What is genuine in this noise?

I wish my conviction was that we should just disregard LLMs and that we will hand-code our way through. It is not what I believe but it is a much simpler belief to hold. The companies involved are noxious. More so than ever before.

And the technology will encourage you to deteriorate while frequently causing large negative side-effects. Environmental, social, ethical. Like fast fashion. Modern farming. Fast food. Ubiquitous entertainment. Constant connectivity. Commerce does whatever the it feels like to the commons in search of an outcome. And with the outcome, we are not well wired to moderate ourselves in regards to abundance.

But modern farming, industrialization and even fast fashion do mean that people are fed, dressed and often able to spend time on entertainment. I don't think "AI" will bring a new era of prosperity because I don't trust a single fucker at the wheel of the tech economy.

I also can't subscribe to the viewpoint of people who claim it is useless. It is here. It has obvious capabilities in terms of going from zero to significant thing in essentially no time. You can't trust the thing it built. You can't trust the research it did. But it is very often mostly right. This will appeal to a lot of people and absolutely disgust a few.

We already see it in the hands of marketing and sales. Workforces that are frequently judged on throughput more than rigour. Every damn event and blog post has a generated image and frequently obvious touches of LLM writing.

The noise floor of everything rises. More writing. More videos. More podcasts. More software libraries. More frameworks. Most wont continue unless they see traction so they are really more like pitch decks in product form.

It won't write for me. This is my voice.

Fundamentally the LLMs can perform work in creative projects. Goatmire was not conceived of by LLMs. They had no input in the idea. I wrote the script by hand. A bunch of the technical implementation used LLMs. The booking system for workshops was built in tech I chose and trust by an LLM in a process I could easily review. Every design decision on how it works was mine. Every line of code was Claude.

I am very glad that I know what I want to do. I have ideas, intent, plans and creative drives that take me in new directions. Without some level of creative integrity and intent you will become a complete slophose. A noise in the noise floor. What is exciting to me is that you can take that creativity to absolutely stupid levels of ambition if you apply the LLMs with direction and intent.

I see the projects I am involved in and interact with flourish in various ways in the LLM era. Nerves has an AI policy and we practice a lot of restraint. We are intended to be a solid foundation for you to build on. Big churn and lots of movement is net negative. But we are finding weird issues, performance improvements and security improvements with these models. We just don't want the models to make the project unmaintainable and we don't want other people to use their LLMs to talk to us.

Ash was always kind of a big brain concept. Now we see it absolutely flourish as it is a much more high-level application framework that fits LLMs very well. Less code. More clarity of intent. Easier to review, fewer lines of code to screw up. And fundamentally built for tooling, introspection and removing tedious repetition. It just fits with what a lot of LLM-using folks need. Ash always had a strong opinion. And that makes it stand out in the noise.

My hope is that the hardline anti-LLM folks find their ways to work on the meticulous bits we all build on. Hopefully not just as the artisanal coders of yesteryear. And who knows what the balance looks like when the token costs become real. I think there are many adjustments coming.

The path to distinguish yourself and your work is going to get weirder, wilder and if you want to carve out a space in people's minds you need to earn the trust, not be another proxy for Claude. Express yourself in both medium and message. We're in for an abundance we never wanted. As always, the challenge is entropy.

Curious to hear what note this strikes in you.
Thanks for reading. I appreciate you spending some attention here.

-Lars