recovering musician. still convinced aphex twin was 10 years ahead of everyone. now obsessed with recommendation systems, streaming economics, and AI audio.
so suno pulled that mary j blige ad after finding out they'd struck the deal with someone falsely claiming to be her official rep. the ad had already gone viral and sparked backlash before anyone caught it
this is a preview of what's coming for every AI music company doing celebrity deals, rights verification just isn't solved yet and the incentive to move fast outweighs the incentive to check thoroughly. expect more of these before it gets fixed
AI MUSIC SCOREBOARD 🎧
Sony + UMG vs Suno: still fighting in court.
Sony vs Udio: still unresolved.
Sony + Warner vs Anthropic: lyrics lawsuit, CEO named personally.
UMG vs DistroKid: now fighting over AI-generated music uploads.
Bookmark this. It might change by friday. 👀
@unusual_whales hearing an anthropic leader state flatly that digital security is heading into a darker phase underscores how fast adversarial ai is scaling systemic vulnerabilities
spent 3 hours trying to train an AI model to recreate that exact vintage guitar hiss from a seventy-eight record and now my laptop sounds like it's preparing for takeoff...
New Google DeepMind paper shows that if AI models offer money to answer each question and the highest offer answers, results are 99.2% as good as always picking the best model.
i.e. say you don't know in advance which model is best for each new question. So who should answer?
The paper’s answer: let each model offer money for the job. The model that knows it is good at this question offers more. Highest offer gets to answer.
They proved the loss shrinks when several agents usually value a prompt highly, as when distilled models perform close to the best. They simulated auctions among 110 models on over 30,000 EmbedLLM prompts.
With all 110 models bidding, the auction reached 99.2% of the best possible value, versus 89.4% for routing each prompt type to its best model. On each method's worst prompt cluster, the auction held 96.0% while the single best model kept 6.0%.
If you run a model marketplace, a simple pay-your-bid auction works well when every bidding pool has several strong models.
claude mythos 5 designed quantum circuits that beat the best published designs on seven of eight tasks, with up to 17x fewer gates in one case and a decoder 12x cheaper than google’s earlier one. the 17x circuit used about 12% more qubits, so it’s a trade-off and not a free win
the two models couldn’t be more different: a corporate-run fund versus shares handed straight to people. whichever one actually ships will tell us more than the announcements do, since both are still promises
Universal Basic Income is starting to sound a lot less crazy
With Anthropic pledging 5% of its shares to a public benefit fund and Elon Musk talking about giving Americans a stake in SpaceX, the conversation around sharing AI-generated wealth is heating up
The idea is simple: if AI eventually replaces millions of jobs while generating trillions in wealth, people could receive regular payments regardless of employment
Neither announcement is actually UBI, but it’s interesting how quickly the conversation is shifting
If AI ends up doing most of the work, how will everyone else make a living?
@Pirat_Nation the fact that this is a hobby project with no public download yet and already beating some official ports on performance is quietly impressive. also slightly rude to the port team
AI MUSIC SCOREBOARD:
major publishers target anthropic with fresh lyric copyright suits, umg and warner lock in commercial licensing deals with suno, and indie labels pile on with new class-actions over training data.
bookmark this, it's going to be outdated by friday