How Large Language Models View Our World: Summary & Review of a Big Think Interview with Dan Shipper, CEO and AI Podcaster

     Dan Shipper, the CEO of Every and an AI podcaster, notes that AI is being used in various ways. He distinguishes the way computers and science see the world, both of which are trying to:

 “…reduce the world into a set of really clean universal laws that apply in any situation. If X is true, then Y will happen.”

     He says that large language models (LLMs) see something different a:

“…dense web of causal relationships between different parts of the world that all come together in unique, very context specific ways to produce what comes next.”

     He says LLMs think, or rather operate, much like human intuition in that they are trained by many hours of direct experience.

     First, he talks about rationalism, its journey from Socrates to LLMs, and its limitations. He posits rationalism as the way we see the world. He calls Socrates the father of rationalism because he was the first to “officially” study truth and what makes things true due to his emphasis on inquiry. He points to one of Plato’s dialogues where Socrates was debating with Protagoras whether excellence, sometimes translated as virtue, can be taught. Here, Socrates stresses the need to define what excellence really is before determining whether or not it can be taught. Defining terms and ideas is one way we know things, expressing ideas through words. Thus, in a sense, to define is to know.     

     Shipper says that the rationalist worldview came to the fore in the Age of Enlightenment due to thinkers like Descartes, Newton, and Galileo, who used it to explain the world. With them came the affirmation that one needed to be able to rationally describe ideas, preferably mathematically.

     Shipper sees the social sciences as following the same framework or structure as the physical sciences. However, he acknowledges that there are limitations to doing that since the “soft” or social sciences have subjective aspects that do not lend themselves well to objective analysis. He mentions the current experimental replication crisis in psychology as a kind of proof that the social sciences are less amenable to understanding in the same way as the physical sciences are.  

     AI is similarly difficult to rationalize, he suggests. Here, he notes that there is no universally agreed-upon definition for AI. AI began in the 1950s with a method known as symbolic AI:

“The idea that you could embody thinking in, essentially logic, logical symbols, and transformations between logical symbols, which is, it’s very similar to just basic philosophy.”

     Early AI, he says, could only solve simple problems, not the complex ones that permeate the world. AI follows rules, but with rules, there are exceptions, and those exceptions must be defined, and some may need continuous training in as the exceptions can be dynamic and constantly changing.

     Interestingly, the idea of neural networks was around when AI was young, but wasn’t taken seriously until the 80s and 90s. Neural networks are inspired and informed by the way the human brain works.

“What you can do with a neural network is you can get it to recognize patterns by giving it lots of examples. For example, if you want it to recognize whether an email is important, what you can do is you can give it an example, say here’s an email from a coworker, and have it guess the answer. And if the answer is wrong, what we’ve done is we’ve created a way to train the network to correct its wrong answer.”

“Language models are a particular neural network that operates by finding complex patterns inside of language and using that to produce what comes next in a sequence.”

     The LLMs can utilize the whole internet to accurately predict what is next in a sequence. He points out that the rules for neural networks are non-explicit, or rather, intuitive, and often, only some rules can even be found. Here, he gets back to the similarities to human intuition:

“And what’s really interesting about neural networks is the way that they think, or the way that they operate it looks a lot like human intuition. Human intuition is also trained by thousands of hours of direct experience.”

     We have long-standing metaphors of the human mind being “like” a computer. Shipper sees this as problematic. Rather, he sees rationality as emerging out of intuition. He then goes on to suggest that intuition is beyond the reach of rationalism and, rather than being overshadowed by it, still retains its importance in human knowledge and understanding. He goes back to the dialogue where Protagoras argues with the aid of myths and metaphors, instead of rational definitions. He also says that neural networks are the first thing we have invented that works like human intuition. It is now widely acknowledged that intuition can be another way of knowing in addition to rationalism. He also notes that many of those who use ChatGPT often have noted that they can intuit what it will be good at or not, and when it is hallucinating, much like we can intuit the feelings of a close friend.

“The interesting difference between how a language model sees the world and how a traditional computer sees the world is this: a traditional computer tries to reduce everything into a set of clean, universal laws that apply in any situation — essentially, “if X is true, then Y will happen.” It relies on clear, context-free chains of cause and effect.”

“And what language models see instead is a dense web of causal relationships between different parts of the world that all come together in unique very context-specific ways to produce what comes next. I think language models do something really, really unique, which is that they can give you the best of what humanity knows, at the right place, at the right time in your particular context, for you specifically.”   

     Thus, he says, neural networks and language models are contextual. They are more based on pattern matching and a kind of fuzzy logic, and they use previous experience to predict the future.

“…the way that a more intuitive relational fuzzy pattern matching type experiential, contextual type way of knowing about the world has to be underneath the rational stuff for the rational stuff to work at all. It’s really about recognizing the more intuitive ways of knowing about the world as being the original parent and partner of rationality, and appreciating that for what it is.”   

     The contextual knowledge of LLMs and neural networks can enable hyper-personalized knowledge that can solve specific problems. They don’t need explicit definitions or rational models to be successful. These networks and models are trained to detect and predict. He says they change a science problem into an engineering problem. He thinks that better and more thorough data access and sharing, especially by the Big Tech companies, will lead to more thorough AI training. Shipper thinks we can utilize our intuition and basically put it into a machine that we can pass around.

     Shipper thinks AI will seriously enrich our understanding of ourselves. He sees AI as a mirror and as a metaphor.  However, limiting AI to only what is provable limits it.

“The thing about it that makes it powerful is that it works on probability, it works on thousands of correlations coming together to figure out what the appropriate response is in this one very unique, very rich context. And allowing it to say only things that are provable, obviously begs the question: what is true and how do we know?”

     AI can be messy, but the key thing to know is that with adequate training, it works, though we may not be able to entirely explain how it works.

“Something to remember is each model builds on the models that came before it. They actually have a dense, rich idea of what it is to be good from all the data that they get. They also have a dense, rich idea of what it is to be bad. But in a lot of ways, the training that we’re doing makes them less likely to do any of that stuff.”

“There’s something very practical and pragmatic about, we have a machine, we don’t know fully how it works, but we’re just going to teach it, and we’re going to iterate with it over, and over again until we basically get it to work.”

     Shipper sees AI as analogous to a gardener rather than as a sculptor who creates something from nothing. A gardener gives his plants the conditions to succeed without forcing that success directly. Gardening is much like a model, he suggests.

“AI does a lot of the more repetitive specialized tasks, and it will allow individuals to be more generalistic in the work that they do. And I think that would be a very good thing.”

     In the last section, he mentions the idea that we’re moving from a knowledge economy to an allocation economy. He also says that an allocation economy will require more of a certain kind of worker with specific skills. Those specific skills will often be the skills of human managers, which include knowing one’s human assets and capabilities, such as “knowing what any given person on your team can do, what are they good at, what are they not good at.”

“In a knowledge economy, you are compensated based on what you know. In an allocation economy, you’re compensated based on how well you allocate the resources of intelligence. There’s a particular set of skills that are useful today but are not particularly widely distributed that will become some of the main skills in this new economy, in this new allocation economy. And that is the skills of managers, those are the skills of human managers, which make up a very small percentage of the economy right now. I think it’s like 7% of the economy is a human manager.”

     Some of the ongoing problems that must be solved include how to manage the other humans working on the problem. He also suggests that managing humans and managing a model are similar. Managing a model, like managing humans, requires some intuitive abilities. Shipper mentions an interesting idea that intelligence is like compression in that it compresses a range of possibilities as answers into a small space, or rather, it can find the right answers quickly. He thinks that is partly how brains and consciousness work as well. He suggests that, in a sense, these models may have consciousness, or at least we can benefit from understanding them in that way. Just in case, he says:

“I always say please and thank you to ChatGPT because you never know when the machine apocalypse is going to come.”  

  Well, now I understand neural networks and LLMs better than I did, and I’m grateful, so a worthwhile article. It was actually a transcribed podcast that I worked from. There is a video as well at the link in the references.

References:

 

How Large Language Models View Our World. Big Think. August 27, 2025. How large language models view our world - Big Think

Nodal Seismology: Geophones, Seismometers, and Nodes: Smart Solo’s Nodes, and ACCEL’s Drop Deployed Nodes Offer Better Signal Quality and Cheaper Acquisition Costs

     For this post, I am relying on Smart Solo Scientific’s July 2024 post: Seismometer vs. Geophone: A Guide to Seismic Sensing Tools, as well as a webinar presented by Sercel/Accel. There are two main types of seismic sensors: geophones and seismometers. A seismic sensor is a device that detects and measures ground vibrations. The vibrations, also known as seismic waves, can detect a multitude of ground vibrations from earthquakes to mine blasting. They need a sound source, typically a blast of dynamite or a vibration truck.

 

Geophones

     Geophones are most common in the field, being the standard seismic sensor for geophysical exploration.

“The basic principle is simple but effective: inside a geophone, a suspended mass (usually a coil of wire) moves relative to a magnet as the ground shakes. This movement generates a small electrical voltage that is directly proportional to the velocity of the ground motion.”

“Because they are relatively simple, durable, and cost-effective, geophones are perfect for deploying in large numbers for seismic surveys hunting for oil, gas, and minerals.”

 

Seismometers

     In contrast, a seismometer contains a geophone for detecting and measuring ground motion, but integrates it into a system of analysis. Smart Solo defines a seismometer as:

“…a sophisticated, integrated system designed for detailed seismic analysis. At its heart, a seismometer typically contains a highly sensitive geophone – the component that actually detects ground motion. However, a seismometer is much more than just the geophone itself. It’s a complete instrument package that includes advanced electronics for signal processing, precise timing (often via GPS), data storage, and communication capabilities.”

     The term seismometer can also refer to enhanced methods utilized in geophysical exploration that go beyond mere sensing and become comprehensive data acquisition systems.

Nodes: No More Heavy Cables Means Faster, Cheaper Surveying with Better Signal Quality

     In the past, conducting large seismic surveys involved physically laying out massive amounts of heavy cables that connected the signals together. Now, signals can be detected and measured wirelessly by what are called nodes. This represents a revolution in seismic since it leads to faster, cheaper surveying, not limited by rough terrain. Nodes have also demonstrated better signal quality. Rivers, mountains, and dense forests are no longer impediments to seismic surveys.

“A seismic node represents a significant evolution in seismic instrumentation, essentially an advanced, self-contained form of seismometer. It ingeniously combines all essential components – a sensitive geophone for motion detection, an integrated battery for power, GPS for precise timing and location, and onboard data storage – into a single, compact, and autonomous unit.”

“Crews can now deploy thousands of nodes quickly and easily, by hand or even by drone, across almost any terrains. This has opened the door to ultra-dense surveys that produce incredibly high-resolution images of the Earth’s subsurface, something that was previously unimaginable.”

     Smart Solo has a 2D node version, the IGU-16, and a 3-component (3C) smart seismic sensor, the IGU-16HR 3C, that collects data in three dimensions for a 3D seismic survey. 

     The company can also pair the nodes with its data logger, a 3-Channel Intelligent Monitoring Unit, the IMU-3C, described below.

     Smart Solo also notes that the massive data sets provided by nodal seismology are very amenable to machine learning methods of analysis.

“The future of seismic exploration and monitoring is all about data—more of it, and better quality. The massive datasets generated by large-N (large number) nodal surveys are perfect for feeding into AI and machine learning algorithms. These advanced tools can help automate the process of picking seismic arrivals, filtering out noise, and even identifying patterns that could predict geological hazards.”

     Company Sercel sells what it calls its Accel drop nodes, nodes that can simply be dropped into place, with spikes to hold them in place or not. These are hand-sized. 

     The company says its nodes can save about 30% in opex. Fewer personnel are needed in the field for deployment and retrieval. Testing has confirmed that its drop nodes have signal quality as good as buried or spiked nodes. The company also claims that its drop nodes can be less affected by noise from wind and rain than buried or spiked nodes. Its nodes are integrated into its QuietSies and Pathfinder data technologies. The EU-based company is currently testing and deploying its nodes in Texas, among other places.

 

    

References:

 

Seismometer vs. Geophone: A Guide to Seismic Sensing Tools. July 4, 2025. Seismometer vs. Geophone: A Guide to Seismic Sensing Tools - SmartSolo

Accel: Accelerate your operations. Sercel. Accel | Sercel

Discover Accel, the world's first drop node. Accel/Sercel. Webinar. November 20, 2025.

 

 

Gunnison Copper’s Johnson Camp Mine Near Tucson, Arizona: Set to Be First New U.S. Copper Production in a Decade: It Utilizes Nuton’s Bioleaching Technology, as Do Other Projects in the Region

     Copper is one of the new minerals put on the updated USGS critical minerals list for 2025. The U.S. has not had new copper production in a decade, and a new, or rather a revitalized, mine reopening very soon will change that. Gunnison Copper’s Johnson Camp Mine near Tucson is expected to begin production in the coming days. Production ceased at the mine in 2010 when the copper ore grade was deemed too low in concentration to produce. However, newer extraction methods are expected to revitalize production. The mine restart, in collaboration with Rio Tinto’s Nuton venture, will utilize microbes and an acid solution to extract the copper and then transform it into cathodes.

     Arizona already produces 70% of U.S. copper. It has a long history of successful copper mining. Now, the state is at the forefront of domestic copper production revitalization.

     The Wall Street Journal’s Ryan Dezember, who wrote a nice article about the mine, notes:

“Gunnison started selling cathodes made using conventional heap-leaching methods of Johnson Camp’s oxide ores in September. The first batch of copper extracted from its sulfide ores using the Nuton technology is expected in the coming days. Ramped up, Johnson Camp should annually produce 25 million pounds of cathode.”

     While it is very true that domestic copper production can benefit from tariffs on imported copper, already at high levels, those tariffs harm buyers and consumers. Copper prices have risen since U.S. tariffs were enacted, but had been rising for years, as shown below.

     The U.S. has abundant low-grade copper ores, but it remains challenging to produce them economically. Nuton’s bioleaching technology is being utilized to extract copper from low-grade ore in new ways.

     Dezember goes on to talk about the very long times it takes to build a mine, to get from discovery to production, which I wrote about recently here, and the need to reduce those timeframes. Some projects in the region have been fighting legal challenges for more than a decade.

     Taseko Mines is trying a different approach at its Florence project southeast of Phoenix, which is utilizing a technique of uranium miners and pumping acid down into wells to separate the copper below the surface, known as an in-situ method. Taseko expects Florence to eventually produce about 85 million pounds of cathode per year.

“Of all the copper that exists in the ground globally, 70% of those resources are comprised of primary sulfides, this mineralogy that we’re looking to unlock,” said Nuton chief executive Adam Burley. “That size of prize is enormous.”

     Taseko has been exploring the idea of in situ mining methods for thirty years, but only now is moving toward production.

“The Nuton name plays on Isaac Newton, the alchemist, as well as the hunt for “a new ton” of copper, which had become elusive via deal or discovery, Burley said.

“We needed a different model because the one that we’d been trying for years wasn’t delivering the results we wanted,” said Clayton Walker, who is responsible for the growth and development of Rio’s copper business in the Americas.

     The Johnson Camp mine is a good place to test the Nuton bioleaching process since it already has two key components: a heap leach pad and a solvent-extraction electrowinning plant, where the copper dripping in solution from the ore is plated onto the cathodes.

     The sulfide ores are crushed, coated with bacteria and acid, then piled onto a new heap leach pad. 

     Nuton also has partnerships with several other copper mines in the region. The company also plans to reopen a copper mine that closed in 1984 and apply the process. This is Arizona Sonoran Copper’s Cactus project. It expects to begin construction in 2027 with the first cathodes made in 2029. That is extremely quick in terms of mining times and underscores the advantages of revitalizing older works over new projects in terms of time to production.

 

     

References:

 

Can Arizona Miners Unleash an American Copper Boom? Ryan Dezember. Wall Street Journal. December 2, 2025. Can Arizona Miners Unleash an American Copper Boom?

The Progress Movement: Culture Feeds Progress, the Importance of Immigrant Entrepreneurs, and How Pragmatists and Purists Work Together: A Review of Three Articles in Big Think’s “The Engine of Progress” Special Issue

      The Progress movement, akin to the Abundance movement, studies past progress to inform and enable the progress happening in the present. I’m still learning about both, but I also know there are other groups that study progress, including the Libertarian Cato Institute, which runs Human Progress, which puts out some very good articles about progress and what makes it happen. It is a very informative online source. These articles from Big Think, another very good online source for many kinds of information, come from the Roots of Progress Institute’s Progress Conference 2025, held in Berkeley, California, in October.  

 

1 - Why culture may be our most powerful lever for progress

     In the first article - Why culture may be our most powerful lever for progress, Beatrice Erkers argues that progress begins with culture. She argues that culture often sets up technological breakthroughs, often by inspiring the right people to ask the right questions. She sees culture as infrastructure. It is below hard infrastructure like roads, bridges, and buildings, and soft infrastructure like laws and institutions.

“Hard infrastructure builds the roads. Soft infrastructure sets the rules of the road. Culture decides which destinations are worth visiting.”

     She calls culture invisible infrastructure and defines it as:

“The stories, narratives, and memes that determine which futures feel plausible and worth pursuing.”

     She thinks she sees Artificial General Intelligence (AGI) following this pattern. She invokes Scientist Michael Nielsen’s idea of hyper-entities, which he defines as “imagined hypothetical future object[s] or class{es} of object{s},” or as something that exists in consciousness before it exists in reality. Past examples include the internet, submarines, and cellphones. Currently, some possible future examples include AGI, space elevators, Martian settlements, the Singularity, and universal quantum computers. She sees William Gibson’s 1984 novel Neuromancer, which coined the term “cyberspace” as an example of a hyper-entity.

     She argues that the Green Revolution arose from stated desires to get hungry people fed. She also notes that not all hyper-entities are benign; some, like bureaucracy (implemented for better coordination but often backfiring) and our outdated education system, may amplify inertia rather than a good idea. She comes back to AGI, which she sees as a powerful hyper-entity of our time.

“It doesn’t exist yet, but the story of AGI already shapes budgets, regulation, research agendas, and the careers of thousands. It may never arrive, but for now, the idea itself is acting as cultural infrastructure, organizing effort across society.”

     While hyper-entities are slow-moving and may exist for decades before becoming real (or not), a smaller packet of culture, known as the meme, a term coined by biologist Richard Dawkins in 1976, is easy to transmit as they often come in simple, contagious forms, such as jokes, slogans, and pictures that spread ideas and influence others.  

     Eskers notes that culture magnifies indiscriminately so that the good and the bad can get amplified through hyper-entities and memes. She cites irrational public fear of genetically modified foods and nuclear energy as examples where culture inhibits progress. She notes that culture can be an unpredictable kind of infrastructure, akin to weather. It can amplify both fear and hope. She sees optimism as passive and hope as an active force. She cites the progress movement, specifically Jason Crawford, founder of the Roots of Progress Institute, who contrasted different kinds of optimism. He noted that blind optimism is not a cure for blind pessimism. He noted that complacent optimism, assuming something will happen automatically, is not as flexible nor durable as pragmatic optimism, which asserts that we must work in order to achieve the results.

     Interestingly, she notes that utilizing cultural references often leads to grabbing those close at hand, which are often bleak. Here, she cites the prevalence of dystopian movies, books, art, etc., and the lack of hopeful narratives. Regarding AGI, there are both dystopian and utopian views. She prefers neither, but a hopeful one.

    In the final section, Investing in Culture, she turns to the subject of funding new ideas. She advocates for funding hopeful narratives, though that seems vague.

“AGI may not exist, but the story of AGI already mobilizes billions of dollars. That shows how culture lays the groundwork before a technology ever arrives.”

“The problem is that most of the cultural ground we’ve laid is dominated by dystopias, and that imbalance won’t correct itself. It needs deliberate work, not only from institutions and funders, but from all of us. Culture isn’t just made in conferences or boardrooms. It is shaped in the stories we tell, the art we share, and the memes we pass along. Everyone participates.”

“Economic infrastructure is the bridge that carries culture into the world: from invisible symbols, to soft institutions, to hard technologies. Culture is unpredictable, double-edged, and too important to ignore. If we want progress to keep moving, we need to balance our culture of fear with visions of hope, and then back those visions with the resources they need to become real.”

     This last paragraph reminds me of Ted Nordhaus and Michael Shellenberger, who in their essay and book “The Death of Environmentalism” argued for a politics of possibility instead of a politics of limitation and grievance.

 

2 - Physical dynamism and the immigrant’s edge

     Afra Wang presented this topic at the Roots of Progress Institute's Progress Conference 2025. Wang ties the Progress movement to the Abundance movement, suggesting they are the same. She posits a societal yearning for “what writer Dan Wang calls “physical dynamism,” the tangible acceleration of the material world that makes tomorrow feel radically different from today.”

     She mentions several bold business and engineering initiatives spurred by immigrant entrepreneurs and technologists.

“Amid all this intellectual diversity, a pattern emerged that the conference rarely named explicitly: The most audacious physical dynamism projects are led by first-generation immigrants. Look closely at the progress movement’s architecture, and you’ll see immigrants everywhere, including at the foundation.”

     She goes on to list some of these immigrant leaders: Patrick Collison, the Irish immigrant who co-founded Stripe and launched “progress studies” in an article in the Atlantic with economist Tyler Cowen in 2019, Heike Larson, who co-founded the Roots of Progress Institute, and grew up in Germany, and Dan Wang, a Chinese Canadian mentioned above, and author of Breakneck: China’s Quest to Engineer the Future. The article’s author, Afra Wang, is a Chinese immigrant to the U.S.

     She notes that immigrants often bring new and unique perspectives:

“Immigrants not only bring non-American-centric mindsets and building speed. They carry lived experiences of both progress and collapse, which breeds a particular kind of vigilance. They see not only what America could become, but what it risks losing. Many arrived believing in promises America made to the world, and now they’re trying to hold the country accountable to those promises, to push it to live up to the dream that brought them here.”

     She also notes what Dan Wang’s book has revealed about China vs. the U.S. in building and engineering: that we in the U.S. are overly burdened with regulatory processes and costs.

“It reveals new facts about China, but also articulates what American builders already feel viscerally: that the U.S. is trapped in lawyerly procedures while physical dynamism accelerates elsewhere.”

     Here she compares the Chinese and California high-speed rail projects. California has spent $120 billion over 17 years and is still a long way from being complete, while China built a comparable project in three years for $40 billion. Surely, we can do better. Permit reform is one sure need. The end result is that America is less dynamic than China in infrastructure building.

     She goes on to talk more about the conference, attended by many engineers, founders, scientists, and policymakers, and about the idea of American dynamism. She emphasizes the American experiment and the idea that America is a flexible and accommodating idea that immigrants can embrace.

“America cannot reindustrialize without immigrants and immigration reform. It also cannot do it alone, without borrowing wisdom from countries like China, Korea, Singapore, etc. I envision a different future: one where immigrants and transnational talent, people carrying know-how and the secret formulas across borders, fluent in multiple systems, fuel this dynamism. Radically pluralistic, ambitious, grounded in the lived experience of people who urge America grow into something it’s never quite been.”

     She praises talks that pondered the importance of “industrial literacy,” increasing knowledge of how the world actually works. She also emphasizes the importance of optimism and a move away from the techno-pessimism that has not been helpful.

“This is a simple yet encouraging belief that progress is good, necessary, and achievable.”

 

3 - How Pragmatists and Purists work together to change the world

     The last article is by Jonny Thomson. He sees activism as a spectrum with purists at one extreme and pragmatists at the other. I prefer to be a pragmatist without being an activist, and I certainly don’t see pragmatism as an extreme. He suggests that when purists and pragmatists work together, activism can be functional and effective.

“Pragmatists see progress in terms of raw numbers. Purists see it in terms of an absolute criterion.”

     He suggests that pragmatism is tied to a “consequentialist” philosophy, which means that measurable incremental gains are acceptable and desirable. Purism, or absolutism, in contrast, is concerned with strict positions with no compromise on certain topics. For example, no amount of slavery or marital rape should be acceptable. Most people can agree with that fairly easily. I would add that the same is not true for many other debates, such as environmental impact or resource use. We have to accept some levels of it. He introduces some hypothetical arguments, including the effects of a piecemeal approach to slavery, for instance, by banning child slaves and allowing adult slaves. It could have saved many from suffering, but could also have delayed the full banning of slavery and led to more suffering, vs. banning it in full earlier. In this case, it would be a short-term gain but a long-term loss. He says, in that case:

“It isn’t about absolutism vs. consequentialism, but longtermism vs. presentism.”

     I agree that we have a moral obligation to future people. However, I would counter that it would be difficult to determine the possible effects of partial vs. full changes, even in the hypothetical slavery case mentioned above.

     Thomson sees purist and pragmatist activists as potentially complementary.

“The Purist calls out the moral horror. The Pragmatist makes the change possible.”

     I have always thought something similar, that activism is great for drawing attention to something, especially a moral outrage or a great injustice. The purists are good for that. However, in other situations, like environmental impact and climate impact, there is often no real justification for the paths activists want to take, such as bans on legitimate economic activity. Here, activists often espouse a minority position, very loudly.

     Thomson cites the bombings of radical groups in the 1960s as a failure of purists, which made people hate leftist movements. While that is true, I am not sure I agree that pragmatism has a tendency to be harmful to those in the future. It certainly could be in certain situations, but the examples he gave, while suggestive, are not convincing.  I see his point, but I’m not sure if it is a relevant issue. That said, I believe pragmatism, as a method based on utilitarianism, is a viable, useful, and very American way of solving problems. Back in February, I published a post on American Pragmatism.

 

  

References:

 

Why culture may be our most powerful lever for progress: Before we can build the future, we have to imagine it. Beatrice Erkers. Big Think. November 19, 2025. Why culture may be our most powerful lever for progress - Big Think

Physical dynamism and the immigrant’s edge: At the foundation of America’s progress movement are immigrants who still believe this country can build. Afra Wang. Big Think. November 19, 2025. Physical dynamism and the immigrant's edge - Big Think

How Pragmatists and Purists work together to change the world: History shows that progress often depends on activists at both ends of the spectrum. Big Think. November 19, 2025. How Pragmatists and Purists work together to change the world - Big Think

4D CO₂ Plume Monitoring: Hyper-Specialization of Interactive Deep Learning Networks using Transfer Learning. AAPG Academy Webinar: Summary & Review

     This webinar was mainly about the applications of deep learning networks trained on seismic attribute data in order to model CO2 plumes in and beyond the reservoir formation. Monitoring data is used to train and update the model. In this case, the model was trained in the seismic response to gas, or the gas signature on seismic. A key goal of CO2 monitoring is the identification of a breached top seal, and the identification of gas signatures can be very helpful in this regard.

     I was hoping that a recording of the webinar would be available so I could add some slides to this post, but that does not seem to be the case. If it does become available, I will update this post in the future.

  

Equinor’s Sleipner Field is the Example for the Study 

     Equinor’s Sleipner Field in the North Sea was used for the study. It is a long-running CO2 sequestration project that injects CO2 into a deep Jurassic sandstone, above and laterally farther away from the zone where the natural gas and formation CO2 are produced. Every two years, a new seismic survey is shot with the same parameters as before. Top seal integrity is of utmost importance.

 

Data, Training, and Learning

     The data consists of three seismic attributes: amplitude, iso-frequency, and relative acoustic impedance. The amplitude data sets are from pre-injection to current. These attributes are used to train the model. There are three ways in which the model “learns.”

Supervised learning – Audio-Visual model using tuning labels

Reinforced learning – this happens while the network trains. It works on feedback and rewards. It allows the network to abandon false positives.

Transfer learning – this refers to reusing knowledge from a previous network in new data sets

     Networks are tuned when new data comes in. The process is very repeatable.

     First, an In-Situ Baseline Model was created from Pre-Injection data (1994). The network learns from the tuning labels that are placed on it. In this case, it is tuned to seismic response to gas, which refers both to injected CO2 and pre-existing natural gas in the rocks. The pre-injection model is then tuned with each arrival of new data, and the model is transferred to the new model. Tuning label placement is predicted by the model. It is important not to “overtrain” the model, which will introduce too much statistical bias. Thus, they utilized the three-epoch method to keep the statistical parameters unbiased. The goal is to get just the right amount of training and bias. The deep learning network can capture hidden structures and relationships. After training and tuning, a geometrical representation of the CO2 plume is generated. This is done every two years when a new seismic survey is conducted. The model can differentiate between the two dry gas types after it learns. In this case, the plume extends through time to the south and then to the north. In this case, some possible top seal breaching occurred, according to the model. The model was trained to distinguish CO2 from existing methane in the shallower reservoir through a technique called latent space drift. Some layering was observed in the plume, which is consistent with geology since sandstones are more porous to gas than the shaley layers in between.

Sequential domain adaptation (SDA) – the most effective training occurs when a network is exposed to information for the first time. Overfitting to a dataset introduces bias. SDA allows a network to continually learn, to evolve. Here, what the network is observing is physical, not statistical. Networks with high levels of user interaction and input can be used and transferred to new models with different data.

 

Q&A

Seismic data must be in a specific format. It was run on data sets up to 1.7 TB.

Importance of a pre-injection seismic study: A study should be started before injection to identify in-place reservoir fluids.

Statistical forcing or bias can affect modeling and should be addressed. He uses the three-epoch method to reduce bias. Bias can also be good in that it creates a geologic context = geologic bias according to geologists. Thus, you want some bias in certain directions, but not statistical bias.

Interactive deep learning – if it is there and can be labeled, it can be trained. Gas signatures are an example.

The plume signature is derived by letting the network show the plume.

Can one train all 1994-2008 to be used for 2010? Better to be trained one at a time. If trained together, the results would be messed up – conflicting and incomplete labels would confuse the network. One can do multiple data training selectively – takes out the tuning labels.

    

References:

 

4D CO₂ Plume Monitoring: Hyper-Specialization of Interactive Deep Learning Networks using Transfer Learning. AAPG Academy Webinar. November 19, 2025.

 

 

Hybrid Cooling Technology for Thermal Power Being Explored for Fermi America’s 11GW Data Center Hub

     Billed as the world’s largest private energy grid, the Fermi America data center campus complex and energy hub, in the Texas panhandle, plans for 6 GW of combined-cycle natural gas power and four AP1000 nuclear units. That is quite a lot of thermal power. Thermal power requires significant amounts of cooling water.

     Fermi America recently signed a non-binding Memorandum of Understanding (MoU) with Hungarian power-cooling specialist MVM EGI Zrt. The collaboration will first involve engineering and feasibility studies for a set of indirect hybrid cooling towers. According to Interesting Engineering:

“The cooling design primarily uses air and closed-loop water circulation to reduce evaporative loss. The companies also plan to explore recycled water, underground reservoirs, and solar-covered retention ponds to further conserve resources.”

“MVM EGI has been on the cutting-edge of power cooling for more than half a century, maintaining the heritage of our founders, Professor László Heller and Professor László Forgó whom the high-capacity water-saving dry cooling systems are named after worldwide,” stated MVM EGI P.L.C. CEO Péter Kárpáti.

     According to the press release:

“The collaboration reflects both companies' commitment to transparent, community-oriented development. With billions of dollars in investment and a 99-year lease with the Texas Tech University System, Fermi America's business model is directly tied to the health of the Panhandle and the long-term sustainability of the Ogallala Aquifer. The MOU reinforces that alignment by putting water conservation at the core of the project's cooling strategy from day one.”

     For a project of this scope and size, an integrated water management and recycling system that limits evaporation will be a very important feature. I would guess that cooling water for the data centers could also be a part of the water management system.

     This is a very ambitious project that could become the largest data center campus in the world. However, with very high costs, initially in the billions, and the slow timeline of nuclear deployment, there is still some uncertainty as to how fast the project will proceed. Fermi notes that they expect to begin construction of the first cooling tower in January 2026, so very soon.

  

 

References

 

US: World’s largest 11 GW private energy grid opts for water-saving hybrid cooling. Sujita Sinha. Interesting Engineering. December 2, 2025. US: World’s largest 11 GW private energy grid opts for water-saving hybrid cooling

Fermi America and MVM EGI Announce Water-Saving Hybrid Cooling Agreement for World's Largest Private Energy Grid, Delivering on Promises Made to Protect West Texas Water Resources. PR Newswire. December 1, 2025. Fermi America and MVM EGI Announce Water-Saving Hybrid Cooling Agreement for World's Largest Private Energy Grid, Delivering on Promises Made to Protect West Texas Water Resources

Fermi signs MoU with MVM EGI to develop cooling systems for planned up to 11GW Texas data center campus: Will comprise a series of indirect cooling towers. Zachary Skidmore. Data Center Dynamics. December 1, 2025. Fermi signs MoU with MVM EGI to develop cooling systems for planned up to 11GW Texas data center campus - DCD

Extensive Brazilian Potash Imports are Twice as Carbon-Intense as Previously Thought, According to New Study

     It has often been noted, by me as well, that carbon accounting has many uncertainties. Unless all aspects of a product, from production to consumption, are accounted for, there can be misconceptions. Life cycle analysis is done to track the full carbon intensity. It was once thought that the subtropical soils in places like Brazil would not be able to support expanded agriculture, but with modern methods of supplying nutrients, it can and does. One major component is imported potassium in the form of potash.

     The new research was led by Newcastle University. The main conclusion was that previous estimates did not fully account for previously overlooked Scope 3 emissions from transport and distribution. 

     The analysis utilized a cradle-to-hub approach. According to Phys.org, the bottom line is that:

“…researchers calculate a weighted average carbon footprint of 530.5 kg CO₂‑eq per ton of KCl delivered to 5,563 agricultural distribution hubs across Brazil—almost double the 273.13 kg CO₂‑eq per ton value widely used today in Brazilian agricultural and biofuel carbon‑accounting tools.”

"As a country that imports almost all of its potash, Brazil is a perfect case study to show how much 'hidden' carbon is embedded in fertilizer supply chains," said Professor Oliver Heidrich, corresponding author of the study and Professor of Civil and Environmental Engineering at Newcastle University. "We've shown that fertilizer producers located close to farming regions tend to have a smaller overall carbon emission impact compared to those located in remote regions. Our hope is that this work will drive more rigorous Scope 3 accounting and accelerate the shift toward lower‑carbon potassium sources for Brazilian agriculture."

     The paper’s authors called for stronger Scope 3 disclosure requirements to update carbon accounting for the potash and for other situations where Scope 3 emissions may have been omitted or ignored.

     It is well-known that tropical soils can generally be depleted faster than temperate soils. It was once thought that the soils in places like Brazil would never be able to support long-term intensive agriculture, but the availability of synthetic fertilizers and mined fertilizers like potash has proved that idea wrong. Brazil imports over 20% of global potassium production and relies on imports for about 97% of its KCl demand. The report also notes that some potential domestic sources of potassium have been identified, which would have much lower carbon footprints.

"Brazilian agriculture feeds close to 10% of the world's population, and KCl is one of the agricultural ecosystem's largest embedded sources of emissions," said Cristiano Veloso, Founder and CEO of Verde. "Studies like this help quantify the challenge and show where innovation and investment should focus. Verde intends to be part of the solution by advancing Brazilian‑made potassium specialty fertilizers which, according to our assessments, can cut carbon footprints by up to 89% compared with conventional fertilizer producers operating from remote, carbon‑intensive locations."

     Data tables and figures from the report, published in the journal ‘Resources, Conservation and Recycling’, are shown below.

  

 

 

References

 

Exposing the hidden carbon cost of potash imports into Brazil. Science X staff. Phys.org. December 1, 2025. Exposing the hidden carbon cost of potash imports into Brazil

The true carbon costs of supplying potassium fertilizer to Brazilian agriculture. David A C Manning, Thiago Ribeiro Siqueira, Mohammad Ali Rajaeifar, and Oliver Heidrich. Resources, Conservation and Recycling. Volume 226, February 2026, 108694. The true carbon costs of supplying potassium fertilizer to Brazilian agriculture - ScienceDirect