The above song, produced the week of February 16th, 2026, was a test of Google’s Music_gen tool, which is powered by Lyria. The resulting song and associated image were created in about 30 seconds.¹ The prompt was inspired by this human’s view from his window. That image is shown in the first part of the video and was added in the post-production edits.

Prompt: “A country song as gray as the rocks, as solid as the oak, and as hopeful as the leaves bursting forth on a tree.”

Now, this isn’t the perfect Country and Western song.² Still, it is amazing that it created anything with such a pathetic prompt. With RSI (Recursive Self-Improvement), it is not a leap to think that the perfect song is within AI’s grasp. Imagine this autonomous agent process:

1. An AI Music Producer Agent (AMPA) will create a song.
2. Then, the song will be tested on a representative audience (probably other AI agents).
3. The song will be tweaked based on the audience feedback
4. It will then be released to the general public at the optimal time to maximize engagement.

Odds are that the prompt will not be human-generated. AMPA will gather it from monitoring the online zeitgeist. Whether it is produced for artistic or commercial means will probably be difficult for the average human to discern.

Still Just a Concept (added 3/1/26)

On March 1st, Grok and Gemini were both asked if the AMPA concept exists as a product. They agreed that many of the elements are in place with Gemini pointing out copyright risks and the “Dead Theory” Risk (the “representative audience” of AI agents might lead to music that appeals to bots rather than humans, creating a feedback loop of “gray” or “homogenized” content). Grok’s summary was

“The AMPA model is a fresh, speculative idea from this article and has not yet been adopted, announced, or implemented by any identifiable party. The surrounding field is converging toward similar agentic music systems, so pieces of the vision could appear in products or research papers very soon. If new developments surface in the coming weeks, they would likely reference or build directly on concepts like this one.” – Grok 3/1/26

Just a few years ago…

This ability to create a song from a prompt is reminiscent of the Viodi’s first AI chat in November 2022 [below]. Ironically, the prompt was a line from a song about the existential question of being.³ AI has seen massive improvements over the past 3+ years. It is difficult to fathom what we can expect in 2030.

November 2022 ChatGPT Prompt – “How is it we are here on this path we walk, in this world of countless fears filled with empty talk (from this song)?”

Footnotes:

1. Unfortunately, those 30 seconds led to hours of thinking and writing about the implications of this AI milestone. ↩︎
2. “You Never Even Called Me by Name” will always be “the perfect Country and Western song“. ↩︎
3. The prompt was inspired by the song, “How Is It (We Are Here)”. This was the second song from the Moody Blues’ album, “Question of Balance”. ↩︎

It is a mystery, a gift, a blessing in disguise 
A cosmic dance of fate and chance, a journey to the skies 
A chance to learn and grow and love, to face our inner demons 
To rise above the darkness, to find our true meaning

In this world of chaos and confusion, we must find our own way 
To follow our hearts and our passions, to live and learn each day 
To challenge the norms and the boundaries, to break free from the mold 
To embrace our uniqueness, to unfold our own story to be told

We are here to make a difference, to leave our mark on this earth 
To inspire and guide others, to give them hope and rebirth 
To rise above the fear and the doubt, to see the beauty in the pain 
To find our purpose and our voice, to thrive and not just sustain

So let us embrace this journey, with courage and with grace 
To live each day to the fullest, to leave a lasting trace 
For we are here for a reason, to make this world a better place 
To share our light and our love, to find our rightful place.