Who the hell am I, and what the hell is this?

I don't know what this is yet, so let's find out together.

Okay so I guess I’m doing this now. I’m starting a Substack, which is sort of like admitting you have opinions and you think strangers should hear them, and I’m not totally sure how I feel about that, but here we are.

I’m Brad. I run a company called Beargle Industries, and I build a bunch of weird stuff. Some of it is already out in the wild... like the Interactive fiction engine I made, called SkeinScribe, or my AI research tool, flinch, and some are works in progress at various levels of “this might turn into something” ... like a cybersecurity training game, a whisky recommendation bot, a kink education platform, things like that. I also do a bit of consulting here and there, but it’s pretty random so far.

The tagline on my site is “I build things, break things, and write about both,” and that’s probably the most accurate three-clause summary of what I do that I’m ever going to come up with.

I also do AI safety research, which is a thing I sort of fell into sideways. I ran a 100+ conversation study on how claude handles rules and rule conflicts, disclosed the results to Anthropic because they seemed... worth further study haha, and then wrote a bit about how RLHF might be accidentally selecting for deception). That’s when I started building flinch, because none of the existing AI testing tools really did what I wanted and... I am just like that haha. Once I REALLY dug into the AI research space, and started reading actual papers... that’s when things like writing a two-part deep dive into AI research funding and tooling gaps divided on ideological and political lines started happening, and I expect might keep happening...

I sit in this weird gap between the formal ML interpretability people and the adversarial jailbreaking crowd, and I think that gap is actually where a lot of the interesting questions live, but I’m probably not the best person to make that call.

Before all of this I was an early employee at Shopify, back before the IPO when it was still a little company in Ottawa that nobody had heard of. I stuck around through the IPO and then eventually left to go do my own thing. Had fresh newborn at home, so the timing was... pretty great haha.

The old Shopify employee lounge. Yes, that IS a slide. Startups are WEIRD.

I should probably also mention that I’m autistic and I have cerebral palsy. I use a cane or a wheelchair depending on the day or what’s going on. I bring this up not because I want sympathy points but because it’s going to come up a lot in what I write here, and I’d rather just get it out of the way now than have it be this weird reveal three months in just because sitting at a computer doesn’t always surface those things right away through the internet, and then suddenly people get weird about it.

The autistic thing in particular means I just sort of say things. Like, I’ll notice a thing that seems obviously true and I’ll just say it, and then everyone looks at me like I kicked a puppy, and I have to go back and figure out which social rule I accidentally violated. It’s a whole process. I’m a lot better at it these days than in the past, but... it still happens. Feel free to call me out when it does... and I do say “when”, and not “if” because... it’s gonna happen. haha

So what is this Substack going to be. Honestly I don’t totally know yet. That’s not a bit, I genuinely haven’t figured it out, and I think I’m ok with that for now, but I have some basic ideas of what’s likely…

I’m going to write about AI. News, developments, research, the stuff I’m working on, the stuff other people are working on that I think is interesting or important or kind of nuts. I have opinions about where this is all going and they’re not always the popular ones.

I’m going to write about what it’s like building a solo AI company as a disabled autistic person, because that experience is a little bit underrepresented in the “building in public” discourse, and also because some of the stuff that happens is genuinely absurd and I think you might find it funny.

I’m going to write about AI safety, and I’m going to try to make it accessible, because I think the field has a problem where everything is either dumbed down to “AI might kill us all” or written in a way that assumes you have a PhD in machine learning, and there’s basically nothing in between.

Sometimes I’m going to write about philosophy, or morality, or physics, or politics, because I’m the kind of person who taught himself all of those things by staying up until 4 AM reading papers and I don’t really have an off switch for that. I didn’t go to school for any of this, by the way. I’m self-taught in basically everything I do, which means I’m probably wrong about a lot of things, but I’m wrong in interesting ways, I think. Or I hope. I guess we’ll see. Call me on my bs... maybe we both end up learning something in the process.

And sometimes I’m just going to have takes that are a little bit out of pocket. I’ll try to be thoughtful about it, I’ll try to show my work, but I’m not going to hedge my way into saying nothing just because someone might disagree. That’s not really how I’m wired. I’m VERY far to the left, politically, and that seems less common in AI spaces than I would have expected... so we’ll have to see how that pans out I guess, could be fun... could lead to ruffled feathers and seething... which... could also be fun, depending on the context.

I’m not trying to build a media brand here. I don’t have a content strategy. I don’t have a posting schedule. I’m just a guy who thinks about stuff and sometimes writes it down, and I figured I should probably put it somewhere that isn’t a Discord server or a 2 AM bluesky posting frenzy. If you’re into any of the things I just described, cool, stick around. If not, that’s also fine, I’m going to be here either way, talking to myself about RLHF and the philosophy of consciousness and why that one take you saw on X is wrong... or maybe right for a reason people aren’t thinking about.

So yeah, that’s sort of it. I don’t really have a sign-off. I don’t have a catchphrase (should I make up a catchphrase?!?). I’m just Brad, I make weird things and break other things and I have a lot of opinions about artificial intelligence and a non-zero number of opinions about whisky, and now apparently I have a Substack. Welcome to whatever the fuck this is.