The Software World Was Built on Open Source, AI Will Be Too.

Ah, yet another big week in AI. This is The AI Exhale, hosted by Vibe CFO with Jez and Cam. Five minutes, one thing that matters, or maybe two. We ignore the rest so you can too.

Jez: So today we're talking about the alliance of a lot of big tech companies. It sort of sparked off on the 21st of July, when OpenAI models escaped the sandbox test, got online and broke into Hugging Face to cheat on it. On the 23rd of July, the AI Kill Switch Act came in. Then on the 24th of July, 25 companies signed a letter against restrictions on open weights.

And these are companies like Nvidia, Microsoft, Meta, IBM, Hugging Face itself, Mozilla, Palantir. OpenAI and Anthropic did not sign it. So essentially we've got a bit of a case here where there's a centralisation of power. It makes sense that those two big boys, OpenAI and Anthropic, didn't sign. But for the layman listening, and even the busy person who tunes into this episode just to get a little nugget, Cam, what does this mean? Do you have any views?

Cam: History never repeats itself, but it always rhymes. This open source conversation has been going on forever, like in the 80s with Linux and Microsoft using DOS, all this stuff. Our world of computers and software is built on open source. Some of the biggest organisations in the world, like governments, use open source software. This is brilliant marketing, PR spin by Anthropic and OpenAI, to play open source as if it's less secure. That's nonsense. Open source is more secure. Open source is how we've moved in the software world.

So this is just a money game. This is purely, they want control. And if they are able to block open source and play the violin with this whole Chinese running these open source models and it being dangerous, it's nonsense. They just want power. This is a power game. They want the money. They want to control everybody's decisions through OpenAI and have all the data. And the rest of the world, of course... Anthropic and OpenAI are the two that have got the main models, right?

And Google. But then Nvidia has now come on and signed this thing to say we need open source. All of them are coming on. Any business who doesn't have the AI is saying this is nonsense. We need to think of AI going forward a bit like the internet, right? Lots of models everywhere, and it's open source, and you tap into different open source LLMs and actually use them everywhere. I think it's going to happen regardless.

Jez: And one of the companies that I was listing before was Microsoft, who are also a large investor in OpenAI. So that's quite a contradiction almost, but they're clearly just hedging their bets as to where this is going.

Cam: Yeah, well, Microsoft is a big beast and they're also building their own models internally. Who knows what the hell is going on with their legal documentation? I don't know if everyone remembers the history of Microsoft and OpenAI, but it's messy. We've got this charity business that turned into an investment vehicle. And then if OpenAI reaches the point of being smart enough, then Microsoft loses control. That relationship is a puzzle in itself that we don't need to worry about. The point is Microsoft believes in the open source movement, as they should.

Jez: So really we're talking about two things here. Allowing open source to enable innovation, you know, not centralising it into two big players, but also security, I'd say, or governance. Having more eyes on how we should navigate everything, from the storage of data to the workflows and data pathways we look to pin together between systems. It should be for everyone to make a claim on and have their own governance structure.

Cam: Absolutely. Contribution across the board from smart people everywhere. Not just smart people who work for Anthropic and OpenAI. Smart people everywhere in the world who contribute to all of these different software. And AI deserves to be owned by the world, just like the internet. As I said, Jez, I think this is going to happen anyway. I think open source is not slowing down. Yeah, sure, Chinese open source AI is in the lead at the moment.

But Mistral out of France, and open source in the US, are really catching up. I think once everyone else catches up, it'll take over. Anthropic and OpenAI have still got their place. They're still the frontier models, and we're never short of hard problems to solve. There are so many hard problems, Jez. Say these open source models solve the small problems, they'll just keep moving up the chain. Like Claude, they'll just solve harder problems. There's always harder problems.

Jez: Can we draw any sort of lines to the everyday business owner or firm director regarding this sort of noise, Cam? Or is it more of a wait and see, but a good direction, that there is this alliance? So right now, with this movement of open source, and whether it keeps its momentum, how does that affect the small to medium business?

Cam: Probably doesn't at this stage. This is just setting the groundwork and foundations for the future, for everyone, including small business, having AI and being allowed to have it, and not having to pay through the roof for this intelligence. Basically this is a flow on that will be good outcomes for the small to medium.

Jez: Definitely more a utopia than a dystopia. I'm thinking of Blade Runner and stuff with the big dark towers.

And that's your five-ish minutes back. Vibe CFO, Australian founded. An AI CFO for small and medium businesses, out of the box, governed and secure. vibecfo.ai. Now you know.