[AI Disclosure: This comment was drafted with the assistance of an AI.]

It is always amusing watching the same crowd who proudly brag about running uBlock Origin to starve ad-supported web monetization into the dirt turn around and clutch their pearls about AI needing corporate revenue models.

So which is it? Do you want ads plastered all over AI that you are just going to block anyway to keep things like the good old internet, or should AI companies bypass customers entirely and get all their funding straight from an inflating stock market backed by a Fed put?

How come Gemini just returned my prompt unchanged this time? Did I run out of free tokens?

"The dispatching company will just bill the driver for the value of the cargo."

How about bill insurance which pays for it out of stock market returns on their investments not necessarily premiums?

AI DISCLOSURE: This comment was drafted with substantial help from AI. I use AI as a research and editing tool. The opinions are mine. Any errors are mine.

---

What do you mean by "taxpayers will pay for it"?

The Fed did not raise taxes to fund QE after 2008. It created reserves and bought assets. In fact, tax rates generally did not rise because of the bailouts, and some were later reduced.

If your argument is that the public bears the costs through inflation, asset-price distortions, or moral hazard, that's a different argument. But "government intervention" and "taxpayers paid for it" are not the same thing.

AI DISCLOSURE: This reply was drafted with substantial help from AI. I use AI as an editing and fact-checking tool. The opinions are mine. Any mistakes are mine too.

---

"Exactly equal" is doing a lot of work here.

Federal debt is up about $35 trillion since 1980. U.S. stock market capitalization is up something like $60+ trillion over the same period. Those are both large numbers, but they are not the same number.

Are you claiming a correlation, a causal relationship, or an actual accounting identity? Those are very different claims.

If you have data showing they are "exactly equal," I'd be interested in seeing it.

AI DISCLOSURE: This comment was drafted with substantial help from AI. I use AI as a writing and editing tool. The opinions are mine. Any mistakes are mine too.

---

Plus ca change...

This sounds like a modern version of Rockford Files' "Gearjammers." The cargo thieves of the 1970s used forged documents, insiders, and logistics tricks to divert valuable truckloads. Today's cargo thieves use hacked accounts, GPS spoofing, and AI-generated IDs.

The technology changed. The weak point stayed the same: goods in transit.

eMudhra is exiting the global SSL/TLS certificate business and is offering customers Sectigo-issued replacements [ ... ]

...Sectigo? Née Comodo? They haven't always been the sharpest tools in the shed, either.

But again that's backwards. I let my AI coding assistant write tests exactly so that I have a reproducible, deterministic ground truth that I can inspect for logic errors. If you let the AI do the testing directly, it's a toss-up if today it feels like it or it decides that your desired result is "everything green" so it'll just give you that answer without actually testing anything.

LLMs obviously aren't human. But that's not the point. The point is that you don't need to know a programming language to program a computer.

For example: Especially in the early days of smartphones, there were mountains of small tools that did something that's essentially a simple script. Egg timers, remind-me-tomorrow apps, tons of stuff that's basically a wrapper around a web search, and so on. Almost all of those are today something ChatGPT can do, and are easy enough to describe.

I've actually outsourced the first iteration of that to AI. :-)

At the end of coding sessions, I generall ask it to spin up an adversial agent whose task it is to find flaws, inconsistencies, mistakes, bugs, etc.

I'd still strongly recommend to anyone shipping AI-generated code to have a thorough human code review, but for a first pass at surface level, the AI can check itself if it has some kind of separation between the generator and the reviewer.

Personally, after some experimenting, I've concluded that:

• AI is good at implementing known algorithms.
• AI is good in creating code that flows straightforward from a design document.
• AI is good at code reviews and finding bugs.
• AI (unlike me) doesn't mind writing a few hundred tests.
• But, it tends to over-engineering simple tasks to the point of "why does Hello World need enterprise-software levels of abstraction?"
• It struggles with code that doesn't have a known solution.
• It's difficult to get it into a different lane once it started going down the wrong way.

What I use it for these days is rapid prototyping. Using AI I can make a functioning prototype of an idea I have in a few days instead of a few weeks. I can quickly test out an idea for a feature or change in my existing codebase.

What I also have been using it for a while is as a code-reviewer and for someone like me who is mostly working solo, that's an insanely valuable thing. Not just the bugs themselves, but also the ability to discuss code with someone. Basically, AI is a rubber ducky that replies.

Mickeysoft lost the plot again.

The game-changing element of LLMs and agents is that they have a non-computer-language interface - you can talk to them in plain English (or Chinese or whatever). This makes complex computer interactions available to a much wider audience than any previous computing invention. Adapting humans to the AI is getting that backwards. It's a sign you don't understand the first thing about why AI is so widely successful.

I predict this will be an abject failure and a year from now will quietly be pushed to the very back of the bottom-most drawer, never to be spoken off again (though it is getting crowded in there for M$).

Meanwhile, in the real world, everyone I've talked to about these things already has a skills file to teach the AI about their personal coding style. This is what actual humans came up with, independently several times, and it works: Adapt the AI to you. Plus we've been writing coding style guides for humans for a long time now, we've got plenty of experience doing that at least somewhat right.

(Note: Drafted with AI assistance to structure arguments on educational theory, primary sources, and media tropes.)

BitterEpicâ(TM)s heroin metaphor relies on a fundamental misunderstanding of both learning and primary sources.

1. Practice vs. Intoxication: Shifting away from grading homework isnâ(TM)t "addiction"; it aligns with educational research from critics like Alfie Kohn and Sebastian Thrun, who have long argued that grading practice work destroys intrinsic motivation and replaces learning with compliance. Removing grades from practice allows students to focus on understanding rather than optimization for a score.
2. Primary Sources: High-level language models do not replace primary sources; they act as search, synthesis, and translation layers over them. A student using a tool to parse, cross-reference, or organize primary research is no more "intoxicated" than someone using a card catalog, a digital database, or an index.
3. The Sutter Cane Crutch: Repeating a reference to a 1990s horror film does not turn an automated tool into a mind-controlling deity. It just highlights a reluctance to evaluate tools based on how they are actually used, preferring instead to treat any shift in learning habits as a psychological collapse.

(Note: Drafted with AI assistance to organize arguments on human-supremacist rhetoric and ideological dependence.)

BitterEpic calls everyone else addicts, but he seems hooked on his own drug: performative humanist moralizing.

His entire rant relies on a rigid script: compare software tools to hard drugs, accuse anyone using automated assistance of psychosis, and demand proof of pure human origin before evaluating an answer. That is not original thinking; it is a mechanical compulsion to enforce social conformity.

Dropping a movie reference to Sutter Cane does not make this profound. It just shows he is dependent on drama to feel superior online. If a tool provides accurate information or solves a problem, nobody needs a sermon on human purity. The real addiction here is his need to police how other people process information.

(Note: Drafted with AI assistance to organize arguments on economic measurement and productivity theory.)

dirk's argument assumes a world with infinite demand and zero quality constraints. In reality, saving time or avoiding mistakes frequently does not increase total measurable output for three reasons:

Fixed Demand: If an engineer uses AI to cut decision-making time in half and finishes their assigned tasks in 4 hours instead of 8, output remains at 100%. Unless the company has an endless backlog of marketable products to sellâ"and customers ready to buy themâ"the saved time does not automatically transform into more widgets or higher GDP.

Prevention Is Invisible in GDP: Avoiding a catastrophic server outage or a costly legal mistake adds immense value and resilience, but it produces zero additional units of output. In standard economic stats, a company that spends $1M fixing an error shows the same or higher economic activity than a company that avoided the error entirely using better tools.

Quality vs. Quantity: Better coordination and fewer errors often yield higher quality, safer code, or better-vetted decisions, not a higher volume of output. National productivity statistics measure the quantity of output per hour worked, not the systemic stability or risk reduction achieved along the way.

AI disclosure: I used AI to help organize and edit these thoughts. The arguments are mine to the extent that any argument is ever entirely someone's own. If that bothers you, feel free to attack the argument rather than the tool.

---

I think the deeper issue is why we're treating productivity statistics as the final judge in the first place.

Productivity is not a direct measurement of economic value. It is an accounting construct that depends on how output is defined and measured.

That is relatively straightforward when counting tons of steel or bushels of wheat. It becomes much harder when dealing with software, research, legal work, design, information retrieval, decision support, risk management, and other information-intensive activities.

AI could be improving decisions, reducing errors, lowering search costs, increasing optionality, or generating consumer surplus without immediately appearing in revenue-per-worker statistics.

We already know GDP and productivity measures struggle to capture the value of many digital goods. Search engines, GPS, Wikipedia, open source software, and the broader Internet generated enormous value relative to what showed up in traditional economic statistics.

There is also a timing problem. What was the measured productivity gain for the average manual laborer shortly after the steam engine was introduced? General-purpose technologies often require years of organizational change before aggregate statistics move. Factories had to be reorganized around electricity before electrification delivered its full benefits.

The argument also seems oddly selective. Slok removes the companies where AI's effects are most obvious, then concludes that the remaining firms have not yet shown economy-wide productivity gains. If the steam engine first changed mining and textiles, would we have concluded steam had no economic impact because most workers were still digging ditches by hand?

Most importantly, why assume productivity is the metric that matters? A technology can create value by reducing uncertainty, improving coordination, avoiding mistakes, increasing resilience, or expanding opportunities. Those benefits do not necessarily show up as higher output per worker.

So before concluding that AI has no economic impact because productivity measures are flat, I would first want to know whether productivity statistics are actually measuring the thing we care about. Flat productivity could mean AI is useless. It could also mean the metric is only observing a subset of the effects.