Discussions about whether we should use Artifical Intelligence helpers for creative work remind me of musician's unions attempting to ban synthesizers and drum machines, back in the early eighties, as told by the articles linked at the end of this post.

Such a ban sounds funny today, we are way past that fortunately.

In the meantime, musicians have learnt to combine these tools with their own skills in fantastic ways, covering the full range from "all electronics" to "no electronics" music, and everything in between, for our enjoyment and at the service of their creativity.

We've come a long way, and it has taken time.

I suppose our evolution regarding AI tools will be similar. We will learn where those make sense, appreciate situations where we are fully separated from them, and use common sense and our increasing experience to decide what the best tool for the job is.

Being a luddite is not my style, but I also want to embrace progress in a reasonable and durable way. There will be mistakes along the way, but my optimistic self is confident that we will collectively learn, as we do with any tool.

I'm looking forward to re-reading this post in a few years and seeing how our use of generated content has evolved.

Links to articles

Here are links to articles about the UK Musicians Union synths & drum machines ban from the early eighties:

• Music Radar: The day the 'Loony' Musician's Union tried to kill the synthesizer (which also happened to be Bob Moog’s birthday)
• Mixdown Magazine: The UK Musicians Union tried to ban synths 40 years ago today
• Far Out Magazine: How Barry Manilow caused the music industry to try and ban synths

---

P.S. Oh - this post uses an AI-generated image !

Yes indeed. It feels a bit like cheating, exactly like using a drum machine or synth does. Which I regularly do, at the service of a "larger picture" project.

Our intern Antoine made me notice the weird arrangement of black & white keys on the keyboard in the above image. You won't be able to unsee it now, and I guess such artifacts are going to be commonplace if we make use of more machine generated content. That's fine with me, depending on the budget allocated to a specific piece of content and as long as that content does its job.

The main point where I think we need progress is in the energy consumption of AI services. I was happy to read about 1.58 bits Large Language Models recently, which perform well for a fraction of the energy cost, and I'm confident we'll progress over time. We need to take our engineers ethics seriously and keep the pressure so that such progress happens.