In partnership with
Dear Sentinels
Artificial intelligence is no longer some far-off promise from a sci-fi novel; it is already busy reshaping how we work, learn, and organise society. Whether it is automating the boring bits of our jobs or making decisions about who gets hired, a loan, or even healthcare, AI is quietly redrawing the lines of opportunity and inequality. Some people see a future filled with productivity and prosperity for all, while others are bracing themselves for mass job losses, wealth piling up in a few hands, and ever-widening gaps between those who build the tech and those who have to live with it. Getting to grips with these socioeconomic twists and turns is not just something for academics to debate over coffee; it is essential for anyone trying to make sense of the world we are cobbling together, one algorithm at a time. But the ultimate question is, who is lying about it, especially with the layoffs being ascribed to AI, and who is telling the truth?
But before we get too carried away, let us see what the internet has for us this week.
News from around The Web
Is Your Training Data Actually Model-Ready?
If you're fine-tuning a speech model, you've probably hit this wall: DNSMOS gives you a score, but it doesn't tell you whether the data behind that score is actually right for your model.
Treat it as a pass/fail gate and you'll end up training on audio that looks clean on paper but drags down real-world performance—while good source data gets tossed for no reason.
Voices' CTO DJ Jalali (with the team's senior audio and voice data engineers) just published a free white paper that breaks down the four-step calibration framework they use internally to set model-specific quality thresholds instead of trusting the raw DNSMOS number. It also covers where DNSMOS breaks down and how Voices validates audio for custom datasets at scale.
