaces, nanoscale machines integrated with biology), DNA as a functional material (DNA origami, data storage, DNA computing), and hybrid systems designed by coding DNA and atoms together, such as organoid “brains on a chip”. He calls this “a little sci-fi” but “not beyond plausibility”. His reason for taking it seriously is the pace of change: “just over 2 years to go from the early iterations of ChatGPT to AI systems that can simulate reasoning”. - Risk. “this does raise the possibility of quite profound risks if we get things wrong.” He praises the Evo 2 team for being “cognizant of the potential unintended consequences”. They left pathogenic virus genomes out of the training data (“Rather smartly”) and red-teamed the model. But he thinks this is too narrow: “the domain of unexpected consequences from this emerging technology to human and environmental health and wellbeing go way beyond harmful viruses.” He wants teams to bring in people with “a deep knowledge and cross-disciplinary understanding of how to successfully navigate highly disruptive and deeply complex technology transitions”. That is his own field, which he advocates for here. - The political moment. He notes that responsible AI is “seemingly going out of fashion at lightening speed at the moment while concepts like “permissionless innovation” take their place”. A footnote says he had planned to write that week about “the upsurge in adoption of the idea of permissionless innovation”. He closes with a sharp line on industry culture: “a world where the name of the AI game is increasingly to go fast and break things in the hope that someone else will clean up the mess.”

Analogies and past technologies. - Asilomar 1975 (recombinant DNA). A literal precedent, since the field is the same, and a structural model of scientists taking “collective responsibility” for a technology’s safe and ethical development. He calls it “a landmark in establishing the foundations of responsible and beneficial genetic manipulation”, which is an unambiguously positive reading of Asilomar. He says the 1975 challenge now looks “much smaller” in scale, and hopes the 2025 “Spirit of Asilomar” meeting will take on AI and biology and Evo 2. - Venter’s 2010 synthetic organism. A literal precedent showing that a designed genome can be turned into a living organism. - DNA coding and digital coding (footnote). He compares the two and notes that digital creativity produced “everything from elegant but devastating computer viruses to the foundations of blockchain”. This is an implicit point about dual use. - Nanoscale science. DNA origami and nanoscale machines appear as parts of a converging technology set, not as risk analogies.

Views. - Nature of AI: next-token prediction across symbol systems (text, code, DNA). It emulates understanding without being grounded in it, yet its capability goes beyond humans’. - AI risk: dual-use biology is the obvious risk. He says the less obvious, wider unintended consequences matter more. - Companies and culture: he credits the researchers but criticises an AI culture of “go fast and break things”. - Governance: scientific self-governance on the Asilomar model, plus cross-disciplinary transition expertise. He regrets that responsible AI is giving way to permissionless innovation.

Quotes. - “DNA as it appears in biological organisms is just another type of language that connects a sequence of symbols with functional outcomes.” - “It is, in effect, a DNA-based stochastic parrot” - “Despite ideas like responsible AI seemingly going out of fashion at lightening speed at the moment while concepts like “permissionless innovation” take their place” - “the name of the AI game is increasingly to go fast and break things in the hope that someone else will clean up the mess.”

2025-01-19 — wef-global-risks-2025 — “WEF: “The global outlook is increasingly fractured”“#

Provenance. His own prose. Quotations come from the WEF Global Risks Report 2025. The ranking tables are images. His image caption jokes that the AI-generated map’s distorted geography is “just AI”.

Argument in his terms. - His relationship to the report. He has followed the report and “occasionally” contributed to its expert survey “for almost as long as it’s been around”. “despite its flaws” he finds it “an essential resource for understanding the emerging global risk landscape”, and has long been interested in “how technological risks are framed and ranked”. He points readers to his two 2024 posts on the report’s technology-risk history and the 2024 rankings. - The data he highlights. - “Adverse outcomes of AI technologies” ranks 32nd on the 2-year outlook and 6th on the 10-year outlook (29th and 6th in 2024). - Misinformation and disinformation, “increasingly being driven by AI”, ranks 4th. - “adverse outcomes of frontier technologies” (brain–computer interfaces, biotech, geoengineering, quantum) ranks 33rd of 33 over 2 years and 23rd over 10. - Enabling technology versus specific risks. He reads the low AI ranking partly as a framing artefact: “while AI is an enabling technology, it’s the specific risks it contributes to that are grabbing attention.” Splitting AI out of the frontier-technology category in 2024 also helps explain its low place. - Main claim: the experts underrate AI. He expects critics to read the low ranking as a sign “that saner minds are seeing through the speculation”. He disagrees: “I suspect it’s more likely that most of the experts polled for the report simply do not grasp how disruptive the technology may turn out to be, and how fast it’s moving.” The report itself warns against complacency, and he cites that. - Critique of method. Survey opinion “tend[s] to regress to the mean in terms of understanding and perception.” He sees both sides. Regression to the mean “helps avoid a dangerously high focus on speculative risks at the expense of more immediate and plausible concerns”. But it also devalues “risks that are poorly understood by a broad base of mainstream experts”. Frontier technologies and advanced AI fall into that category. The coming disruptions will be ““I told you so” moments for people immersed in the various fields, but blindsides to many other professionals and experts”. - Conclusion. The report shows “a deeply complex and interconnected risk landscape”, and a positive future will need “deeply coordinated global action at unprecedented levels”, including on AI and frontier technologies. - How firm. Hedged (“I suspect”, “probably”), but the direction is clear.

Concepts. Risk landscape. Framing and ranking of technology risks. Regression to the mean in aggregated expert perception. AI as an enabling technology whose risks show up under other categories. Insiders’ foresight against mainstream experts’ blind spots. Interconnected risk that needs coordinated global action.

Expertise and publics. An important nuance on expertise. He trusts people “immersed in the various fields” over a broad panel of “mainstream experts” for risks that are new and fast-moving, while granting that the broad panel protects against speculative excess. The same week, he frames his PhD-experiment sub-question around “thinking at the edge of the distribution rather than in the mainstream” (2025-02-09).

Continuity. This continues his 2024 WEF critique of expert surveys as a “risk perception zeitgeist” rather than foresight (see B13). It is more explicit this year that the experts are underrating AI.

Quotes. - “most of the experts polled for the report simply do not grasp how disruptive the technology may turn out to be, and how fast it’s moving.” - “the opinions it represents tend to regress to the mean in terms of understanding and perception.” - “it does have a tendency to devalue risks that are poorly understood by a broad base of mainstream experts.” - ““I told you so” moments for people immersed in the various fields, but blindsides to many other professionals and experts”

2025-02-04 — openai-deep-research-ai-scholarship — “Does OpenAI’s Deep Research signal the end of human-only scholarship?”#

Provenance. - His own prose: the main text, the Feb 4 update and the footnotes. - AI-written: the abstract of the attached paper (Navigating Advanced Technology Transitions: Toward New Frameworks for Human Flourishing in an Era of Radical Convergence), and Deep Research’s replies in the transcript. Neither is evidence. - His prompts in the transcript are his own words and are useful evidence of how he frames “advanced technology transitions”. - Mark Daley’s line (“you have to use tools like this to stay in the game”) is Daley’s.

Argument in his terms. - The core claim. Even with “the debates and discussions around whether current advances in AI are substantial, or merely smoke and mirrors hype”, he finds himself “beginning to question the value of human-only scholarship in the emerging age of AI.” He predicts that “reasoning research agents like this will eventually make non AI-augmented scholarship and research look intellectually limited and somewhat quaint.” - What Deep Research is. “in essence a “research agent”” that “can iteratively reason its way through complex questions”, draws on the web and checks its work. Using it “feels like giving a team of some of the best minds around PhD-level questions”. In some respects it goes beyond most PhD students. This is not because they lack intellect, but because it can synthesise across “a vast array of disciplines”, past “the restrictive training and perspective that comes with disciplinary boundaries”. This fits his long-standing complaint that disciplinary fields are too narrow. - The test. He asked for a framing paper on “navigating advanced technology transitions”, his own area, which he describes as “not a widely recognized area of scholarship”. It took about 12 hours. His verdict is “Brilliantly frustrating”. - Strengths: good narrative arc and cross-domain integration. Much of it matched his own thinking and his colleagues’. He was surprised by its “focus on responsible innovation, the need to draw on non-technical areas of knowledge”. - Weaknesses: the prose is “somewhat monotonic” and lacks “the subtle nuances and modulations of expression that a good human writer would insert”. It is too brief in places. It says too little about “what makes this point in human history unique”. It deferred too much to his prompts rather than “thinking” for itself. It is “not as generative and insightful as a leading transdisciplinary and polymathic scholar”, “But such scholars are rare”. - Verification failure: the final paper dropped its in-text citations, which is “a big deal as there’s no way to validate claims”. Later revisions got worse. He adds that most citations were valid: “we’ve come along way from hallucinated sources and references!” - Speed has made his own work obsolete. He had used o1-pro the day before to draft seven papers on technology transitions. Now “pursuing them any further would be a waste of my time — such is the rate of AI progress at the moment.” - The seed of the “artisanal intellectual”. “it’s hard to imagine serious scholarship without tools like this surviving in the future.” Then: “Unless, that is, the future value of the scholarship lies primarily with the process of creation rather than the relevance and impact of what is created.” - Where humans stay in the loop. The tool “does not eliminate people as it’s only as good as the directions its given, the sources it has access to, and the subsequent human assessment of what it produces.” “as a scholarship and research accelerator, it’s a potential game changer.” - Institutional stakes. Daley says tools like this are needed to stay “in the game”. Maynard goes further: soon you will need them “to even get into the research and scholarship game”. He adds: “And if you find that a little disconcerting, you probably should.” - A hedge the next day (Feb 4 update). The quality is “very variable”, and some runs are “just mediocre — and with plenty of false references”. “Seems that things are still being worked out.”

His own framing of advanced technology transitions (from his prompts). - The concept means “we need new models, frameworks etc to understand how to navigate transformative, convergent and synergistic technologies”. - Its technology set is AI, neurotechnologies, gene editing, synthetic biology, Michael Levin’s work, biohybrids, automation, robotics, nanoscale science and technology, and “synergistic convergences between them”. Hidden trends are “equally as important”. - It must be historically grounded but “tackle what is unique about this point in human history and whether this justifies new thinking”. - It includes “the nature of value within a human context, even down to what it means to be human”. - He insists on humility: “No recommendations at this point – remember the humility bit.” - The disciplines he wants drawn on are philosophy, STS, responsible innovation, policy and governance, sociology, behavioural science and futures studies.

Tone and language. He uses a lot of relational, human-like language for the tool: “a talented polymathic AI”, “a wonderful synthetic research partner”. He “felt bereft” when the session broke down. He also says it produces “simulated thinking and scholarship”. This is his most enthusiastic AI-augmentation writing so far.

Views. - Nature of AI: reasoning agents are a real step change, not hype, though flawed and unreliable. - Scholarship and higher education: human-only scholarship may become an anachronism, and institutions that do not adopt these tools risk being shut out. - Risk: the main risks he names are verification (citations) and reliability. He flags his unease in one line and does not develop it.

Quotes. - “I suspect that reasoning research agents like this will eventually make non AI-augmented scholarship and research look intellectually limited and somewhat quaint.” - “Unless, that is, the future value of the scholarship lies primarily with the process of creation rather than the relevance and impact of what is created.” - “it’s only as good as the directions its given, the sources it has access to, and the subsequent human assessment of what it produces.” - “And if you find that a little disconcerting, you probably should.”

2025-02-09 — can-ai-write-your-phd-dissertation — “Can AI write your PhD dissertation for you?”#

Provenance. - His own prose: the main text, his description of the process and the footnotes. - His own words: the driving question (“Can humanity survive the emerging polycrisis of environmental stressors, sociopolitical upheaval, and technological acceleration?”) and its three sub-questions. - AI-generated and not evidence: the dissertation’s abstract, the roughly 400-page dissertation, and the “Originality Score” block (~70%).

Argument in his terms. - Verdict. “To be very clear, this is not a PhD dissertation — AI isn’t there yet. But it is frighteningly close, and in some respects this document exceeds the depth, breadth and insight that many human-written dissertations demonstrate.” - Flaws he names. Deep Research is “lousy — really lousy” at in-text citations and bibliographies. It relies on easy websites rather than primary sources (“it even used Goodreads at one point!”). He is “not sure how far I trust its discernment and critical “thinking” as it selects and weighs sources”. It has a weak through-line, repetition, and “a lack of originality in many places”. - The comparison of effort. He describes his process: a roadmap first, then eight chapter prompts, about 3 hours of AI “thinking time”, and 4 days in all. He compares the result to an in-depth analysis “a research consultancy or an organization like the National Academies would compile”, which would take one researcher at least 3 months. - Rethinking the PhD. “if I was doing my PhD now without the help of AI, I would be deeply worried.” “we will have to critically rethink the purpose and value of a non-AI augmented PhD.” - He sets out two views of what a PhD is for. The first is “more about the journey than the new knowledge it produces”, in which case “not using AI would make sense, at least to a degree”. The second is “pushing the bounds of what is known”. On the second view, an AI-augmented PhD “has the potential to generate more new knowledge, and to generate it much faster”. - This holds most clearly for scholarship that does not involve lab work or human subjects. Even there, AI is getting close to suggesting research strategies “or even carry out research on its own”. - Engaging sceptics. He expects readers who think “any semblance of scholarship exhibited by AI is a mere smoke and mirrors illusion”, and asks them to read the dissertation “with an open mind”. - The limits of AI (footnote 2). He is explicit: “I don’t think we are heading for a future where AI can independently research and write a PhD dissertation — that would require independent intent and understanding on the AI’s part.” In the same breath he coins the artisanal intellectual idea: AI as “a powerful catalyst and accelerant in research that will relegate human-only research to a class of artisanal intellectualism where the primary purpose is the provenance and process, not the product.”

Concepts. - Artisanal intellectualism (first use): human-only research valued for provenance and process, not product. - AI-augmented versus non-AI-augmented PhD. - Polycrisis: his chosen question, bringing together environmental, sociopolitical and technological acceleration. - “thinking at the edge of the distribution rather than in the mainstream” (his sub-question 2). This echoes the WEF post’s point about regression to the mean.

Views. AI lacks intent and understanding, so it is an accelerant, not an independent scholar. The purpose of higher education and the doctorate must be re-examined. He voices concern for students (footnote 1 apologises to his current student).

Quotes. - “it is frighteningly close, and in some respects this document exceeds the depth, breadth and insight that many human-written dissertations demonstrate.” - “that would require independent intent and understanding on the AI’s part.” - “relegate human-only research to a class of artisanal intellectualism where the primary purpose is the provenance and process, not the product.” - “we will have to critically rethink the purpose and value of a non-AI augmented PhD.”

2025-02-16 — the-artisanal-intellectual-in-the-age-of-ai — “The Artisanal Intellectual in the Age of AI”#

Provenance: mixed. - His own prose: the introduction and the closing “Notes”. - Co-written with AI: the central article, “The Artisanal Intellectual in the Age of AI”, with its sections on origins, contemporary trends, future speculation, optimistic versus cautious visions, and conclusion. Deep Research drafted it. Maynard then went through it “line by line and link by link”, tracked references to primary sources, added sources and context, and smoothed the style. Individual sentences cannot be reliably attributed to him; for example, the article refers to “Maynard” in the third person in places, which suggests the draft began as AI text. The article is therefore treated as what he chose to publish and stand behind, not as his own prose. - AI-generated and unchecked: the attached “Missing Perspectives” paper, which he flags with a “caveat emptor”.

Argument in his terms (intro and notes). - The tool is making him “rethink what it means to be someone who makes a living by thinking.” - AI-only intellectual labour is “less interesting to me than what a person and a powerful reasoning/research AI might be able to achieve together.” - “For the first time in my experience as a pretty well established and respected academic, it feels like AI is capable of extending what researchers, academics and scholars can achieve beyond anything we’ve seen before.” - The result is “substantially better than anything I could have pulled together on my own in the time I had”, better than the AI’s draft, and “genuinely generative”. The process “significantly advanced my thinking” about how an academic working with AI “can far surpass what either could achieve on their own.” - He admits that “The idea of the artisanal intellectual was a bit of a throwaway at the time”, and says he has since used it in panels and on the podcast. - The concept shifts. In the notes, the exercise “demonstrated the value of the craft of the “artisanal intellectual” (i.e. me in this context) in assessing and building on the raw material produced by Deep Research.” So within a week the artisanal intellectual has moved from someone who does human-only work valued for its provenance (2025-02-09) and “someone who thinks without using AI” (the 2025-02-11 podcast show notes) to the human craftsperson inside an AI-augmented workflow: judging, checking, adding. The meaning is not yet settled. - Closing judgement. After reading Deep Research’s “Missing Perspectives”: “it’s hard to imagine a future of effective intellectual labor that is not AI-enhanced.” - Asymmetry. It took him “several hours” and Deep Research “a few minutes”. He was impressed that the tool found and cited his own day-old podcast and post without prompting.

Themes in the co-written article he endorsed (attribute with care). - Intellectual craftsmanship: C. Wright Mills, Polanyi’s tacit knowledge, and Dominic Boyer’s 2002 definition of the artisanal intellectual as a knowledge-maker with an “immediate and sensuous relationship” to epistemic forms. - Heidegger’s enframing and “standing-reserve”, and Sennett’s craftsman. - Deskilling and alienation from “academic personhood” (Watermeyer et al.), Lindebaum’s “hard thinking”, “algorithmic conformity”, and a “slow scholarship” movement. - The “cyborg craftsman”, and the point that analogies to the microscope or word processor are becoming “weak”. - Scholars as “knowledge curators”; a dystopian risk of scholars being “proletarianized”. - A new inequality, in which only well-resourced or tenured scholars can afford the slower artisanal route, and academic authority splits in two. - Transparency statements about AI use, and human credibility as “intellectual branding”. - Neither “pro-AI” nor “anti-AI”, but “an age of choice and redefinition”. These fit his wider concerns about formation, equity and being human, but they come from a hybrid text.

Views. Human–AI collaboration is the frontier. Human craft is redefined as judgement, verification and synthesis. He is openly enthusiastic about generative augmentation and puts risk to the craft second.

Quotes (his own prose). - “it’s forcing me to rethink what it means to be someone who makes a living by thinking.” - “it feels like AI is capable of extending what researchers, academics and scholars can achieve beyond anything we’ve seen before.” - “demonstrated the value of the craft of the “artisanal intellectual” (i.e. me in this context) in assessing and building on the raw material produced by Deep Research.” - “it’s hard to imagine a future of effective intellectual labor that is not AI-enhanced.”

MEDIUM#

2025-01-26 — i-asked-chatgpt-to-create-three-video-games — “I asked ChatGPT to create three video games – this is what happened”#

Provenance. His own prose. The code for the three games (Invaders, Jellyfish, Windmill), and the DeepSeek and o1 pro variants, is AI-generated. The quoted prompt is his.

Argument in his terms. - Conversational coding. He borrows the term from David Wright (Forbes, 2022). The user “just need[s] some imagination and an ability to communicate in plain everyday terms.” “This is a huge deal. It begins to redraw the line between what someone can imagine, and how that’s recreated on a digital platform.” He treats it as a creative opening for non-experts, like AI image and video generation, and says outright that he is not assessing it for professional coding. - Limitations. - ChatGPT kept dropping large sections of code. - It got “locked into” an approach. - Above all, it lacks “spatial intelligence, meaning they don’t have a learned understanding and intuition of how things behave and interact in an environment constrained by physical rules.” - The games are “limited by ChatGPT’s simulated understanding of the world.” - Asked for an original game, “It failed miserably”. But Windmill, which grew out of one of its ideas, came “from somewhere deep in ChatGPT’s synthetic imagination”. - Four lessons. - Speed and ease. - It is “not professional-level coding, but I don’t care”. The joy, and his memory of coding in BASIC as a teenager, matter. - Conversational coding has a way to go but will be “profoundly transformative”, because it means “essentially creating functional machines through conversations with AI”. - It shows “how far generative AI has come from the idea that it just regurgitates smooth but vacuous text”: “the generative AI of today is most definitely not the generative AI of 2022.” - DeepSeek R1 test. R1 was fast and its visible reasoning was “intriguing to watch”, but it failed on long code, where o1 pro succeeded. - Footnotes. ChatGPT was “a good tutor” when it walked him through small changes. Coding’s flow state was “scary”, nearly “bring[ing] everything crashing down” in his schedule.

Views. AI as a language-driven interface that lets ordinary people build machines, which prefigures his Evo 2 framing of DNA as another language. He is enthusiastic about democratised creativity and joy, while noting that the models’ understanding is simulated and not grounded in physical reality. There is no risk analysis in this post.

Quotes. - “It begins to redraw the line between what someone can imagine, and how that’s recreated on a digital platform.” - “they are still limited by ChatGPT’s simulated understanding of the world.” - “the generative AI of today is most definitely not the generative AI of 2022.”

2025-01-30 — ai-at-a-crossroads — “AI at a Crossroads: The Unfinished Work of Aligning Technology with Humanity”#

Provenance. - AI-written and excluded as evidence: the title, subtitle and whole article body, written “entirely by ChatGPT” (GPT-4o) from the International AI Safety Report and the Vatican’s Antiqua et Nova. ChatGPT also wrote the Midjourney prompt and picked the image. - His own words: the italic preface, the “Postscript”, “A word on the process”, and his prompts.

What he says in his own prose. - Breaking his rule. “I’ve resisted doing this for so long”. This is a deliberate first: a whole analytical article written by AI and published under his masthead. He gives four reasons: - there is an important synergy between the two reports; - synthesising them would take a lot of time; - ChatGPT “articulated 80% - 90% of what I would have teased out … but far more articulately than I could have done”; - above all, the insights need to be read “irrespective of whom or what wrote them”. He adds that the ideas are “ultimately drawn from human authors” but that “generative AI was instrumental in revealing insights”. - Writing as craft and identity. “As a writer, using generative AI to create copy scares me profoundly. I know it can write better than me, faster than me.” His craft is “the very human piece of me I bring to my writing”, which includes “the idiosyncrasies that make my writing an expression of who I am, and even the flaws”. “I hold on to my craft in the belief that there’s unique value to what I write because it’s written by a person.” This anticipates the artisanal intellectual posts two weeks later. - Self-description in his prompt (his own words, useful as his stated position): “I am well known for my informed and nuanced reflections on the responsible and beneficial development of emerging technologies, and thinking around navigating advanced technology transitions.” He does “not write pieces that are polarizing, that are preachy, that push an ideology or an agenda”, and places “human wellbeing and flourishing at the heart of my work.”

What he chose to publish (ChatGPT’s text, endorsed as “profound and timely”). - AI safety and AI ethics “must evolve together”. - Intelligence as function versus intelligence as relational, embodied and ethical. - “The Myth of Neutrality” (a section heading). - The need for “moral imagination”. - Theologians, artists, anthropologists and humanists need a seat at the design table. - AI as a “civilizational question”. These match his long-standing positions, but they are not his prose. What is notable is that he engages a scientific safety report and a Vatican document side by side.

Quotes (his own prose). - “As a writer, using generative AI to create copy scares me profoundly. I know it can write better than me, faster than me.” - “I hold on to my craft in the belief that there’s unique value to what I write because it’s written by a person.” - “I do not write pieces that are polarizing, that are preachy, that push an ideology or an agenda”

LOW / NONE (one line each)#

• 2025-01-14 — are-your-cars-autonomous-features-unsafe (low). Modem Futura episode promotion with ADAS expert Brunno Moretti. His short intro flags a “hidden” danger: driver-assist sensors often not recalibrated after crashes. Podcast, not examined further per the user’s instruction.
• 2025-01-21 — is-this-the-year-of-agentic-ai (low). Modem Futura promotion (Altman’s “Reflections”, AGI, agents, synthetic personas, the “dead internet”). A footnote describes playful but grounded conversation as his main source of ideas. Podcast, not examined further.
• 2025-01-28 — the-500-billion-ai-gamble (low). Modem Futura promotion recorded amid Stargate, the rescinded Biden AI Executive Order and DeepSeek R1. Title generated by DeepSeek. It mentions responsible innovation and steering AI away from “a whole flock of black swan events”. Podcast, not examined further.
• 2025-02-02 — toe-tapping-ai-pizza-joy (low). A student’s (Jacob Seipert’s) ChatGPT/Suno song about his “Pizza and a Slice of Future” class. The lyrics and the explanation are not his. His only framing is delight, plus a quip about “AI-driven viral class marketing”.
• 2025-02-11 — ai-humility-artisanal-intellectuals (low). Modem Futura promotion with timestamps (Deep Research, “The artisanal intellectual as someone who thinks without using AI”, Heidegger, Reid Hoffman’s Superagency, humility and AI, the value of a PhD). Useful only as a record of the artisanal-intellectual definition. Podcast, not examined further.
• 2025-02-18 — ai-the-medium-is-the-message (low). Modem Futura promotion on McLuhan’s The Medium is the Massage. Timestamps include “Technological playgrounds vs playpens”, “Permissionless Innovation” (twice), Postman, and Socrates on technology. Podcast, not examined further.
• 2025-02-25 — this-is-not-a-podcast-about-ai (none). Modem Futura “AI fatigue” episode on narrative, technology and art, and “the art of doing nothing”. Podcast, not examined further.