In the premiere episode of our Expert-Led Fireside Series, we sat down with Hobson Lane (CTO of Tangible AI), an AI researcher and ethical AI builder, to look past the hype and dust off the reality of modern artificial intelligence.
We unpack AI terminology, explore practical alternatives to general-purpose chatbots, and discuss the risks of outsourcing our thinking. From smaller models and private data to human expertise and responsible adoption, this candid conversation helps business leaders look beyond the hype and jargon.
We explore in three parts:
A FISTFUL OF DOLLARS:
What's the potential harm of using Large Language Models (LLMs) like ChatGPT, Claude, and Gemini: What they are bad at, and What they're good at.
FOR A FEW DOLLARS MORE:
What is "good old fashioned AI" (GOFAI) and other forms of Artificial Intelligence and Machine Learning (AI/ML). What they do better than LLMs, such as reliability or logic.
THE GOOD, THE BAD, & THE UGLY:
A candid discussion on the "doom and gloom" scenarios of the future of AI and the potential silver linings.
[Recorded December 17, 2025 with a live audience online]
Key Takeaways
AI isn’t synonymous with ChatGPT or other Large Language Models (LLMs). Understanding the differences between language models, machine learning, and rule-based systems helps you choose the right tool for the job.
Fluent answers aren’t proof of sound reasoning. Hobson encourages listeners to treat chatbot responses critically rather than mistake confident language for expertise.
Bigger isn’t always better. Smaller, specialized models paired with clear rules and quality data can be worth exploring for focused business applications.
Your internal knowledge can be a valuable resource. Retrieval-augmented generation (RAG) connects language models to selected information sources; local deployment can offer more control over sensitive data.
Human judgment still matters. The conversation highlights the importance of preserving expertise, developing skills, and using AI to support—not replace—critical thinking.
Key Timestamps
These timestamps refer to the original VTT recording, not the edited MP3. Update them against the final audio before publishing.
04:33 — What is AI, beyond the chatbot hype?
07:52 — What is a large language model? What is Machine Learning? Explained through everyday examples.
11:03 — Agentic AI and the limits of language-based decision-making
18:04 — Smaller models and specialized business applications
19:35 — RAG: connecting language models to your internal knowledge
21:03 — Privacy, local models, and keeping control of your data
21:51 — “Good old-fashioned AI” in everyday life
25:24 — AI risks, technical debt, and the erosion of expertise
32:53 — Final takeaways: choosing the right tools and trusting your judgment
Links and resources
About Hobson Lane and Tangible AI
Tangible AI — Learn more about Hobson’s work.
Natural Language Processing in Action, Second Edition — The book mentioned in the episode, authored by Hobson Lane and Maria Dyshel.
Email Hobson: hobson@tangibleai.com
What’s one task where AI genuinely helps your business—and one where human judgment remains essential? Share your takeaway, send this episode to a fellow founder or business leader, and follow The Collective for more candid conversations with mission-driven experts.
Subscribe to the Aurelius Newsletter to get notified when new episodes drop and learn about other resources and opportunities for entrepreneurs.