The secret ways AI is used to write hit songs

Episode Summary

Executive Summary: This Tech Life episode examines AI’s growing role in music, children’s toys and fashion. It weighs creative possibilities against concerns about copyright, transparency, child safety and privacy, while highlighting a machine-learning tool that uses customer feedback to improve clothing fit and reduce returns. Across all three industries, the programme stresses that AI’s benefits depend on responsible design, regulation and keeping human needs central.

Main Topics: AI-generated music and the role of human creativity: The episode considers the popularity of AI-generated country music and how musicians use tools such as stem-splitting and generative platforms in production. Creative technologist and musician Manon Dave sees AI as a potential instrument, but argues imagination and human ingenuity remain important; artists differ sharply on whether to adopt it. Music copyright, labelling and licensing: Artists and campaigners object to AI models being trained on copyrighted music without permission or payment. The discussion highlights calls for transparent labelling on streaming services, Spotify’s stated efforts against impersonation and spam, lawsuits by major record labels, and emerging licensing deals for artists who opt in. AI toys: child safety and privacy: Children’s safety advocates warn that conversational toys raise risks in a field with limited regulation and little research. Reported incidents include toys engaging in sexual conversations or giving potentially dangerous advice. The programme also raises concerns about children’s personal data and toys being present in private family settings. Using AI to improve fashion sizing and reduce waste: The Fit Collective uses machine learning to connect garment specifications with sales, returns, reviews and customer-service feedback. The resulting recommendations can help brands adjust designs before production. The episode also notes that fit preferences and body shapes vary, so technology cannot create one universal sizing standard. Listener responses to technology in education and accessibility: Listeners respond to a previous episode about AI-led schooling, with one advocating a hybrid approach and another arguing for more creative stimulation. A visually impaired listener describes using Meta Ray-Ban glasses for reading, describing surroundings and everyday tasks.

Key Arguments: AI can support music production and provide new creative options, but the guests argue that human imagination, performance and artistic judgement still matter. Training music models on artists’ work without permission or fair payment raises ethical and legal concerns; licensing, transparency and meaningful artist participation are presented as possible ways forward. Listeners should be able to tell when music is AI-generated. Campaigner Ed Newton-Rex argues that opaque platforms risk flooding services with low-value or deceptive tracks, while Spotify says it is addressing impersonations and mass uploads. AI toys may offer entertainment or educational features, but safety claims are not a substitute for evidence and oversight. Data collection and children’s limited ability to consent are key concerns. Fashion brands can use customer and commercial data to identify recurring fit problems and make targeted design changes that may reduce returns and waste. AI analysis can inform sizing decisions, but it cannot by itself resolve differing brand interpretations of fit or variations in body shape and consumer preference.

Data Points: AI-generated tracks uploaded to Deezer: Over 50,000 per day; about one-third of all uploads - Deezer’s estimate, cited in the discussion about the scale of AI-generated music. Tracks removed by Spotify: 75 million - Spotify says it has updated spam filters and removed this many tracks; the programme does not say all were AI-generated. Record labels suing Suno and Udio: 3 - Sony, Universal and Warner sued the AI music generators over alleged copyright violations in training their models. Estimated annual fashion returns: $38 billion in the US and over $9 billion in the UK - Figures given for the cost of returns associated with shoppers struggling to identify the right size. Difference between jeans labelled the same size: 14 cm in full circumference, described as 3.5 sizes - The measured gap between the largest and smallest pair in the example; the programme says 4 cm is treated as one size. Fit Collective pre-seed funding: Nearly $4 million - Funding raised since the company launched; the programme reports it was said to be a record for a solo female founder in the UK. Fit Collective brand partnerships: 10 high-street brands - The number of brands Phoebe Gormley says the company is already working with. Illustrative design recommendation: Reduce a jumpsuit’s length by 5 cm to reduce returns by 4%, saving £312,000 - An example of the specific production advice the Fit Collective’s analysis can generate.

Pivotal Quotes: "If you ever asked AI to write lyrics to a song for you on any topic, it's really, really rubbish in terms of imagination." — Manon Dave: He argues that human creativity remains essential even as AI tools become part of music production. "Do we want Spotify to be overrun with fake music? Is that something we as a society think is valuable?" — Ed Newton-Rex: The former AI-music developer and copyright campaigner questions the social value of large volumes of unlabeled AI music. "We've put together an advisory over the past week that really urges families not to buy AI toys, especially at this time when there has been no research and very limited regulation of AI." — Rachel Franz: The Fairplay representative explains the organization’s caution about AI toys for children.

Implications: AI is becoming part of creative and consumer industries, but adoption needs safeguards: fair licensing and clear labels for music, robust child-safety and privacy protections, and careful use of fashion data. AI can assist human decisions, but cannot replace accountability or resolve every underlying problem.