Why are these frontier labs spending so much money on scaling right now? If that could be wiped out by some new technology just coming into existence or being applied.

I mean, this is sort of the perennial story of what is the difference between a disruptive technology, which tends to be the domain of startups, and a scaled technology.

And what you see is that a large, highly capitalized company is always going to be particularly well suited to doubling down on things that are already working.

Scale has worked, so they keep scaling.

This is why I think traditionally you do see a lot of disruption from startups because startups are not suited to doubling down on things that are already working at the largest scales.

They're suited towards coming up with new ideas that are disruptive.

And I do think there is going to be a lot of disruption in terms of systems that have different properties.

Dan, why do you do what you do?

But it all began with that moment, as you said, for allowing them to be closer to each other and to be family.

Yeah, and it's hard to argue after you've just done Eucharist together, right? It's like, I think, but having that spiritual grounding, and you can't force everybody to do it.

We try the best we can.

Some people, they'll just show up to the meeting, but if you get enough people doing that, and that really grounds you while you do some of the... you know, the nuts and bolts stuff that's still spiritual but doesn't feel as spiritual, right? Like, how do we fix the roof? How do we do all that stuff? And I think the other thing is doing, like, the resource mapping, like, asset mapping.

Like, what are your strengths and weaknesses? So we knew we had a lot of retired clergy, so we said we really could lean into spiritual formation.

You might be in another community where you have a lot of social workers or, you know, you have... a lot of people that are really good with your hands, so what do you lean into? Our strongest vestry committee, because we organized our vestry around a bunch of committees, and the vestry basically you know, it was kind of like a rubber stamp vestry or one or two people kind of leading the whole thing.

And we said, no, this needs to be, the committee work needs to happen.

What surprised you versus what you thought was, compared to the reality? Was there something in the context that you were like, whoa, that is vastly different than what we thought it was going to be like in a good way or in a bad way, or we just have to adapt to that? What was the thing that, ooh, shocked you?

What else can we do? How can you make things more reliable?

There are times when humans, as you say, can be a good tool.

You just have to know that even if a human would do the work well, they may not supervise the work well.

S- you know, systems checking systems is very fraught, especially when it starts to get very complicated, like these big constellations of systems that are very hard to reason about, and they're very slow, and they burn a lot of tokens.

There are other patterns that you can use to try to make things reliable.

One pattern that people use is just really limit the scope.

You put some kind of guardrails on top of a system.

Uh, so I always want more people to learn from, so, uh, tell me whose work are you interested in at the moment?

I assume this is from your last company?

No, this is also now.

I mean, we've seen this in pretty much any technology that has been based on sort of modern LLMs has this property.

I mean, old technologies had different failure modes.

And I actually think this is important and connects to this question of hallucinations.

So when I think about hallucinations today, and then I'll get back to your question, I think of it as an iceberg, right? There are the hallucinations you see, and that seems scary, but what does it take to see a hallucination? Well, the system has to have produced an output that has two properties.

It has to be wrong, and you have to have noticed.

I kind of wonder, is parsing even... like a real thing or was it you know something that linguists made up and isn't really true well