Gary N. Smith

_{Senior Fellow, Walter Bradley Center for Natural and Artificial Intelligence}

Gary N. Smith is the emeritus Fletcher Jones Professor of Economics at Pomona College. He is the author of more than 100 research papers and 20 books, including The AI Delusion, (Oxford University Press, 2018; translated into Korean, Vietnamese, Chinese for sale in Taiwan, and simplified Chinese for sale in mainland China), The 9 Pitfalls of Data Science, with Jay Cordes (Oxford University Press, 2019; winner of the 2020 PROSE Award for Popular Science and Popular Mathematics), The Phantom Pattern Problem: The Mirage of Big Data, with Jay Cordes (Oxford University Press, 2020), Distrust: Big Data, Data-Torturing, and the Assault on Science (Oxford University Press, 2023), and Standard Deviations: The truth about flawed statistics, AI and big data (Duckworth, 2024)

Archives

What Can We Learn From the Chinese Housing Bubble?

_{Anyone can use our theory, which has been confirmed by events in foreign real estate markets, when determining whether to buy a home}_{Gary Smith
September 2, 2026}

Bubbles are created by speculators who don’t care about an asset’s income because they expect the price to keep rising. For most buyers, income is what counts.

Might Now Be a Good Time to Buy a House?

_{I asked ChatGPT, Gemini, and Claude and they confidently spewed bucketloads of well-written but barely relevant prose. There is a better way to find out}_{Gary Smith
August 28, 2026}

All real estate is local but, despite bubbles, it is generally better to buy than rent.

UFOs and Crop Circles: Believe It — or Don’t

_{Alleged evidence of extraterrestrial visits often evolves from local news headlines into a game, an art, and an industry}_{Gary Smith
August 26, 2026}

Even when the pranksters who create the often elaborate crop circles are identified, "croppies" insist that only ET could have done it.

Colleges and Universities: Still on a Road to Nowhere

_{One cost-saving proposal has been to eliminate courses with ten or fewer students, irrespective of what the course is teaching}_{Gary Smith
July 31, 2026}

Significantly, no administrators who were contacted recommended reducing schools’ budget woes by reducing the number of administrators.

Parkinson’s Law Is the Key to Understanding Bureaucratic Bloat

_{The experience of one small university offers a useful example}_{Gary Smith
July 20, 2026}

I spent 55 years as an eyewitness to this disaster, watching a relatively lean and efficient enterprise turn into a distended, barely recognizable caricature.

Addicted to Social Media: Are We the Customer or the Product?

_{A lawsuit has revealed social media company strategies that seem to promote addiction}_{Gary Smith
July 13, 2026}

Meta and Google recently lost a bellwether case accusing them of exacerbating children’s mental health struggles by intentionally addicting them to unhealthy social media.

Plagiarism and Defamation—Two More Bad Things LLMs Are Good At

_{The authors’ and publishers’ copyright claims against the chatbot developers are only one of the legal problems the industry faces}_{Gary Smith
July 1, 2026}

The LLMs’ hallucinations could end up getting them severely restricted in Europe, depending on the outcome of a recent legal challenge for defamation.

Speculation in Large Language Models: The Worst Sort of Business

_{Speculators buy an asset because they expect the price to keep rising, not because they want the income generated by the asset}_{Gary Smith
June 15, 2026}

The inventions that changed our lives for the better were usually accompanied by speculation bubbles and busts. LLM firms will likely attract the same fate.

Are AI Developers As Smart As They Are Made Out To Be?

_{My confidence in the intelligence of AI developers was shaken by two recent events}_{Gary Smith
June 1, 2026}

Many commentators seem to have forgotten Goodhart’s Law: When a measure becomes a target, it ceases to be a good measure.

AI-Lawyers Will Have Fools for Clients

_{When tech prophets promise everything—except accuracy.}_{Gary Smith
February 17, 2026}

Lazy lawyers have no doubt filed hallucinated content orders of magnitude more often.

Computer Intelligence Versus Human Intelligence

_{Professor Joseph Weizenbaum created a chatbot he named ELIZA that conversed with users the way a psychotherapist might}_{Gary Smith
January 13, 2026}

Even though the users knew they were interacting with a computer, many were convinced that the program had human-like intelligence and emotions and they happily shared their deepest feelings and most closely held secrets.

Large Language Models (LLMs) Are Inherently Frail and Unreliable

_{From my recent tests, things have not improved much. Even an apparent improvement did not last long. See for yourself}_{Gary Smith
January 5, 2026}

Post-training might stabilize the responses to specific inquiries but then the bot can just go off the rails again, as my tests showed.

The Core Problem with Large Language Models

_{LLMs are inherently unreliable, which means that failures are not incidental, easily fixable glitches}_{Gary Smith
January 1, 2026}

The data deluge exponentially increases the number of coincidental, useless statistical patterns — so the probability of useful patterns approaches zero.

Is OpenAI Approaching the Valley of Death?

_{Overpromising is still a big problem; in any event, the fate of Netscape looms}_{Gary Smith
December 22, 2025}

OpenAI needs to show that ChatGPT is more than just the first publicly available LLM. It has not done that and maybe never will.

Home Ownership: Madness Over 30-Year Mortgages

_{It is tempting to put borrowing and investment decisions in separate mental buckets, but they are intimately related}_{Gary Smith
November 21, 2025}

When considering a mortgage, the correct comparison is not total payments but the return on the borrowed money versus the loan’s annual percentage rate.

Illusions: No, Large Language Models Do Not Understand

_{A recent New Yorker article is mistaken about this. For one thing, the LLMs have trouble distinguishing between causation and mere correlation}_{Gary Smith
November 14, 2025}

A real-world example of such struggles is the poor performance, in general, of AI-powered mutual funds.

Scaling Up is Not Going to Make Large Language Models Intelligent

_{ChatGPT-5 misunderstood the jab in a simple Will Rogers joke five out of six times}_{Gary Smith
October 30, 2025}

Bigger data centers are not going to change the harsh reality that LLMs are prone to generating unreliable blather.

Large Language Models: Inconsistent Because They’re Unintelligent

_{Here’s what happened when I tested popular LLMs on student exercise questions I have been using for over fifty years}_{Gary Smith
October 28, 2025}

Claims that an LLM is intelligent don’t square with the fact that it may answer a question one way and, a few seconds later, simply contradict itself.

For a Sounder Approach to Investment: Deep-Six the 60/40 Rule

_{Sometimes, good decisions require more than blindly accepting advice from humans as well as computers!}_{Gary Smith
October 20, 2025}

The conventional belief that a sound investment portfolio should be 60% stocks and 40% bonds does not stand up to careful scrutiny.

Intelligence Demands More than Pressing a Lever to Obtain Water

_{I continue to be astonished by how willing people are to assume that LLMs are intelligent because they give glib, confident answers}_{Gary Smith
October 7, 2025}

Human trainers can nudge LLMs to give good answers for specific questions but this training doesn’t prepare them for subtle variations.