Podcast Summary

Podcast: Something You Should Know
Host: Mike Carruthers
Episode: The Serious Problems with AI & Why Humans Drink Alcohol
Date: March 19, 2026

Overview

This episode dives into two insightful topics:

1. The growing concerns and misunderstandings surrounding artificial intelligence, particularly large language models, with guest Emily Bender, Professor of Linguistics at the University of Washington.
2. The societal, psychological, and biological roots of why humans drink alcohol, with Dr. Charles Knowles, Professor of Surgery at Queen Mary University, London.

Both guests challenge prevailing narratives: Bender critiques the “AI hype,” questioning both the technological progress and the societal impacts of current AI trends, while Knowles unpacks cultural myths and scientific facts about alcohol, drawing on both research and personal experience.

Part 1: The Problems with AI Hype

Guest: Emily Bender, Professor of Linguistics, UW; Author of “The AI: How to Fight Big Tech's Hype and Create the Future We Want”

[06:02] What Is “AI Hype”?

• Definition: Hype is essentially overblown salesmanship, aiming to convince people they must get on board with technology for fear of missing out, often without substance.

  “Hype is this, like, salesmanship, the sense that if you don't get on board, you're going to be missing out, you've got to jump on the train with everybody else without necessarily there being a real there there.” (Emily Bender, 06:17)

[07:04] What AI Actually Is – Not a Unified Technology

• AI ≠ General Intelligence: “Artificial intelligence” covers a variety of technologies, but current hype centers mostly on chatbots, which are actually large-scale language simulators (not true cognition or understanding).

  “The main thing that everyone is super excited about right now, chatbots, or you might call them conversation simulators…systems designed to mimic the way we use language...what we have is a system that mimics the way we use language.” (Bender, 07:04)

• Many users mistakenly attribute deeper understanding or information-synthesizing abilities to these systems, which is an illusion created by their user interface and outputs.

[08:08] Retrieval-Augmented Generation (RAG) and Perceived Value

• Chatbots sometimes “retrieve” web search results and summarize them, but with no guarantees of accuracy or transparency over sources.

  “There’s no accountability there. There’s no reliability...it's not searching the whole Internet any more than an old-fashioned search engine does, but...” (Bender, 08:08)

[09:19] Difference Between Classic Search and AI Chatbots

• Key difference is user experience: chatbots allow natural language queries, which fosters an illusion of understanding.
• The summarization and conversational style can mislead users into trusting information that has neither clear origins nor responsible authorship.

[10:24] Why This Is Dangerous

• The lack of transparent sourcing and accountability can be harmful, especially in sensitive domains (e.g., medical advice).
• Example with medical query:

  “If instead, I put this into a chatbot and I get back something that might have come from one of any of those sources, I don’t know where it came from, so I don’t know how to situate it. I am cut off from my ability to continually build up my understanding of these different sources...” (Bender, 10:24)

[11:52] On Asking Chatbots For Sources

• “There’s no guarantee that the summary is an accurate representation of those links...because there’s no accountability there. It has no understanding of what it is processing.” (Bender, 11:52)

[13:03] Lack of Consistency and Accountability

• Chatbots’ answers can vary day to day; they cannot explain or defend their responses in a human sense.

[13:52] Why Emily Avoids Synthetic Media

• Synthetic media: Any text, image, or video generated by AI.
• Prefers “artistic effort of a person...or the thought processes and accountability of someone writing text.” (Bender, 13:59)

[16:09] Personal Use Case: ChatGPT as Writing Assistant

• Host shares he sometimes uses ChatGPT to help polish podcast segment intros.

• Bender argues this risks losing personal skill, originality, and voice, and perpetuates negative impacts (environmental, economic, and labor).

  “Every time we do the cognitive offloading...we lose that opportunity to practice. We also lose the opportunity to really hone our own unique voices and...end up towards this average, which I think is really quite bland.” (Bender, 17:02)

• Bender objects to calling AI a “collaborator”: it’s not a person; it’s just math processing words, most of which were “stolen” from the internet without consent (see [19:20]).

[20:10] What’s Driving the AI Boom?

• Not technological breakthroughs, but enormous amounts of capital invested by large tech companies attempting to create new monopolies.

  “What is powering this wave is not major breakthroughs in technology. What is powering the wave is enormous amounts of capital...They have to push it at us every single minute...because they have invested so much money in it.” (Bender, 20:10)

[21:24] Is This Inevitable?

• Bender rejects “inevitability” arguments. She calls for public accountability, both for the technology’s outputs and for maintaining human expertise.

  “...anything that encourages us to turn towards machines and towards these products rather than towards each other and towards...community and connection is anti-progress, really.” (Bender, 20:55)

[21:34] Bender’s Wish List for the Future

1. Companies must take full responsibility for all synthetic media outputs (libel, medical errors, etc.).
2. People should take pride in and maintain their own expertise rather than assuming the computer is superior.

[23:22] The Calculator Comparison – Why It Falls Short

• Calculators are reliable, focused tools; chatbots are inconsistent, often unreliable, and not subject to a clear or auditable performance standard.

[24:57] Is ChatGPT Worse Than "Dr. Google"?

• Chatbots tend to give direct, plausible-sounding but untraceable and often arbitrary answers, further hindering evaluation skills and increasing risk of misinformation.

[26:18] Conclusion of AI Segment

• Bender offers a sobering counter-narrative to the optimism about AI, emphasizing critical thinking, source evaluation, and the importance of human connection.

Part 2: Why Humans Drink Alcohol

Guest: Dr. Charles Knowles, Professor of Surgery; Author of “Why We Drink Too Much: The Impact of Alcohol on Our Bodies and Culture”

[30:36] Is Alcohol Pure Poison?

• Knowles is not anti-alcohol; he emphasizes its deep historical and cultural roots.
• “If [alcohol] were just a poison, we wouldn’t do it.” (Knowles, 31:02)
• Serious health consequences are real (including its status as a Class 1 carcinogen), but for most people, long-term, high-level consumption is the primary danger.

[32:56] Alcohol and Cancer Risk

• Even small, regular amounts can increase breast cancer risk in women and risk of stroke.
• For other health outcomes, risk rises usually at higher consumption levels.

[33:59] “Safe” Limits Are Fuzzy

• Definitions and recommended limits vary by country (and sometimes by gender). There’s wide variability in the relationship between consumption and harm.

  “It’s not a straightforward relationship at all...as an individual, it’s only a question of probabilities.” (Knowles, 34:53)

[35:45] Any Good News About Alcohol?

• Some evidence suggests a small daily intake (especially moderate wine) may reduce risk of heart attack in some white European populations, but not other benefits.
• Nevertheless, WHO and medical consensus is that “there is no safe level of alcohol consumption.”

[36:38] Does Alcohol Kill Brain Cells?

• “True...ish.” Chronic heavy drinking is related to brain changes and some cell loss, but for most moderate drinkers, clinical significance is uncertain.

[38:05] Spirits vs. Wine or Beer

• The form of alcohol usually doesn’t matter; it’s about total intake (and rate of consumption).

[39:43] Dr. Knowles’ Personal Story of Dependence

• Ten years sober after struggling with dependence for years.
• Describes the transition from controlled to problematic drinking, highlighting that some people have a much harder time stopping than others.

[42:17] Why Do Some Develop Alcohol Dependence?

• About 2–3% of the population develop “alcohol dependence” (aka alcoholism).
• Causes:
  • 50% heritable (genetics)
  • Environmental and psychological factors: e.g., childhood environment, psychological traits (ADHD, depression, neuroticism, PTSD, etc.)

[43:34] Familial and Early Exposure Factors

• Drinking at an early age (before 14) increases risk.
• Being raised in a drinking household does not, by itself, strongly predict adult drinking, once age of first use is accounted for.

[44:00] Alcohol’s Social Function

• Alcohol production coincides with human civilization—likely serving to ease social interaction and group formation.

  “One can speculate that's no coincidence...back in those days...sitting around the campfire, having to socially interact with other people, one wonders whether alcohol's fulfilled a purpose really since that early time.” (Knowles, 44:17)

• For many, alcohol is a “social fitness” aid, helping to reduce anxiety and increase connection.

[45:57] First Experiences and Individual Risk

• Early affinity for the effects of alcohol can be a warning sign of future problems.
• For Knowles, even as a child, alcohol provided immediate relief from social anxiety and other underlying issues.

[47:22] Gray Area Drinking

• Not a medical term, but describes hazardous patterns short of dependence—about 20% of the population fall into this gray zone.
• These individuals may not suffer major consequences (yet), but struggle to moderate intake and worry about future harm.

[49:13] Why He Quit

• Ultimately, crisis and profound personal suffering led Knowles to stop drinking—“sick and tired of being sick and tired.”

Notable Quotes

On AI Hype:

“What is powering this wave is...enormous amounts of capital. And these companies aren't making money. They are basically telling us we have to use it because they have invested so much money in it.” (Emily Bender, 20:10)

On Synthetic Media:

“I don't want to spend time on things that don't reflect the...artistic effort of a person drawing an image or the thought processes and accountability of someone writing text.” (Emily Bender, 13:59)

On Alcohol's Enduring Power:

“If it were just a poison, we wouldn't do it. I think what has changed is an appreciation of the health consequences of alcohol.” (Dr. Knowles, 31:02)

On Gray Area Drinking:

“The situation has reached one of...hazardous alcohol use rather than harmful alcohol use. So it's a person who worries that those harms may be coming down the line...but the yets are yet to happen.” (Dr. Knowles, 47:53)

Key Timestamps

• [06:02] – Defining AI Hype and Public Misunderstanding
• [07:04] – Chatbots and the Illusion of Intelligence
• [09:19] – Old vs. New Search (Illusion of Understanding)
• [10:24] – Dangers of Sourcing & Accountability
• [16:09] – AI as Writing Tool: Skill Erosion & Ethics
• [20:10] – What’s Driving AI: Money, Not Breakthroughs
• [21:34] – Bender’s Wish List for Tech’s Future
• [23:22] – Calculators vs. Chatbots: Reliability and Trust
• [26:18] – AI Segment Wrap; Bender’s Challenge
• [30:36] – Myths and Realities of Alcohol as “Poison”
• [32:56] – Alcohol and Cancer Risk (Especially for Women)
• [34:08] – Varying Global “Safe” Limits
• [35:45] – Benefits? Very Limited and Contextual
• [36:38] – Brain Health and Alcohol
• [39:43] – Personal Story: Knowles’ Dependence
• [42:17] – Causes: Genetics, Mental Health, Environment
• [44:00] – Alcohol as Social Lubricant
• [47:22] – Gray Area Drinking Explained
• [49:13] – Rock Bottom and Quitting

Tone, Flow & Engagement

The tone is candid, practical, sometimes critical and always grounded in real-world implications and personal experience. Both guests use stories and evidence to challenge assumptions—Bender with sharp skepticism about “AI progress” and Knowles with a nuanced, personal understanding of why alcohol has both tempted and harmed humans for millennia.

Summary Takeaways

• The so-called “AI revolution” is more a product of investment and hype than true technological progress. The risks of misunderstanding, uncritical use, and lost human skills are serious.
• Synthetic media lack origin, accountability, and reliability, and their proliferation is driven by business interests, not social need.
• Alcohol consumption presents a blend of risk and reward, rooted deeply in our biology and cultures. Dependence is partially genetic and psychological, while most people remain in a gray area where risk can be managed, but never erased.
• Both segments urge listeners to think critically: about their sources of information, the seductions of technology, and their own habits and health.

For More:

• [Emily Bender’s Book: "The AI: How to Fight Big Tech's Hype and Create the Future We Want"](link in show notes)
• [Dr. Charles Knowles’ Book: "Why We Drink Too Much"](link in show notes)