Limits to Growth of LLMs w/ David Helmer

Yeah.

So, uh, before I was a jet engine guy, my dissertation was actually about validating uncertainty quantification for computational models.

So how do we decide if these big fancy models are right? Because, uh, in fluid mechanics world, it's really expensive to do testing, as we talked about.

Um, so after I was at GE, where I developed a lot of methods, uh, at BCG, where I did a lot of work on analytics, uh, I went to the applied physics lab, where I spent basically a decade advising the government on advanced tech.

A lot of that conversation was about AI.

Um, so you started to have a lot of the machine learning entry.

Uh, since I have the background in engineering and I understand the math, but also could understand the big picture, I've done a lot of work advising the government, uh, on things like whether autonomous systems should be invested in or different AI capabilities.

Uh, I've led teams that develop these kind of capabilities, uh, for the government.

Uh, and then, uh, I also taught, uh, introductory AI and some other topics in that space, uh, at the academy.

Uh, so I've spent the last 10 years in that space.

Uh, you're not gonna hear me rambling about, uh, details of mathematical methods.

I can if you need me to.

No, please no.

Um, but a lot of my conversation, uh, tends to be about the fundamental principles.

Uh, and my company now advises, uh, generally non-technical experts, although technical experts as well, on what these systems can really do.

So I do legal education, senior leader education, uh, diligence activities for, for, uh, investments and things like that.

Uh, so that's the world I live in now.

Awesome.

Well, again, looking forward to, uh, to jumping into this.

Uh, you know, I was, uh, being interviewed by Oliver Stone's team for a, uh, part two of their Nuclear Now movie.

Um, and one of the things I was reflecting upon was, you know, as a, uh, bright-eyed, bushy-tailed nuclear advocate, I sort of got into nuclear out of, you know, these really profoundly sort of humanistic concerns, uh, partially around things like climate change, you know, air quality.

Um, we had a big coal phase out in my province, uh, which was accomplished through nuclear power.

You know, good, solid, you know, working class jobs that could, you know, a single income could support a family, sort of that Homer Simpson ideal.

Medical isotopes.

God, I don't know what else, but it was a really sort of idealistic, you know, in retrospect, slightly naive.