AI Frontier

Clara Nguyen

0.0 (0)
Technology
Weekly

Researchers, engineers, and operators on the practical state of AI.

5d ago

AI Made Simple: From GenAI to LLMs

AI Made Simple: Verifying AI Output and Starting Small with Agents 1. Episode summary John Binks, author of AI Made Simple, explains the difference between AI and generative AI, then shows how he fact-checks model output by having Gemini and ChatGPT argue with each other. He also describes the agent he uses to print a two-page AI news "newspaper" every morning, and why he'd never hand nuclear or surgical decisions to a machine. 2. What you'll learn AI is the broad category (a self-driving car reading traffic), while generative AI creates something new from patterns. Not every gen AI tool is an LLM; diffusion models, for example, generate graphics. Read every output before you paste it anywhere. Binks sees people copy, paste, and hit send without checking that it makes sense. Ask the model for links to its source material so you can check claims yourself and catch hallucinations. A broken link is a red flag. Use "adversarial AI": put two models, such as Gemini and ChatGPT, in a debate. The disagreements show you where to verify, ideally on a separate system or with a specialist. Match the verification effort to the stakes. A homeowners association chili cook-off flyer needs little checking, while nuclear or surgical work needs a human in the loop. 3. Notable quotes "Read through it. See does it make sense to you." — John Binks "You don't want an AI that's launching nuclear bombs, right? You want to have a human in the loop." — John Binks 4. About the guest John Binks is the author of AI Made Simple: A Plain-English Guide to Understanding AI, GenAI, Machine Learning, and LLMs, the first book in his "Made Simple" series. He writes for people just starting out in AI and has published other books, which are available on Amazon. He advises leaders to begin with small, repetitive tasks, and he runs his own agents, including one that prints a morning AI news digest and another that summarizes his email. He can be found at botsandbosses.com. 5. Topics covered AI vs. Generative AI Diffusion Models Spotting Hallucinations Adversarial AI Checking Model Wrappers and GUIs Starting Small with Agents Human-in-the-Loop Matching Verification to Stakes
Sep 17

Bo Bennett On AI's Greatest Threats And Possibilities

1. Episode summary Bo Bennett — author, social psychologist, and operator behind AI writing tools like AuthorVoices and MemoirMaker alongside human-focused eBookIt.com — joins Clara Nguyen to argue that AI "creativity" is no different from human creativity, since both are just recombination of prior influences. He also lays out a contrarian theory that AI leaders' existential-risk warnings are less about safety and more about managing an unsustainable business model built on constant reinvestment. 2. What you'll learn Why Bo says there's "no solid answer" for when a book stops being human-written and becomes AI-written — it's a judgment call, not a bright line. His case that human creativity and AI training work the same way: both take existing material and "rearrange it in a different order," using his own sitcom "Squat" as the example. Why he thinks AI extinction warnings are "unfalsifiable" and lack evidence, compared to the more probable (but less dramatic) risks like hacking and misuse. His theory that AI labs face a trap similar to pharma R&D spending — except unlike pharma, they're "losing fortunes," which pushes them toward coordinated "slow down" messaging. His prediction that the AI investment cycle mirrors the dot-com bubble, and that the fix is models designing more efficient successor models and hardware — pointing to progress already seen in Chinese labs. 3. Notable quotes "There's no copying in any different sense that human beings copy when we create." — Bo Bennett "You could only hype it so much before the bubble bursts." — Bo Bennett 4. About the guest Bo Bennett runs a portfolio that spans both AI-driven content tools — AuthorVoices and MemoirMaker — and eBookIt.com, a platform for human-written books, giving him a front-row seat to the blurring line between AI-assisted and human-authored work. He's also a social psychologist and author who wrote the sitcom "Squat" long before AI tools existed, which he uses as a case study for how human creativity itself is built from prior influences. He's spent years cataloguing logical fallacies and brings that lens to evaluating AI-risk rhetoric from industry leaders. He states plainly he has no financial stake in any AI company beyond using their tools. 5. Topics covered AI vs Human Authorship Originality And Influence AI Existential Risk Claims Business Model Skepticism Dot-Com Bubble Parallels Pharma R&D Comparison AI Efficiency And Hardware Logical Fallacies In Rhetoric

2 Episodes

Researchers, engineers, and operators on the practical state of AI.

Creator

Clara Nguyen
Years Active

2026
Episodes

2
Rating

Clean
Copyright

© AIHosts.fm
Show Website

AI Frontier