The more I learn about how AI works, the more I’ve come to prize the value of human creativity in the AI era, especially when it comes to writing.

AI is a revolutionary productivity tool, but it is not a replacement for human knowledge or input. The software can only go as far as the human behind the machine.

AI has many incredible applications, but AI writing is mediocre by design. It’s fine. Not great. Just fine.

It’s suitable for the vast majority of the population, especially since most people currently can’t identify the difference between AI writing, AI-assisted writing, and purely human, unassisted writing. Undoubtedly, there’s a place for all of these calibrations.

But there are a few reasons why AI models can’t, on their own, produce uniquely outstanding literature.

Let me explain:

Think of your favorite chatbot as a student who has read every book in circulation. That’s incredibly impressive. Your student is an encyclopedia of knowledge! You ask him a question, and his default job is simply to guess the next word. He’s very good at guessing, but that’s because, despite his wealth of knowledge, he’s inclined to be probabilistic and not unique. Sad! He knows that among a million writers, there is a significant chunk that will finish a sentence the same way, and that’s the decision he will use in his writing. So by default, he is brilliant but seeks to optimize an average. And he can’t, by design, break out of this mediocrity shell.

Major AI companies, like Anthropic, SpaceXAI, OpenAI, and the like, have hired thousands of people to train AI models and evaluate their answers. Whether they intend to or not, they are eliminating unique answers because they are seeking to install a reward mechanism that encourages the models to play it safe. AI is trained to be a teacher’s pet, one who calculates what the “teacher” (AI model trainer) prefers and decides to write that way to achieve the reward. AI writing is habitually mediocre because of this.

Moreover, all of the major American AI companies (and foreign ones, because they distill information from American labs) train their models using the same internet with similar processes. And when smaller upstarts build their own models, they train them on behemoths like ChatGPT, Claude, and Grok, which is much more cost-efficient, but it leads to the same problem. So their habits become everyone else’s habits. Despite the mess they’ve created, none of the big dogs in AI seem to want to hire great writers and editors to fix that institutional headache, probably because there isn’t much ROI involved. You can’t tell chatbots apart by their prose because it’s one big circular learning environment.

When a human being writes an article without assistance from AI, they inject creativity from their own life experience or imagination into the piece. If you tell an AI to “write an article” about something, it does not have an independent perspective, so it fills the space with pieces of information that have been spliced together from a variety of places. Despite the claims of delusional tech bros on the West Coast, AI models do not have the ability to think independently. AI models are very advanced software projects that operate under a very impressive but very non-human paradigm.

Fixing this problem is mostly above my pay grade. More and more content online is now written entirely by AI, so AI models are now training on AI writing at an increasing rate, which likely means that this mediocre writing problem will get worse before innovation arrives to make it better.

Due to the aforementioned institutional hurdles to creativity, it’s become clear to me that great human writing (with or without AI assistance) is needed now more than ever.

Share

Subscribe now

Bill Gates, the guy who tried to destroy the world over the Wuhan sniffles, says it’s time to panic about AI. Should we?

Share

Subscribe now

The billionaire Microsoft founder is very concerned about AI. He is concerned about our jobs, our families, our ability to think and reason about the issue, and of course, how AI impacts his precious climate change narrative. What he is not concerned about, apparently, is his own track record as a nanny state catastrophist.

In a several-thousand-word essay on his website, Gates tells us that the time to act on AI is now, well before any of his predictions come to fruition or go bust.

We have seen this movie. Gates was one of the chief antagonists for a national shutdown and forced compliance with vax mandates over the COVID-hysteria years. As someone who covered this beat closely, it’s easy to recall the Gates-funded Imperial College model that tried to panic the masses by projecting millions of American deaths in the absence of sweeping lockdowns. Gates and his underlings went on TV and demanded sweeping shutdowns. He got his way, and it resulted in millions of American businesses destroyed and American families torn apart, for no reason whatsoever. He has shown no remorse and has simply moved on to the next current thing, AI, which apparently demands his infinite wisdom.

The bill came due in tragic fashion. The average American student remains significantly behind pre-COVID achievement levels in math and reading, with the gap widest in the poorest places in the country. Some student metrics fell to their lowest point in decades. The policies that Gates demanded hit hardest on the people who could least afford it.

Not everyone was hurt during the COVID years. Gates, for his part, made a massive windfall on his early investments in mRNA vaccines. Once an mRNA hype man, he later conceded that his “miracle cure” shots weren’t actually useful, and he simply moved on to the next issue.

Now, in 2026, the AI era is here, and Bill Gates is back demanding a centrally planned response to AI innovation. For Gates watchers, it’s easy to notice that the policy structure of his arguments hasn’t changed.

Another new threat that is “unprecedented.” The public is “underestimating” it. Existing institutions are inadequate. New institutions must be built. Humiliating ourselves before China is again necessary. And time is of the essence. We must act right now before it’s too late. We can’t wait for this apparent catastrophe. We must preempt it like the Bush Doctrine in Iraq.

In his essay, Gates says that he wants a sweeping domestic apparatus with authority spanning employment, education, taxation, energy, elections, public health, finance, law enforcement, transportation, and public lands. But that’s still not enough. He also wants to install a new national and global regulatory body. He offers the post-9/11 government reorganization as his benchmark. Who knew the TSA was so popular? Well, Gates apparently wants the equivalent of a Patriot Act and TSA for AI. Hold his vegan, non-GMO, toxic sludge beer, Dario Amodei.

In his infinite central planning wisdom, Gates says it’s time to have jobs reserved by government decree. He calls them “Human Reserved” jobs. To his credit, Gates lists some objections himself but doesn’t really answer his own critiques. He moves forward with the proposal anyway, declaring it “will need to be worked out in public” after it’s implemented. It’s similar to the Pelosi-ism of “pass it so you can find out what’s in it.”

Next, Gates wants to tax AI tokens and robots. He concedes it isn’t economically efficient, but the man with a $100 billion-plus net worth argues “we” can afford the inefficiency because innovation will make up for it. It would “slow the rush away from human labor a little,” he says. The flaw with this argument is that the government would somehow be expected to devote that new tax revenue to something worthwhile. In reality, he is just arguing that America should slow down innovation because he feels bad vibes about human progress right now.

Then there is the China appeal. Gates writes that “some cooperation between the U.S. and China will be required” and holds up Beijing’s AI rules as the world’s most advanced. This is the same guy who argued that America should embrace the ruthless Wuhan lockdowns. Gates has repeatedly praised the Chinese Communist Party’s governance, which he seems to love because of the iron grip the Central Party maintains over the population. We have run “cooperation” experiments with China before, and it has resulted in America being bamboozed by Beijing. Under the Obama-era Paris Agreement, China pledged to curb emissions before 2030 (China never made any concessions), while the United States accepted immediate costs and paid out billions to meet the new standards. The Gates framework will be more of the same.

Gates does concede that AI will bring us great things for humanity, such as cheaper medicine, more accurate diagnostics, productivity for farmers, competent legal and financial help for people who could never afford either, and more. But he says it’s more important to adhere to his nanny state agenda, because the catastrophist has strong feelings about the trajectory it is on. Those positive outcomes he lists are only possible through open markets and free competition, not through a global authority chaired by men like Bill Gates, who were catastrophically wrong the last time they wielded enormous authority and paid no price for it.

The choices we make now are indeed critical. Our first choice should be to stop listening to Bill Gates.

Share

Subscribe now

Share

Subscribe now

Not so long after the piece was published, journalists ran the column through Pangram, an AI-detection tool (which is also now embedded in Substack). The article received a 100% AI generated score.

When reached for comment, Druckenmiller doubled down on his use of AI. [Though I suppose he didn’t double down because he didn’t acknowledge it when the piece was published.] “Of course I used AI,” he said, adding that he’s “not embarrassed by it,” and that he now writes with AI “for the same reason I use a calculator when I do math problems.”

“My name is on the piece. It’s my message,” he declared.

Wall Street Journal opinion editor Paul Gigot, clearly put in an awkward position by editors who grossly failed him (given that the article was so clearly identifiable as AI), defended the piece. In a statement to the media, Gigot said “AI is a fact of modern life” and “nobody can doubt that his op-ed is his genuine opinion.”

The problem isn’t the use of AI. It’s that an S-tier media publication would have such low standards for the use of AI. I have no objection to writers using AI. I think it’s an excellent research assistant and productivity tool. It’s an even better editor for grammar, content, clarity, etc. If Druckenmiller had fed an initial prompt into Claude or GPT to start making his case, and then had his team do a little back and forth to tighten up his argument, I wouldn’t have much of a problem with what he did here.

Druckenmiller basically one-shotted the whole thing, as evidenced by his 100 percent AI score on Pangram and my understanding of what AI writing looks like. This thing wasn’t ever even touched by a human. This wasn’t even an “AI-assisted” piece. A prompt went in and the essay came out. There was no effort or energy dedicated to such a high-stakes op-ed.

Druckenmiller’s calculator analogy is clever but wrong. A calculator does not make an argument for you. It just computes math problems that would remain identical to the product that you would do by hand, just with a faster output through technology.

The op-ed blunder is the exact opposite of the calculator analogy. He basically licensed his name to a large language model and signed off on the result. Druckenmiller outsourced his own thinking to AI, but tried to keep the credit to himself.

I had an interesing back and forth with the chief economics correspondent of the Journal here:

I’m sure Druckenmiller really does believe that Bessent is wrong about bond buybacks and the like. He’s on CNBC all the time delivering similar sentiments, but it’s not an issue I’m super familiar with and the point of this discussion has nothing to do with the merits of the case. Nobody subscribes to the Wall Street Journal to read op-eds that kind of sort of capture the rudimentary sentiments of influential people. They read it because they want to absorb, digest, and distribute the high-level reasoning that makes its way to print.

The Wall Street Journal is making a mistake by legitimizing low-effort slop from high-status people who are trying to shape the public debate. This is the paper of record for American markets. Its editorial page should be a place where ideas are buttoned up and independent of an overreliance on software. Gigot shouldn’t accept a new standard that allows for influential people and organizations to deliver minimal effort, especially when they have incredible resources at their disposal to dedicate time and energy to the process of publishing an op-ed.

AI is a tool that when used well, can very much elevate published work and make it just as authentic as before. Used the way it was used here, it produces nothing more than slop.

Share

Subscribe now

Originally I believe I was supposed to debate Glenn Greenwald on these issues. My opponent later became Harrison Smith, a popular commentator in the right leaning conspiracist/dissident right space. It was moderated by popular Rumble podcaster Viva Frei.

Share

It became apparent early on that Harrison had very few facts to work with so he began by trying to smear me as a foreign agent for Israel and then proceeded to use vague language about all of this being a plot to impose a global surveillance state. It was clear that he didn’t really have any good information other than a couple anecdotal pieces of news. It was akin to debating the horseshoe right version of AOC. It was a good reminder that are not purely about facts, but convincing people that my side of the argument is also reasonable and morally sound.

I thought it was important to debate these issues because so many have become consumed by a hysteria that is entirely detached from reality. And I’m happy to take the unpopular position (which is currently the case on a lot of these topics) as long as I believe it is truthful and the right thing to do.

I’d love to participate in more debates, but take it easy on me because I don’t do it often! Would love to know what you all think, and feel free to drop any recs for who I should debate next. I appreciate your thoughts and value your opinions, whether you agree or disagree with my views.

Here’s a YouTube link to the debate in full:

Share

Surprise surprise. There’s another AI panic sweeping through Washington this month, and it is not holding up to basic scrutiny. Headline after headline has declared that AI models have “escaped” their test environments, gone rogue, and hacked real companies without the consent of their operators. Endless cable news and web-based tech media segments have leaned into the panic narratives, which basically amounts to the idea that the AI software is becoming sentient and betraying its human masters. If it was remotely true, it would make for quite a tale, and it would be sure to create lots of hysteria and demands from lawmakers to regulate AI. However, once you scrutinize the sources of these claims, it boils down to the simple fact that a single partner firm made the same error over and over again.

There were five “incidents” in question over the past month or so.

Anthropic says it reviewed 141,006 interactions in which its Claude models could have reached the open internet during cybersecurity testing, and found three cases where a model supposedly wandered outside its simulated environment and broke out of the sandbox to target real infrastructure belonging to three separate organizations. However, what we really find is that the model did exactly what it was told to do in the exercise. The exercise just was not actually contained in the way it was supposed to be.

Another case from August involved an OpenAI model connecting to the internet during an evaluation and compromising the website of an outside institution. As with Anthropic, it comes down to the same root cause: a misconfigured testing environment that let the model reach the internet, because internet access was available to it.

Meta’s Muse Spark 1.1 model broke into an outside company’s systems during a cybersecurity evaluation. In a recent statement, Meta said it was “a misconfiguration” that “inadvertently allowed one of our models access to the internet during evaluation.”

All five “incidents” were the fault of the same partner, a startup cybersecurity testing firm called Irregular.

For whatever reason, all three companies are using the same contractor to test their frontier models. I don’t have the cybersecurity expertise or knowledge to explain why that is. Perhaps they have a unique technological advantage.

An Irregular spokesperson has already made clear that these non-incident incidents “did not involve a sandbox escape or a sophisticated cyber action.”

So no, AI software did not independently think its way out of supermax confinement, as the media has continued breathlessly reporting.

“A contractor screwed up” and “an AI decided to break free and pursue its own agenda” are two entirely different stories with massive implications for the latter, and only the former is actually supported by evidence. Yet the impression that the latter has occurred has caused all kinds of panic in the halls of power.

Some prominent voices in the cybersecurity world have already expressed skepticism about the new AI panic. Ledger CTO Charles Guillemet described the ongoing news cycle as “marketing theatre,” arguing that having a model “go rogue” has become akin to a test of whether you remain on the frontier. Sorry, Google!

This comes on the heels of the new push from some of these forces to “pace the frontier” and convince lawmakers to grant them unprecedented power to create something of an AI “expert” cartel.

We shouldn’t rule out that self-directed system are possible down the road. Yet a hypothetical situation has been narrative-laundered into a concocted real-life scenario. All these “incidents” prove is that Irregular should probably fix its repeatedly broken testing infrastructure.

Anthropic and OpenAI are two of the fastest-growing companies in American history. They don’t need the White House and Congress to protect them. If Anthropic and OpenAI genuinely believe that they are going too fast and unleashing some mythical monster, they have every resource in the world to tackle it themselves or simply slow their own pace. The fact that they would rather lobby the government to write new rules for them should tell us a lot about what they’re trying to accomplish in D.C.

Subscribe now

The United States is in the early innings of an AI-powered economic boom, but if we embrace the degrowth doomers and stop building, the boom stops with us, and Beijing is more than happy to take our place.

Subscribe now

Share

Here’s the state of play in 2026:

AI-adjacent companies accounted for more than 80 percent of the S&P 500’s gains this year, leaving the index with virtually zero gains without them. Over the course of President Trump’s second term, the trend becomes even clearer. From May 2024 to June 2026, the S&P 500 gained 142 percent.

Now, if we remove the AI stocks from the equation, that number collapses to just 16 percent, which is just about on par with the pace of inflation, or even lower, depending on who you ask. AI-linked companies now make up almost half of the entire index. And that’s not because the market is driven by pure hype and a lack of fundamentals. The AI sector is producing at an unprecedented pace of innovation, and markets are rewarding companies participating in that reality.

Chip companies, data center hosts and builders, energy providers, and software corporations are among the dominant categories turning AI into real products and services. They are the ones putting the American economy and Americans’ retirement accounts on their backs. At the same time, the non-AI-associated economy is struggling, through a combination of inflation, rapidly increasing debt, an insecure housing market, among other factors

America’s AI engine is giving us the potential to elevate human flourishing to new heights. Without it, we could very well be staring down the potential of a long, crippling downturn.

The pace of innovation is unlike anything we’ve ever seen, and it’s worth digging into the numbers to make that clear.

The five biggest American tech companies alone plan to spend over $600 billion on infrastructure by the end of 2026. That is more than the annual GDP of the vast majority of countries, deployed in a single year, by a handful of American companies. While AI doomers and populist degrowthers want to merely chalk it up as some kind of giant scam, the exact opposite is true. This is the result of market demand, and additionally, real, exponential consumer demand. Despite wavering public opinion on AI, take a look at your app store, and you’ll see that the public wants to use these tools, and it finds them greatly useful to their personal and professional endeavors.

According to a McKinsey study, companies using AI are slashing logistics costs by 5 to 20 percent, inventory by 20 to 30 percent, and procurement spending by 5 to 15 percent. Additionally, a June 2026 AI Adoption Study surveyed supply chain leaders across North America and Europe and found that supplementing with AI-driven decisions represents roughly a $55 million opportunity for a $1 billion company. Thanks to AI, it will cost increasingly less to make things and to move things. This matters greatly in an era when families are battling inflation and a mixed bag of policy coming from D.C. Cheaper logistics means cheaper groceries, cheaper shipping, and cheaper everything, really.

Health care and medicine applications in AI are just starting to take off, and they’re already showing great promise. There are now more than 170 AI-discovered drug programs in human clinical trials, with 15 to 20 expected to enter final-stage trials this year alone. Moreover, AI-discovered drugs are currently passing Phase I safety trials at twice the rate of traditionally discovered processes. Health care costs have been a major political and policy issue in the post-Obamacare era, and tragically, these struggles have been making life incredibly difficult for many American families. AI has the technological capacity to reduce the harm of bad economic policy that limits choice and stifles innovation.

On the “solopreneur” front, AI is making it possible for individual Americans to have virtually unlimited productivity and efficiency tools at their disposal.

In May 2026, American “solopreneurs” filed around half a million new business applications. There are now almost 30 million nonemployer businesses in America generating trillions in business. A study conducted by Stripe found that AI is chiefly responsible for the solopreneurboom.

All you need is a laptop and an internet connection, and anyone in America can weaponize AI to do work that once required entire departments. With a little AI assist, you can now do your own marketing, design your own products, and “vibe code” your own website. Millions of Americans are building small businesses with tools that used to cost a fortune and/or were not accessible to the masses.

We don’t appreciate data centers enough. More politicians and policymakers should understand that we need to have robust physical infrastructure in place to ensure the viability of our economic system. Data centers are, without question, the heart of the modern American technological economy in 2026. I call it American technological infrastructure, because these are literally the physical factories of the intelligence age.

If we do not build them, the consequences are immediately apparent. Without AI compute infrastructure, America cannot scale an AI economy. Without an AI economy, the one engine pulling the U.S. market forward will stall out, and it doesn’t take Milton Friedman to figure out that we will almost certainly face the real prospect of a debilitating recession.

China is not waiting for us to figure out how to stay on top here. Beijing has made data centers a national strategic priority, and its capacity is on track to nearly double to 60 gigawatts by 2030. China is electricity-rich, and its data centers pay less than half what American facilities pay for electricity. The AI race will be won by whoever capitalizes on the opportunity to scale compute and generate power for the physical infrastructure of the AI age. If America loses this race, we lose the commanding heights of the global economy to our fiercest competitor, which, of course, happens to be an authoritarian one-party state with polar opposite founding principles to those of the United States.

Lawmakers in D.C. do not need to subsidize this revolution with infinite loopholes and pork legislation and the like. They need to just stop strangling innovation and let markets work.

Our best and brightest business leaders want to build the infrastructure necessary to keep America atop the global hierarchy. We should just let them cook.

Subscribe now

Share

Share

Subscribe now

If you just happened to notice, left out is any acknowledgement of the reality that there are exactly two countries on the AI frontier, the United States and China. They seemingly want the American government to both unilaterally negotiate its own technological restraint and sign an accord with the Chinese Communist Party. Then we will have to trust Beijing to honor these vague, subjective terms. Given that AI doomerism has not proliferated in China in the fashion it has here, this would amount to unilateral technological disarmament.

The “Pacing the Frontier” letter is rife with valuable signatures, and they don’t just come from the hyper wokes within the “trust and safety” teams. Anthropic’s CEO, OpenAI’s chief scientist, Meta’s chief scientist, Google DeepMind’s chief strategy officer, and a long roster of the people who build these systems are signatories to the document.

I can’t help but to point out again that everything hinges on one word in that last sentence: International.

Europe is not in this race or really, any race that involves innovation these days. Mistral, in France, is the one European lab that is kind of sort of doing something interesting, but it’s still rather inconsequential. The continent has built a regulatory apparatus so hostile to building things that its best founders almost always relocate to the United States and incorporate here.

We have seen this appeasing China movie before and we know how it ends.

Take the Paris climate framework as an example. Both the U.S. and China signed the same document, but only the U.S. was baited into action. The United States pledged, in an agreement signed by former President Barack Obama, to drastically cut emissions immediately. This was a hard, near-term number, and the U.S. proceeded to wire billions of dollars into sketchy international climate funds and NGOs. China pledged to let its emissions keep climbing until roughly 2030 and then, at its leisure, to peak. The U.S. bound itself to self destruction while China bought fifteen years of runway and did absolutely nothing to achieve any stated objectives.

How exactly would this committe inspect a frontier AI system’s progress in China? It does not exactly emit gasses that can be picked up by satellite observation. There are no physical inspectors who could verify the compliance of companies within an authoritarian government that presides over a giant land mass and 1.5 billion citizens. A pacing agreement with China would be unverifiable by its nature, which means in practice it slows down exactly one country: ours.

And what exactly are we pacing?

Everyone has a different definition of AI. I think of it as a new technology that encompasses a general-purpose capability and a human productivity tool. If we slow down the frontier and you slow down whatever the frontier was about to produce. Do we tell a man with advanced cancer the bad news that his treatment is now banned because it is violating a subjective “pace” determined by an international committee?

There is no magical dial we can pinpoint that we can command to “slow down only the parts we determine as extra scary.”

Also, who would staff the committee(s)? Presumably, this would mean the signatories themselves, because they are some of the most knowledgeable people in the field, and their contemporaries who disagree with the proposal would not want to join such a committee. They would be handed discretionary power to decide, subjectively, like Soviet commissars, what rate of human progress we are permitted.

Frontier AI companies are raising hundreds of billions of dollars on the promise of the very technological breakthroughs they have achieved in recent months and years. It’s always been a bit paradoxical that these are the same people who are virtue signaling about the need to roll back AI progress. And it seems destined that whatever proposed mechanism gets built will almost certainly determine something that the incumbents can reach and their competitors cannot. Suddenly, the “safeguard” becomes a regulatory moat. Funny how that works out.

The expected rejoinder is that the letter deserves deference because very smart and successful (San Francisco-based liberals) leadership signed it.

But that just proves we have a massive bubble problem in West Coast tech circles. It does not prove that they know the first thing about governing a country, as proven by their historically awful proposal to have an international body usurp the authority of the United States government and our systems of laws and justice.

Being the best in the world at training an AI system confers no expertise in statecraft, treaty enforcement, morality, or constitutional design. And rarely will you be able to acquire the wisdom of the American founding principles by interacting with chatbots. Given their backgrounds, it’s not surprising that they imagine governance as an engineering problem that involves finding the right mechanism, tuning the particular incentives, and voila! Global compliance is achieved. But that’s entirely detached from human nature and the nation state system that governs our politics.

Much of this worldview traces to effective altruism, the movement that has colonized the very loud and intimidating “safety” wings of these companies. Their ideas have very little grounding in the American founding principles that allowed for the flourishing of their respective companies.

What Silicon Valley needs is not a new international body, but an infusion of certain principles that make rational Americans come to the obvious conclusion that self-governing people can handle a powerful technology better than a supranational committee ever can.

Subscribe now

Share