Alarmingly Shallow Ideas

As one New York financier told Financial Times journalist Gillian Tett, new hires who were seen as “AI natives” are turning out to have alarmingly shallow ideas. So much so, the anonymous finance worker admitted, that his firm now actively avoids seeking out AI-literate STEM graduates, and opts to comb through humanities students instead…

The state of higher education is so bad that many of today’s higher ed students are not only offloading their coursework to AI chatbots like ChatGPT — a shortcut, educators say, that’s even impacting their ability to participate in face-to-face discussions. Bosses Horrified as “AI Native” College Graduates Hit the Workplacehttps://futurism.com/future-society/college-critical-thinking-ai

Yikes.

Deeply disturbing. So what can a concerned parent do?

Ask ChatGPT what to do!

My college graduate son has no critical thinking skills after relying upon ChatGPT. What should I do?

Treat this as a skills-and-habits problem, not a character flaw or a “ChatGPT ruined him” problem. A lot of recent graduates learned to use AI as a shortcut before they learned how to think through ambiguity on their own. The good news is that critical thinking is trainable.

A few things matter here:

  1. Don’t turn this into a fight about technology
  2. Separate competence from confidence
  3. Create situations where he has to reason, decide, and defend conclusions

What often happens is:

  • Students get very good at producing polished answers
  • But weak at:
    • evaluating evidence
    • making tradeoffs
    • spotting bad assumptions
    • persisting through confusion
    • forming independent judgment

Those are different skills.

Here’s what tends to work better than lectures about “kids relying on AI.”

Change how he uses AI

The goal is not “stop using ChatGPT.” It’s “use it as a sparring partner instead of a vending machine.”

Encourage rules like:

  • No copying answers directly
  • He must form an opinion first
  • He must explain why an answer is correct
  • He must ask AI to critique his reasoning instead of replacing it

Good prompts:

  • “Challenge my assumptions.”
  • “What are the strongest counterarguments?”
  • “What evidence would disprove this?”
  • “What am I overlooking?”
  • “Give me three competing interpretations.”

Bad pattern:

  • “Write this for me.”

Put him in environments that punish shallow thinking

Critical thinking develops fastest when reality pushes back.

Good environments:

  • Jobs with responsibility and consequences
  • Sales
  • Operations
  • Managing projects
  • Technical troubleshooting
  • Apprenticeships
  • Debate-heavy workplaces
  • Entrepreneurship
  • Volunteer leadership

Real-world feedback is hard to fake.

Ask questions instead of giving conclusions

If you immediately correct him, he stays mentally passive.

Instead ask:

  • “How did you arrive at that?”
  • “What evidence supports it?”
  • “What would change your mind?”
  • “What’s the downside of your approach?”
  • “What assumption are you making?”

That forces metacognition.

Make him defend positions he disagrees with

One of the fastest ways to build reasoning is:

  • argue both sides
  • compare tradeoffs
  • steelman opposing views

You can do this casually over dinner or current events.

Encourage slow thinking

A lot of AI-heavy habits create intellectual impatience.

Critical thinking often requires:

  • sitting with uncertainty
  • reading longer material
  • comparing sources
  • wrestling with incomplete information

Books, long-form journalism, analytical podcasts, and difficult projects help rebuild that stamina.

Watch for learned helplessness

Sometimes the issue is not inability but reflexive dependence:

  • “I don’t know immediately”
    → “I need AI to think for me”

A useful rule:

  • Spend 15–30 minutes thinking independently before asking AI.

Also be realistic about the transition to adulthood

Many graduates — long before AI existed — left college without strong judgment skills. Those usually develop through:

  • responsibility
  • mistakes
  • work
  • relationships
  • financial pressure
  • repeated decision-making

You do not need to panic because he’s behind in one area at graduation.

The biggest mistake parents make here is trying to “convince” an adult child that he lacks critical thinking. That usually triggers defensiveness and dependency at the same time.

A better posture is:

  • respect him as an adult
  • expect reasoning and accountability
  • ask probing questions
  • stop rescuing poor decisions too quickly
  • encourage environments that require independent judgment

If you want, I can also help you with:

  • signs that this is normal vs genuinely concerning
  • exercises that rebuild critical thinking quickly
  • how to talk to him without sounding accusatory
  • ways young professionals can use AI without weakening reasoning skills

Colorectal Cancer and Mortality Risk Among Older Adults

Conclusions and Relevance  Adults 75 years of age or older with adenoma at prior colonoscopy were more likely to experience subsequent CRC and CRC death compared with those without adenoma, but cumulative risks were low and were far exceeded by competing risks for non-CRC death. Older adults may consider deprioritizing surveillance colonoscopy relative to other health concerns. Colorectal Cancer and Mortality Risk Among Older Adults With vs Without Adenoma on Prior Colonoscopyhttps://jamanetwork.com/journals/jama/article-abstract/2847311

(Less Than) Random Thoughts on Retirement – 05.03.26

Yes dear reader, yet another post in the never ending series of random thoughts on retirement. https://lifeunderwriter.net/tag/random-thoughts-on-retirement/

Though the traditional retirement age in the U.S. typically falls between 62 and 67, many Americans continue working beyond that point. As of 2024, slightly more than 22% of adults aged 65 and older are still employed, either full-time or part-time. Though the traditional retirement age in the U.S. typically falls between 62 and 67, many Americans continue working beyond that point. As of 2024, slightly more than 22% of adults aged 65 and older are still employed, either full-time or part-time.https://financebuzz.com/working-in-retirement-data

I’m not the only Old Guy who is still working past age 65.

But despite the fact I’m not the only Old Guy who is still working past age 65 more people are starting retirements earlier than they expected (always have a Plan B and maybe even a Plan C).

2026 EBRI/Greenwald Retirement Confidence Survey https://www.ebri.org/content/2026-ebri-greenwald-retirement-confidence-survey

I just learned I’m not the only Old Guy who still has a mortgage.

Over the past three decades, the share of homeowners ages 65 to 79 with a mortgage rose from 24% to 41%, while median mortgage debt surged by 400%, according to the Joint Center for Housing Studies at Harvard University. More Americans aging into retirement are still paying down mortgageshttps://www.marketplace.org/story/2025/06/09/more-older-americans-continue-to-pay-mortgages

I also learned about the disappearance of structured cognitive demand in retirement.

A 2025 systematic review published in Health Psychology Review found that retirement is associated with measurable cognitive decline, not just because people age, but because structured cognitive demand disappears. Researchers have called it “mental retirement”: The brain follows the body’s example and withdraws from challenge. Gary Has a Plan for Retirement: Crash on the Sofa and Veg. Here’s the Problem With That …https://www.kiplinger.com/retirement/retirement-planning/your-long-term-retirement-plan-needs-a-purpose

I’m now paranoid about the disappearance of structured cognitive demand with a mortgage to pay off and too scared to retire.

Incubation of Craving (in rats)

Changes in neural connections due to substance use and withdrawal are long-lasting, and craving can peak well into abstinence.

Incubation of craving in rats — in other words, an inverted U-shaped curve where craving rises, plateaus and then declines — holds across drug classes. So although it was initially demonstrated for cocaine, incubation of craving occurs in rats after self-administration of methamphetamine, opioids, nicotine and ethanol. Incubation of cue-induced craving has also been demonstrated in humans — so far, this has been shown during abstinence from cocaine, methamphetamine, nicotine and alcohol. So that long plateau phase that we see in the animals is a relevant model for the persistent vulnerability to craving and relapse in humans who are trying to recover from substance use disorder. What addiction does to the brainhttps://knowablemagazine.org/content/article/mind/2026/what-addiction-does-to-synapses-in-brain

The article is an interview with Marina Wolf, a behavioral neuroscientist at the Oregon Health & Science University. Good stuff if your brain is into brain stuff.

Medscape on Obesity – 04.26.26

Medscape on Obesity  
  Latest Stories  
  Study Exposes Risks in Compounded GLP-1s  
  GLP-1s: Are We Losing Too Much Muscle?  
  Can Weight Loss Cut Cancer Risk? New Data  
  A Looming Crisis in Diabetes  
  Non-Medicinal Satiety Tips for Weight Loss  
  Trending  
  New Obesity Medication Guidance: Will It Reduce Stigma?  
  Why GLP-1 Results Vary: New Genetic Data  
  Is Tallow Shallow? The Skinny on Unsaturated Fats  

“Medical” Advice for the Masses

The AIs’ failure rates exceeded 80 percent when provided with given ambiguous symptoms that could match more than one condition, and for more straightforward cases that included including physical exam findings and lab results, they still failed 40 percent of the time. The researchers also found that unlike human clinicians, the “LLMs collapse prematurely onto single answers,” resulting in “weak performance” across all models. Millions of Americans Are Talking to AI Instead of Going to the Doctor, and It’s Giving Them Horrendously Flawed Medical Advicehttps://futurism.com/artificial-intelligence/millions-americans-ai-instead-doctor-bad-advice

Wow.

From the study discussion section:

Our evaluation suggests that despite rapid advances in pattern recognition and knowledge retrieval, current LLMs still lack the reasoning processes needed for safe clinical use. The consistent gap between differential diagnosis and final diagnosis highlights how differently these systems process information compared with physicians. Clinicians preserve uncertainty and iteratively refine differential diagnoses, whereas LLMs collapse prematurely onto single answers, a limitation that persists across model generations. Their weak performance on differential diagnosis, consistent with a prior study from authors of the current work,8 suggests these limitations persist across early and state-of-the-art models. The risk is not just that LLMs are sometimes wrong but that their reasoning is brittle precisely where uncertainty and nuance matter most. Benchmarks that reward only correct final answers risk reinforcing this shortcutting, widening the gap between marketing claims and the skills actually required at the bedside. Large Language Model Performance and Clinical Reasoning Tasks – Rao AS, Esmail KP, Lee RS, et al. Large Language Model Performance and Clinical Reasoning Tasks. JAMA Netw Open. 2026;9(4):e264003. doi:10.1001/jamanetworkopen.2026.4003 https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2847679

Wow.

Should you really trust health advice from an AI chatbot? https://www.bbc.com/news/articles/clyepyy82kxo. Dr Nicholas Tiller explains: “They are designed to give very confident, very authoritative responses, and that conveys a sense of credibility, so the user assumes that it must know what it’s talking about.” He thinks chatbots should be avoided for health advice unless you have the expertise to know when the AI is getting the answers wrong.

The study’s Conclusions The audited chatbots performed poorly when answering questions in misinformation-prone health and medical fields. Continued deployment without public education and oversight risks amplifying misinformation. Tiller NB, Marcon AR, Zenone M, et al

Generative artificial intelligence-driven chatbots and medical misinformation: an accuracy, referencing and readability audit BMJ Open 2026;16:e112695. doi: 10.1136/bmjopen-2025-112695 https://bmjopen.bmj.com/content/16/4/e112695

Wow.

Now go read this thread posted on LinkedIn https://www.linkedin.com/posts/gratuz_ai-llm-activity-7358862577512165376-Q7AA

Yikes.

Scary Charts 04.17.26 (random thoughts on retirement and longevity risk)

There is a significant chance that you will live for many years beyond the average, and you should consider this possibility when thinking about your retirement. The Actuaries Longevity Illustrator (“ALI”) has been developed as an educational tool by the American Academy of Actuaries and the Society of Actuaries to help you gauge what those chances are. Reflecting on your longevity will allow you to consider the risks of outliving your financial resources, i.e., the chance of running out of money during your lifetime, which we refer to as retiree financial longevity risk . The ALI helps you to consider those risks by letting you see how long you might live. https://www.longevityillustrator.org/

Try this calculator to see your probability of living to certain ripe old age.

America’s Largest Hospital System Ready to Start Replacing Radiologists With AI

“We could replace a great deal of radiologists with AI at this moment, if we are ready to do the regulatory challenge,”
Mitchell Katz, president and CEO of New York’s 11-hospital public benefit corporation

Mohammed Suhail, a radiologist at North Coast Imaging in San Diego, told Radiology that Katz’s comments are “undeniable proof that confidently uninformed hospital administrators are a danger to patients (and are) ..“easily duped by AI companies that are nowhere near capable of providing patient care.”

America’s Largest Hospital System Ready to Start Replacing Radiologists With AI, Its CEO Says – https://futurism.com/artificial-intelligence/hospital-ceo-ai-radiology

Confidently. Uninformed.

Yikes.