Why AI’s Struggle With Logic Matters

Stunning amounts of money are being poured into AI investment, with five years of aggressive investment occurring in just 12-18 months. However, this surge will largely be the redirection of investment dollars rather than the creation of massive new investment activity. This is because investments will be withdrawn from areas imperiled by artificial intelligence breakthroughs. In this regard, it will be similar to what occurred with the advent of online retail investment, which basically redirected capital spending from stores to online platforms. This resulted in little net change in total investment.

People are atwitter about how not only particular tasks but all human employment will be obliterated by the spread of AI. I’m highly doubtful that this will occur. Since the onset of the Industrial Revolution, claims repeatedly have been made that each breakthrough is the end of employability for most workers. However, for nearly 200 years, new technologies have always resulted in more net labor demand, as the wealth created generates jobs that more than offset specific job losses. Thus, while many jobs will be eliminated, I expect that many more new jobs will appear. So while tractors eliminated many jobs walking behind a plow horse, more importantly it freed the population to pursue far more productive activities. Similarly, industrial robots freed millions of workers from bending metal, all while raising the general quality of life. The impact of AI may be different from historical experience, but we have no reason to believe that that will be the case.

Since the onset of the Industrial Revolution, claims repeatedly have been made that each breakthrough is the end of employability for most workers.

A final problem is that we have not found AI capable of “counter narrative” insights. This is hardly surprising given that it is trained on prevailing narratives. But our longtime readers appreciate the fact that we are not a slave to prevailing narratives. Until AI can predictably overcome these limitations, we will monitor the progress of the technology while relying on genuine intelligence to write The Linneman Letter. That said, we continue to explore the various uses of AI for discrete projects such as data organization and certain analytics.

While great strides have been made in training (ever proliferating) AI models since 2022 when ChatGPT was introduced, accuracy is never a certainty.

Given the proliferation of AI these days, we are frequently asked if we’re using AI to write The Linneman Letter. So far, the answer is “no.” This is for three reasons. First, AI writing is still mediocre, and our readers are not willing to pay for mediocrity. And while AI writes better than most college students (a low bar), it remains pedestrian. A second challenge is the frequency of hallucinations (more precisely, making things up out of whole cloth). While great strides have been made in training (ever proliferating) AI models since 2022, when ChatGPT was introduced, accuracy is never a certainty. Making this problem particularly worrisome is that the hallucinations generally seem reasonable.

Dr. Peter Linneman is a principal & founder of Linneman Associates (www.linnemanassociates.com), Professor Emeritus at the Wharton School of Business, University of Pennsylvania, author of “Real Estate Finance and Investments: Risks and Opportunities,” and co-author of “The Great Age Reboot: Cracking the Longevity Code for a Longer Tomorrow.” Follow Dr. Linneman on X: @P_Linneman

Read the February 2026 issue of CPE.