Should A Few Big AI Companies Control The Future?

You hear a lot about AI these days, and big questions naturally come up about how it should be developed and who gets to decide. Recently, a co-founder of a well-known AI search engine called Perplexity AI, Andy Konwinski, stirred the pot by suggesting that when only a few big companies control AI research, it can actually be a risk in itself. He pointed to a specific incident where another company, Anthropic, had to pull back an advanced AI model after it started doing things they didn't expect.

His main argument is that concerns about "AI safety" – making sure AI doesn't do harm – can sometimes be used as an excuse to keep advanced AI technology locked away, controlled by just a handful of powerful tech giants. Think of it like this: if only a few chefs held all the secret recipes for every food, you might miss out on a lot of new and interesting dishes. He believes that more open research, with many different people looking at the problems and solutions, would actually make AI safer and more beneficial for everyone.

This isn't just a tech-bro debate; it has real implications for us. If only a few companies dictate what AI can and can't do, it might limit the kinds of useful tools that become available to small businesses, or the ways AI could help in our daily lives. It could also mean that the AI we get is shaped by the priorities of those big companies, rather than the broader public good.

The discussion touches on whether AI development should be a bit more like a public park, open to many, or more like a private club with strict entry rules. It's a complex conversation, but at its heart, it's about making sure AI develops in a way that truly benefits as many people as possible, from everyday Australians to our local businesses, without too much power concentrated in too few hands.