CEO & Co-founder, MusicalAI. Published author, data & analytics (O'Reilly). Indigenous (Anishinaabe / Eagle clan).
As my co-founder Matt said - "There, we fixed it! Generative AI for Music that is. Now with transparent, auditable Attribution right down to the writer and performer level from the training data." We think it's insane that companies are training on art that they don't own the rights to. And we think there is no industry / vertical that understands this more than music. And we've decided to do something about it. Here's how: - Our customers train their own base models on the music they own the rights to. They have full visibility on what is in their model. - For each generative outputs made through our system, an attribution report is made that accurately identifies which training model objects (ie - a particular song) influenced the generative output, and by how much. The lie that you cannot trace back the origin of anything in Gen.AI has to stop. We must acknowledge the human creative DNA that the industry is using to train their AI models with. And we're going to do just that, by example. We're so proud to introduce https://www.somms.ai. This is what I've been up to with Matthew Adell, Nicolas Gonzalez Thomas and James Maxwell for the last year+. Proud to be building this partly in the east coast. Proud to be building this mostly in Canada (with Matt in Los Angeles, the epicenter of music). I'm so proud of the work we're doing with SOMMS.AI. #MusicAI #Innovation #MusicAttribution #AI #Tech #MusicIndustry
This is so cool! There must be so many connections between this work and the work you do through ICCE, in terms of using AI as a tool but also considerations around IP. Really interesting!
Vik Chaudhary
12mo
Congrats Sean Power! Were your music attribution here in 2012, Taylor Swift wouldn’t have had to re-record her music because of ownership disputes! 🎉
Tim Estes
11mo
You need to come to Nashville with this. Lots of interest and certainly the other Hub in the US next to LA.
Bryan Watson
1y
💯 💯 💯- “The lie that you cannot trace back the origin of anything in Gen.AI has to stop. We must acknowledge the human creative DNA that the industry is using to train their AI models with.”
Sean Power this is fantastic. I can't wait to learn more about what you're doing. I just came from Ireland Music Week (industry conference & showcase) and the topic of AI & IP rights came up multiple times. As an advocate of getting songwriters, producers, and musicians paid, I'd love to be involved in what you're doing. I'd love a demo and a chat.
This IP consideration is one of the core reasons why Pinecone and vector databases are so crucial to the future of AI. Leaning on RAG rather than full 'black box' AI generation w/ ambiguous training data lets you build a solution that can withstand a lot more scrutiny on the sourcing side.
Georges Lariviere
12mo
Very impressive and much needed Sean !
That's awesome Sean Power ! SOMMSing tells me that this is going to be a killer app 💪💪
Agentic Engineer / aiCTO / Consultant
1yThis is really interesting. Would love a demo