S27:E8 - Learning AI (Matt Eland)

May 22, 2024·51 min·4 clips

Matt's AI squirrel learned to herd a rabbit into a dog to survive, surprising even him.

1. CodeNewbie S27:E8 features Matt Eland, AI specialist at Leading Edge consulting firm, discussing how he developed his career in AI and now teaches those skills to others. 2. Eland grew up with one of the earliest IBM PCs in his household in the 1980s and credits King's Quest text-adventure games with showing him that computers could tell stories, sparking a lifelong interest he pursued starting around age seven. 3. In high school, it dawned on him that his programming hobby could be something people might pay him for, at which point he decided on a computer information systems degree in college. 4. Eland struggled with depression from high school through his mid-20s, which made college difficult, but he completed a degree in computer information systems and later returned for a data analytics master's degree at Franklin University, expected to complete in summer 2024. 5. His first full-time job was writing Java as a contractor at a large telecommunications company, which he left after eight months when his boss departed, transitioning to the .NET ecosystem and smaller startup-culture companies that he found more fulfilling. 6. His specific fascination with AI began with playing the original Half-Life, where enemy grunt AI flanked him and flushed him out with grenades — prompting him to wonder 'how did they know to do that?' and connect this to an AI technology book his father had given him in the late 1990s. 7. He built a genetic algorithm squirrel simulation on a checkerboard — squirrel, acorn, tree, dog, rabbit — and evolved a weight function scoring the squirrel on survival and acorn retrieval over approximately 20 generations. 8. When he added bonus points for the rabbit dying without programming any attack behavior, the algorithm evolved an 'attack squirrel' that herded the rabbit from the opposite side into the dog, then collected its acorn — a solution Eland says he did not anticipate and that convinced him to pursue AI further. 9. He shared the squirrel project on DevTo approximately a decade after building it, receiving community feedback about libraries he had missed and language suggestions for F-sharp implementations. 10. Eland's decision to pursue his data analytics master's degree was triggered partly by going through Coursera specializations on Python and AI after a family loss, and partly by his wife telling him 'you're going through harder classes than I ever did for my master's — just get a master's degree already.' 11. He is currently working on a data science in .NET book — his second book — and about to start his second LinkedIn Learning course, alongside creating short tutorial videos on topics ranging from for loops in C# to how GPT-4 Turbo analyzes images. 12. He taught at Tech Elevator boot camp for three years starting in early 2020, working with cohorts of approximately 18 students over 14-week programs, which he describes as among the most fulfilling work of his career. 13. During teaching, almost every cohort's first multi-class design project produced five or six students in tears saying they could not figure out how to make their classes interact, and Eland describes his approach as asking what specific part is frustrating them, confirming they understand for loops and classes, and reframing the difficulty as 'lifting a new kind of weight.' 14. Teaching during the 2020 pandemic meant students were learning remotely for the first time while uncertain whether there would be jobs to graduate into — Eland describes it as 'trying to keep doing what we're doing while the air raid sirens are going off.' 15. He observed in 2022-2023 that students were asking whether it was worth learning to code given AI advances, and describes these as 'very real conversations to have.' 16. His core argument about AI and developer jobs is that the fear assumes writing code is the bottleneck, but software engineering is primarily maintaining existing code, fixing bugs, understanding business intent, integrating new features into existing architecture — tasks at which current AI tools are weak. 17. He notes that AI is also weaker with new technologies that have little training data and stronger with well-established patterns — meaning developers working with cutting-edge tools face a different risk profile from those maintaining legacy systems. 18. The interview is structured and conversational, with Saran frequently asking 'what did that feel like?' to draw out the emotional and motivational dimensions alongside the technical narrative. 19. Eland's tone is self-aware and reflective throughout, willing to name personal struggles including depression, imposter syndrome, and the irrational role of credentials in granting self-permission. 20. This episode is well-suited for developers who are mid-career or considering tech education who want to understand what AI tools actually change about software engineering — and what they do not.

As heard by us

A grounded AI-learning conversation about fundamentals, small projects, community feedback, and knowing when to move on.

CodeNewbie's conversation with Matt Eland treats learning AI as a developer practice, not just a career label. Eland, an AI specialist at Leading Edge, argues for keeping the fundamentals close enough to read code and reason about software design, while accepting that modern…

Read the full review in PlayNext →

Why you'd press play

If AI feels too big, this gives you a smaller way in.

Read the full recommendation in PlayNext →

Listen to the show on

Apple Podcasts Spotify

Play clips from this episode More from CodeNewbie