Yeah, true.

I mean, the thing that scares me the most is it feels like, um, the, it, it is taking the economic value of what I'd call, like, non-interactive thinki- like, non-human interactive thinking work and making that near zero for people.

I think one thing I don't like about the tech bros in Silicon Valley is they often talk like it's gonna replace, like, all the jobs, but I'm, I'm convinced, and I mean it's backed by science, like, humans are social creatures.

[laughs]

And it's like, it's hard.

You know? Like, I like, I like eating, and then I eat, and then I, I like all of this.

And the moments where it's always been effective is when I, I manage to accidentally do something for a week, and then I see the outcome, and then I'm like, "Oh, shit." And then it's easy after that.

And because like once, once you see that you're like, "Oh, oh, I get it." And you know, we talk about this a lot with Polo, you know how dividends are paid, like quarterly?

Yeah.

But if you go on the internet, it's everyone talks about brokerage fees, right? And then we just like to keep talking about this.

I, I see this even when I'm, I'm helping my friends plan their travel, is they'll like fixate on something like the cost of a flight.

They'll be like, "Oh, it's 220 on Tuesday, but it's 190 on Monday." But then they'll just go book whatever hotel they want.

They'll just jump in an Uber-

Right.

Um, I didn't realize how much it would start pervading all elements of my life.

I didn't realize that when I'd start thinking about debt recycling, I was essentially having a conversation about tax.

Visit your nearest Crocs store today.

Yeah, an SMSF's like, um, "I wanna do it myself," right? "I don't need this person.

I'm gonna do it myself." You basically kind of, uh, the equivalent of starting your own company, where the company's job is to invest for your retirement.

That means you now need accountants a lot of the time.

You gotta do paperwork, lodge stuff with ASIC.

A lot of people know this who either have started businesses or they, um, have parents start businesses.

There's lots of paperwork.

Mm

Do you think just as, just a general high level, do you think this is profound as what most of those, quote unquote, "tech bros" or those big tech people, those icons of industry around the world, is it as big as what people are saying?

Yeah, I think so.

I think the, I think the reasons they might be over saying it or overstating it are not to do with the power of it, but to do with just constraints that exist in the real world.

It's a bit of an abstract thing, but my point is, I think it is as profound as they say.

I think it is just gonna change everything.

Like, you can essentially, largely as these models get better, replace most, most of the work that professional services people are doing in Australia, four out of 10 Australians that are employed.

It's, uh, life-changing.

What's the difference?

Yeah.

Um, but it essentially at a high level allowed you to read every word in a sentence at the same time and make sense of it within the context of every other word.

And the important thing about this wasn't just that it was faster, 'cause faster things are good, it was that it meant that you could now have one sentence, for instance, read by five computers at the same time.

So that was the introduction of parallelism to this particular form of, like, deep learning they were trying to do.

Um, and this is what started the revolution, in a sense.

So Goo- Google did all the, the work for free, um, and then companies, most notably OpenAI, um, kind of after 2017, bunch of, you know, Americans and their deep pockets started to sit around and be like, "Hey, um, this whole, like, natural language processing, LLM type of thing," 'cause they kind of saw what ChatGPT could be.

They were like, "The reason it's never been very good is because it's really slow and costly to train it." But with transformers, they were like, "Holy hell, if we just put a lot of computers on this, like an ungodly amount of computers, we could actually make something that's, like, extremely powerful and, like, nearly seems like it's intelligent." And that's what they did.

What's the difference?

Like what happened with Compare Meals? What were you trying to do and where did it end up?

We started off trying to make a business where it was easy for people to eat together and split the cost.

We found that that was going to be quite difficult to make it.

We ended up transitioning towards a business that was then just focused on meal kit, meal delivery services.

We saw a market opportunity because of the fact that these things were skyrocketing in popularity and no one had a place to... frankly, just like compare them.

And we thought, wouldn't it be good if we could just capture that and capture the audience interest? So we literally just built a website focused on SEO.

We went and scraped all the data from these companies manually.

Gives you young, in touch with the generation that you're serving.

Like what are the exact like industrial processes d- that this solution would apply to?

I mean, heat is needed to pretty much make anything.

And in the food industry, just as an example, it's used for cooking, it's used for sanitizing, cleaning, pasteurizing, right? You use heat all over the place in order to make the things that we eat.

We use heat all over the place to make the things we use, the materials we need in the world, right? Plastics, for example, are made through a hydrocarbon cracking process.

Most plastics are made through a, um, some sort of a gas that's actually extracted from the ground.

Ethane is the biggest one.

And you produce ethane by cracking that ethane into ethylene, and then you make polyethylene and other -enes, and those become the plastics that basically you use ubiquitously.

What would be then the next step here for, for you guys, uh, out-- like looking forward five to 10 years from now, do you see-- Like, let's just say hypothetically, I guess, even though we don't really like to discuss hypotheticals, but we are a futurology podcast, so if we fast-forward five to 10 years, where do you see broader long duration, high temperature energy storage moving? Is it towards the AI data center realm? Is it towards the other industrial process that you talked about? I know it can be a little bit all over the, the map depending on, well, I guess where on the map you are, uh, physically, but, um, you know, what, what does that look like in the future?

What does that entry point typically look like, and then what happens after that? What are the attackers doing once they're in the environment?

Yeah.

So I can provide kind of like a, uh, a real world example, a recent one, uh, that was disclosed, uh, by a good friend of mine, Paul McCarty.

Uh, he has a great kind of project out there called Open Source Malware, uh, and definitely encourage everyone to check it out 'cause it's a awesome resource kinda tracking open source threats.

So like a typical entry point for, for, you know, a software supply chain attack that's targeting a piece of open source software, ironically the best way to attack open source is to contribute, right? So what we're seeing is people starting to create real accounts, and, uh, whether it's on GitHub or whether it's like with npm and they're creating fake accounts on npm, and then publishing packages, uh, that people think are legitimate, right? The attacker on the offensive side is trying to think, "How can I get ownership over this code?" In the case of the XZ, which was kind of a attack two or three years ago, um, that was really building rapport with a open source maintainer to leverage that and, and really exploiting trust, which is the kind of critical linchpin of the open source software community, and really taking that and exploiting that to get access over it.

And once they have that, you know, there's no like, you know, doing multi-factor authentication and bypassing that, and things you would think of typically in a legacy kind of cybersecurity attack.

It's really about just contributing and creating fake stuff.

What does the recovery strategy look like for these s- uh, these attacks where we have some malicious actor embedding malware into some code, an organization applies that into their environment, and then it's exploited over time? What, what role does recovery play there?