Beyond the Hype: What Works in AI Right Now (And What's Complete BS)

q01Shawn, tell us how you got into AI engineering and what sets your approach apart?

Shawn: I have a long history in computer science and systems engineering, went back for an MBA, and realized we often build systems for the system, not the user. Now I focus on identifying human problems and building systems for those humans.

We also upcycle furniture as a hobby—real hands-on work that grounds me. AI is similar: you can’t just throw AI at a problem and expect magic. You need engineering rigor.

q02So many claim AI will replace engineers. Do you believe that?

Shawn: No. Engineering is about breaking complex problems down into solvable pieces. AI tools alone can’t do that decomposition, especially because LLMs try to swallow entire complex problems at once. They can hit 95% but getting to 100% is where real work is.

Systems thinking and engineering discipline are irreplaceable. Engineers will always be needed.

q03Explain the “flashlight vs electrical grid” analogy in practice.

Shawn: Flashlights shine on small problems, but electrical grids power entire homes. Building AI into business is like building a grid—embedding AI in SOPs, knowledge bases, videos, docs, parsed and made queryable.

For example, a new employee can ask a chatbot housing all SOPs instead of scrolling through manual after manual. That’s the power of AI embedded into systems.

q04You work with offline AI tools. Why is that important?

Shawn: Offline or local AI means air-gapped models that never connect to the internet. This is mission-critical for fields with strict data privacy—lawyers, doctors, government.

Removing internet connectivity slashes attack vectors, fulfilling compliance like HIPAA or EU AI Act automatically. Your data literally can't leak if it never leaves the premises.

q05What about companies that claim they anonymize or scrub data but still generate realistic voices or images?

Shawn: That’s a contradiction. You can’t scrub data and then reproduce personalized voices accurately. There’s something wrong in their claims. Real data is being retained and used.

It shows how privacy promises don’t always align with reality.

q06What’s the difference between “human-first” and “AI-first” approaches?

Shawn: Human-first builds automation to augment humans, inspired by Henry Ford’s assembly lines. The goal is better tools for people.

AI-first tries to replace humans wholesale, often ignoring human needs, leading to brittle, overhyped products. Human-first respects the complexity of problems and human behavior.

q07Which AI trends would you bet money on surviving the hype?

Shawn: Small, domain-specific models tuned for niches will dominate. Larger models remain for general tasks but won’t replace specialized ones.

Image and video generation are expanding rapidly and have clear value.

q08One thing AI skeptics get wrong?

Shawn: That you don’t need to understand how AI works. Ignorance fuels hype and mistakes.

Final Thoughts: Building Real AI in a Hype World

Shawn David cuts through AI hype with a human-first engineering approach. Practical success demands systems thinking, privacy-first offline AI, and embedding AI into real business flows.

For businesses and builders: focus on niche, domain expertise, and robust engineering. AI is a tool, not magic—and the future belongs to those who wield it wisely.

Related Shorts

Explore short-form videos that expand on the ideas covered in this blog.