Vox

Leitura de 14 min

Using LLMs To Expand Your Working Vocabulary

Summary

The episode focuses on the importance of building your conceptual vocabulary as a fundamental, high-leverage skill for navigating a career in tech. The host argues that vocabulary goes beyond just words to include mental models, paradigms, design patterns, and diagram types (like causal loop diagrams or C4 diagrams). This expanded vocabulary helps developers find clarity, perspective, and purpose by providing the mental tooling to understand complex systems and make better decisions.

The discussion positions vocabulary as the building blocks for understanding. Learning a new paradigm, like asynchronous programming or event-driven architecture, or a new diagram type gives you a new lens through which to view problems. The host shares personal examples, like exploring causal loop diagrams to understand system dynamics and leverage points, to illustrate how new vocabulary pieces can be combined to form a more complete picture of any subject or challenge.

The host then introduces LLMs as a particularly useful tool for expanding this conceptual vocabulary. Because LLMs are trained on vast corpora of human knowledge, they can help identify the standard tools, models, or visualizations (like a scatterplot for X,Y data) that are commonly used for a given set of primitives or problems. This allows for efficient exploration of new domains by asking questions about fundamental system components (inputs, outputs, flows).

The final message is to make the conscious pursuit of vocabulary expansion a key part of your learning journey. The high-leverage skill is not just learning how to use a tool, but understanding why and when to use it versus another, and how different pieces of vocabulary fit together to generate insights. This meta-skill of continually adding to your conceptual repertoire is presented as one of the most valuable career assets you can develop.

Recommendations

Concepts

  • Mental Models — Referred to as a form of vocabulary that helps in attaining clarity, perspective, and understanding purpose. They are part of the conceptual latticework you build.
  • Causal Loop Diagram — Mentioned as a specific piece of vocabulary the host has been working on. It’s a tool for understanding system dynamics, identifying leverage points, and seeing what kinds of things are happening in a system.
  • C4 Diagram — Cited as an example of a piece of conceptual vocabulary for software architecture. It’s distinguished from other types of charts or diagrams, like a pipeline.
  • Asynchronous Paradigm — Given as an example of a programming paradigm that constitutes vocabulary. Not having experience with it means lacking access to that way of thinking about problems.
  • Event-Driven Architectural Paradigm — Another example of a paradigm that serves as vocabulary, providing a specific model for thinking about system design and interaction.

People

  • Paul Graham — Referenced for discussing how understanding the Smalltalk programming language provided a unique advantage due to its fundamentally different paradigm, illustrating the power of specific vocabulary.

Tools

  • LLMs (Large Language Models) — Recommended as a useful tool for expanding conceptual vocabulary. They are good at helping you discover standard models, diagrams, or visualizations commonly used for a given set of system primitives or data types.

Topic Timeline

  • 00:00:52 — Introducing the fundamental mindset of building vocabulary — The host introduces the core thesis: building your vocabulary is a fundamental mindset for career resilience through transitions, disruptions like AI, and role changes. This vocabulary is positioned as essential for finding clarity, perspective, and purpose—the core mission of the podcast. The concept is expanded beyond literal words to include mental models and conceptual tooling.
  • 00:03:05 — Vocabulary for self-understanding and system primitives — The discussion connects vocabulary to self-understanding, using the analogy of languages having specific words for nuanced feelings. Developing a nuanced vocabulary helps you understand your own values, stress responses, and operating modes. The host then shifts to concrete examples, starting with understanding different forms of diagrams and system primitives (inputs, outputs, flows) as foundational vocabulary pieces.
  • 00:05:29 — Programming paradigms and design patterns as vocabulary — The host uses programming paradigms (asynchronous, event-driven) and design patterns as clear examples of conceptual vocabulary. Learning these provides new ways of thinking that may not have analogs in other contexts, citing Paul Graham’s example of Smalltalk providing a unique advantage. Access to these vocabularies allows for more nuanced and powerful expression of ideas.
  • 00:07:14 — Using LLMs to explore and refine vocabulary — The host suggests LLMs as a practical tool for refining and exploring vocabulary. By describing the primitives of a system or subject you’re investigating, you can ask an LLM questions. Due to its training, it can likely surface commonly used models, diagrams (like a scatterplot for X,Y data), or concepts that many people have applied to similar primitive components, accelerating your vocabulary acquisition.
  • 00:11:48 — The high-leverage pursuit of categorical vocabulary — In the conclusion, the host urges listeners to make vocabulary expansion a deliberate part of their learning. The high-leverage change is not just adding discrete skills, but learning the categorical ‘why’ and ‘when’ behind tools and concepts. Understanding how to piece together different vocabulary elements to gain insights is framed as one of the most valuable career skills you can develop.

Episode Info

  • Podcast: Developer Tea
  • Author: Jonathan Cutrell
  • Category: Technology Business Careers Society & Culture
  • Published: 2025-06-25T07:00:00Z
  • Duration: 00:13:26

References

Podcast Info

Transcript

[00:00:00] one of those fundamental skills that will carry you through

[00:00:15] the most difficult parts of your career it’ll carry you through the transitions in your career

[00:00:23] it’ll carry you through disruptions like what we’re experiencing right now with AI

[00:00:30] it’ll carry you carry you through from title changes from hopping tracks let’s say you’re

[00:00:41] an engineer and you want to become a manager

[00:00:43] the fundamental mindset that I want to reinforce to remind you of today

[00:00:52] is the mindset of building your vocabulary building your vocabulary and to be clear

[00:01:03] uh I partially mean building your literal vocabulary but it goes beyond that and it’s

[00:01:13] part of the fundamental mission of this show of this podcast

[00:01:20] to

[00:01:22] to

[00:01:22] to

[00:01:22] to help developers to help engineers software engineers data analytics engineers

[00:01:28] product engineers front end developers people who are building things whoever you are

[00:01:37] you probably count in this okay to help you find clarity perspective and purpose in your career

[00:01:44] and finding clarity finding perspective

[00:01:52] and even finding purpose

[00:01:55] those efforts rely on

[00:02:01] the the mental tooling that we talk about so much on this show how do you find clarity

[00:02:11] how do you find perspective how do you find your purpose

[00:02:15] and the first building blocks are your values

[00:02:22] and your vocabulary

[00:02:22] so when we talk about mental models on this show

[00:02:29] that is a form of vocabulary

[00:02:34] the mental model concepts that you build that lattice work that is part of your vocabulary

[00:02:45] because it will help you in attaining for example

[00:02:51] clarity

[00:02:52] clarity. A mental model can help you in attaining perspective.

[00:02:59] A model can even help you in understanding purpose.

[00:03:05] Building your vocabulary is also about understanding your own self, the granularity

[00:03:14] of your own self. You know, we’ve heard stories about how there are different words

[00:03:21] in different languages. You can Google this. I don’t have a good specific example for this, but

[00:03:27] I’ll make something up here. For example, there may be so many words for the underlying concept

[00:03:36] of love, and a given language may have a really specific articulation of a feeling, an idea

[00:03:48] that another language doesn’t have.

[00:03:51] So this nuanced explanation, the nuanced vocabulary, the ability to combine these

[00:04:03] ideas together, this is a good analogy for understanding the variety of, for example,

[00:04:13] personal values you hold, understanding the way that you respond under stress,

[00:04:21] the kinds of things that you care about over time. How do you operate in a high intensity

[00:04:28] environment? These are all pieces of your personal kind of intellectual, emotional,

[00:04:37] spiritual, whatever, you know, AL word you want to use here. These are all types of vocabularies

[00:04:48] that you can develop for yourself.

[00:04:51] A more concrete example of this that I’ve experienced in my personal life is understanding

[00:04:57] different forms in my personal life. I mean, in my personal career life, understanding

[00:05:03] different forms of diagrams, very simple, you know, vocabulary that I’ve been working

[00:05:11] on recently. Understanding, for example, what is a causal loop diagram? Now, what’s really

[00:05:18] interesting is that we can build our vocabulary.

[00:05:21] And then we can take advantage of this vocabulary in new and unique ways. Another good example of

[00:05:29] this, there are paradigms in programming, paradigms and design patterns, those kinds of

[00:05:37] kind of models, ways of thinking about things. So if you have never experienced, for example,

[00:05:44] the asynchronous paradigm or the event-driven architectural paradigm,

[00:05:50] then this is a good example of a causal loop diagram.

[00:05:51] And those pieces of vocabulary, which may be useful, again, these are kind of like mental

[00:05:56] models, right? They may be useful in multiple contexts, but you won’t have access to those

[00:06:03] until you learn that vocabulary, right? So this is something that even Paul Graham talked about.

[00:06:10] He talked about the idea that, you know, understanding something like small talk

[00:06:16] provided a unique advantage that was a,

[00:06:21] kind of a leap ahead of other programming languages, because it has a fundamentally

[00:06:27] different paradigm. There are parts and pieces of how you would use small talk that

[00:06:36] don’t have a good analog with other languages. This vocabulary has been useful to Paul Graham and many,

[00:06:50] many other,

[00:06:51] engineers who have found tools that express what they’re thinking in more nuanced or more powerful,

[00:06:58] higher leverage ways. So if you are able to develop this vocabulary, and I would say

[00:07:05] you have an opportunity to continue refining this vocabulary by learning about it and exploring the

[00:07:14] nuances of it with, with an LLM, for example, you’re going to be able to wield,

[00:07:21] a more complete kind of picture in whatever endeavor you’re, you’re moving into. So for

[00:07:32] example, if I wanted to explore a subject that I’ve never, that I’ve never thought about before,

[00:07:38] now I have some, some tools at my disposal. I will, you know, search for various systems

[00:07:47] and try to develop an understanding of the causal loops.

[00:07:51] In the system that’s at play. This helps me get a picture for what kinds of things are happening.

[00:07:58] Where are the leverage points? What might I do to change things, to balance something out?

[00:08:04] You know, what does the system look like? And I can do a lot of this work now,

[00:08:09] just by having the vocabulary. I don’t necessarily have to have,

[00:08:13] you know, every piece of skill, underlying skill or perfection of that vocabulary.

[00:08:20] But,

[00:08:21] but having that top level, and understanding those concepts so I can piece them together.

[00:08:29] This is probably the most valuable skill that I’ve learned, that I’ve picked up in my career,

[00:08:35] the ability to add to your repertoire, to add to your overall vocabulary, some of these

[00:08:43] ways of thinking. I won’t limit it to mental models because I do think it goes beyond that,

[00:08:50] that the concept of,

[00:08:51] of a mental model captures some parts of this, but it’s not entirely complete.

[00:08:57] You know, for example, understanding what a, what a C4 diagram is.

[00:09:02] This is a piece of vocabulary. It might be, you know, you could kind of squint and call that a mental model,

[00:09:09] but really it’s just some vocabulary. And, and here I, you know,

[00:09:15] I want to be careful that you understand what I mean when I say vocabulary.

[00:09:19] I really mean conceptual vocabulary.

[00:09:20] I don’t mean just one word, but also what’s the connotation of that word?

[00:09:27] Where does it come from?

[00:09:29] When is it useful?

[00:09:30] How does it piece together with other things that are adjacent?

[00:09:35] You know, when, when would you care about seeing a C4 diagram versus say a pipeline?

[00:09:42] You know, that kind of, you know, charting, there are plenty of opportunities to develop your vocabulary.

[00:09:50] As you learn about these various things, what, you know, one,

[00:09:56] one way that I would suggest that you pick up as soon as possible is to start with the fundamental kind of primitives,

[00:10:05] the principal pieces of your system.

[00:10:09] Think about the inputs, the outputs, the flows.

[00:10:12] Is this a static system?

[00:10:14] Is it a dynamically changing system?

[00:10:16] You know, the thinking in the systematic terms,

[00:10:20] can help you understand what are the primitive, you know, interacting pieces.

[00:10:25] And then you can think about, okay, you know, what, what am I trying to get out of this?

[00:10:29] And am I trying to understand, you know, how these things work over time?

[00:10:34] Maybe there is some kind of time series graph that would be a piece of vocabulary.

[00:10:39] So, you know, one, one incredibly useful tool in this regard, and I won’t recommend using LLMs for everything in your life,

[00:10:47] but this is a place where an LLM does a fairly good job.

[00:10:49] It’s a place where an LLM does a fairly good job.

[00:10:50] And the reason for that is because you can describe these primitives and you can ask essentially ask questions.

[00:11:00] And because of the way LLMs are trained and because of how the kind of lookup will work,

[00:11:07] you’re going to find something that many people have likely used for that type of primitive.

[00:11:16] So if you had X, Y data, you’re likely to find a scatterplot graph.

[00:11:20] As a tool that you might use to visualize that X, Y data.

[00:11:25] This, this is a remarkably good tool for expanding your vocabulary for that reason.

[00:11:34] Thank you so much for listening to today’s episode of developer T.

[00:11:37] I hope you will make it a point, you know, in your learning journey in particular,

[00:11:44] make it a point to always be expanding your vocabulary.

[00:11:48] This means that you’re looking for the,

[00:11:50] the leverage points at the categorical level.

[00:11:53] You’re not just looking to,

[00:11:54] you know,

[00:11:55] add a little skill here and there.

[00:11:57] You’re not just looking to learn how to use a thing,

[00:12:00] right?

[00:12:01] That is a lower leverage change to your skill level,

[00:12:08] right?

[00:12:08] The much higher leverage is to know why you’re using that thing in that context.

[00:12:14] Why would you use that thing instead of another thing?

[00:12:17] When would you use something different?

[00:12:20] Uh,

[00:12:20] you know,

[00:12:20] how can you piece together your various pieces of vocabulary to get the things

[00:12:26] that you care about,

[00:12:27] to get the insights that you care about,

[00:12:29] et cetera.

[00:12:29] Thanks so much for listening.

[00:12:31] Uh,

[00:12:31] if you enjoyed this episode,

[00:12:32] please let me know.

[00:12:34] Uh,

[00:12:34] you can join the developer T discord community,

[00:12:36] head over to developer T.com slash discord.

[00:12:39] I’m in there.

[00:12:39] And so are a bunch of other engineers who listen to this show regularly.

[00:12:43] Um,

[00:12:44] you can also email me.

[00:12:45] We don’t talk about the email very often,

[00:12:47] but developer T at gmail.com.

[00:12:49] Uh,

[00:12:49] you can email me questions,

[00:12:50] you can,

[00:12:51] uh,

[00:12:51] send me comments.

[00:12:53] Another great way to leave a comment,

[00:12:55] uh,

[00:12:56] about the show is to leave a review,

[00:12:58] leave a review in iTunes,

[00:12:59] whatever podcasting app you use,

[00:13:01] but iTunes is the most impactful to help others find that review and make a

[00:13:07] decision about listening to the show.

[00:13:08] Of course,

[00:13:08] we’re on Spotify.

[00:13:10] We’re on,

[00:13:10] uh,

[00:13:11] we’re actually on YouTube as a podcast.

[00:13:14] Um,

[00:13:14] so you can find that there as well.

[00:13:16] Thanks so much for listening.

[00:13:16] And until next time,

[00:13:17] enjoy your tea.

[00:13:20] Bye.

[00:13:24] Bye.

[00:13:24] Bye.

[00:13:24] Bye.

[00:13:24] Bye.

Visão de gráfico