My Position on AI, LLMs, and "vibe coding"

I’ve been a strong AI skeptic since the current hype wave started. I’ve seen a few of these hype waves come and go in my twenty-odd years as a software engineer. I understand that climate change is the single biggest existential crisis facing humanity, and the current GPU-farm landrush is not helping with that.

I worry about the weaponisation of misinformation, propaganda, and manipulation. I lament the creatives who spent a lifetime honing their skills and developing their style and creative vision, who are being replaced by slop machines. And while I think copyright is deeply flawed, it makes me bitter to think of all the legal action against bittorrent users, Pirate Bay, SciHub, to think of Aaron Swartz, and to then see big tech doing the same things but at an unfathomably larger scale, unchallenged, with impunity. The double standard is staggering.

And yet, you are going to see me post in the near future about LLMs, about AI and AI-assisted coding. For a little while now I have actively been exploring that space. Does that cause me a good bit of cognitive dissonance? Absolutely.

But as a software professional, as a consultant who helps people navigate the industry, I have to have an informed opinion. The only way to do that is to really dig in and get first hand experience. Unfortunately, I don’t think this one is simply going to blow over.

It’s also a scary time in the industry. We’ve noticed first hand how budgets have dried up, how the mass layoffs over the last few years have altered the landscape. And yet there’s still as much capital as before looking for good investments. There’s ample funding available, if your company does something AI. Gaiwan is a small company, but including dependents there are a dozen people who depend on my business decisions for their daily income. I owe it to them to think about these things, and to make sure that we stay relevant.

Perhaps this bubble will pop dramatically, but I’m getting the sense it’s going to have a lot of staying power. People are throwing anything and everything AI at the wall, and some of it is sticking.

And so I am trying to be informed, dispassionately. Neither a hype man nor a hater. Fully conscious of the externalities, of the risks, of the shortcomings. And as I’ve done over the past two decades, charting my own path. Never accepting best practices or industry standards as gospel, but critically evaluating, and helping people around me make better more informed decisions as well, including my colleagues at Gaiwan, our clients and partners, and the community of practice with whom I’m hoping to share my learnings.