LR Friberg
The idea is simple. NotebookLM offers a service called Audio Overviews, which transforms various types of documents, slides, charts, or websites into interactive discussions. Recently, they introduced a feature that allows users to customize these Audio Overviews by giving specific instructions to the AI hosts on what to emphasize. The AI technology driving this service is surprisingly empathetic. While it may not always get the facts spot-on, it does create an impression that "they really understand you".
You might be thinking, "What kind of empathy can you expect from ChatGPT or your ordinary kitchen appliance? That's just absurd". But let's consider this: experimental underground artists often struggle to get their work noticed and appreciated. When was the last time you did more for these artists than just liking their posts on social media or donating a few dollars on Bandcamp, probably without listening? While there are certainly plenty of artists working hard to create platforms for experimental and underground music communities, there are many others on the fringes who struggle to get any attention at all, let alone a chance for people to really explore and appreciate their work. And, even though it's not 2022 anymore and we're talking about seemingly ordinary AI technology, these systems can really engage with the content they receive. They provide feedback in a way that's surprisingly empathetic and understanding, something that people often overlook or don't take the time to offer. For many artists who can easily get overlooked in the vast sea of music available today, even a small acknowledgment from AI can have a significant impact. That's the purpose of the project Electronic Empathy.
Every track on the album features NotebookLM-powered discussions about the work of various underground musicians, with their actual music mixed in. It serves as a tribute to these artists and could be a way for newcomers to discover these artists' work. The selected musicians have collaborated with blockchain-based netlabels Aurora Compilations and Motivational Music, as well as their predecessors like INTENT compilation series, dating back to before the pandemic.
It's worth considering that robots are ultimately created by humans, and there's often an awkward but well-meaning data scientist who decides to make AI models more empathetic. But, do we really need to rely on machines for empathy? Maybe we should focus on valuing each other's creative contributions while we're still here, creating art and music? Sure, we all have our own commitments and limited energy to explore new things, and let's face it — attention spans are shorter than ever with so much competition for our focus. In an ideal world, projects like Electronic Empathy wouldn't be necessary. But, until we reach that level of perfection, if you're interested in exploring something like this, feel free to let AI dive into your work — it's more accessible than ever, just like all the music out there.
Finally, some further thoughts. For those who know me well, you might be wondering how I reconcile my concerns about sharing personal data with big tech and social media with a project like this. The truth is, while controlling our personal data is extremely important, art and music are meant to be shared openly and willingly. The reality is that large language models (LLMs) are here to stay, and with the overwhelming amount of content online we should take the opportunity to represent ourselves on our own terms. If someone asks an LLM about a particular artist, it can highlight that person's contributions to underground music and encourage others to explore their work. This is a small but meaningful way we can make our voices heard as LLMs continue to evolve. Then, you might be curious why I decided to host this project outside of the permaweb. The truth is that the focus of creators can evolve over time, and these AI-generated discussions may not always reflect reality. This is a work in progress, and the aim is to experiment with these technologies to gather feedback on your work, especially since they are widely accessible and free at the moment. Finally, if you’re wondering how to generate these discussions, there are plenty of tutorials available online, so I don’t think it’s necessary to create another one.
Here's the music that was used in this project. If you're interested, feel free to look up the artists online and check out their work.
{AN} EeL - Future Conversations [? / panpanpanaviandistresscall / 2024]
{AN} EeL - Banned Ear [Red Flags / panpanpanaviandistresscall / 2024]
Brainquake - Slowed By Gravity [Excess Denied / Mahorka / 2021]
Chris DeBry (MedMera Foundation) - Episode 110 [MedMera / 2020]
D-Fried - Over The House-Tops [Peak Shift Effect / Clean Error Records / 2014]
D-Fried - Peak Shift Effect [Peak Shift Effect / Clean Error Records / 2014]
Deborah Fialkiewicz - Dazzle [Daughters Of The Industrialists / z0 / 2023]
Deborah Fialkiewicz - Orange [Daughters Of The Industrialists / z0 / 2023]
Hidden Glow - Fimiam, Part 1 [Fimiam / hvedrungrsmil / 2018]
Ivo Petrov (Skylined) - Somebody Did All That [Music For Elevators Vol. 4 / Mahorka / 2012]
Jon Shuemaker - Time To Start Living [Time To Start Living / self-released / 2024]
Jon Shuemaker - Forward Thinking [Forward Thinking / self-released / 2024]
Linn Friberg - Illuminate (The Path) [Aurora 2024: Luminous Path / Aurora Compilations / 2024]
Linn Friberg - Rise From The Shadow [Aurora 2024: Luminous Path / Aurora Compilations / 2024]
Linn Friberg - Higher Standards (Original Version) [Higher Standards / Motivational Music / 2024]
MUWN - S K Is In The Shadow [Dream 7 / self-released / 2021]
Pete Swinton - D2 Two [Drifting Vol. 2 / self-released / 2024]
Slavek Kwi (alfa00) - 23:13 kwa-crickets-WSi [Ho'oponopono(02) / self-released / 2024]
Thomas Park - Concrete Expanse [Concrete Expanse, Submerged Version / self-released / 2020]
Wayne DeFehr - Transmission [Ode To Life / Attenuation Circuit / 2024]
Wilfried Hanrath - Starseed [Starseed / self-released / 2024]
You can find the album on the Bromtol Largesse netlabel.
