Will AI mean better adverts or 'creepy slop'? (www.bbc.co.uk)

Brands and ad agencies are already using large language models to generate hyper-personalised advertising that mimics an individual’s tone, phrasing, preferred music and colour palettes. Firms such as Cheil UK and startup Spotlight combine LLMs with signals from public social media, search histories and even what people enter into ChatGPT to infer personality traits (introvert/extrovert, mood, taste) and overlay those insights on traditional demographic data. The tech can produce millions of unique creative variants at scale, and a US study found ChatGPT-written copy tailored to personality traits was more persuasive than generic text — promising to reduce the roughly 15% of digital ad spend that’s currently wasted. That capability matters because it shifts targeting from broad segments to one-to-one psychological profiling, creating more relevant ads but raising data‑privacy, ethical and regulatory alarms. Critics warn this will fragment creative memory into “one-person” ads, feel intrusive or be abused for political persuasion; proponents argue it can deepen relevance if used ethically. Key implications for the AI/ML community include improving model fidelity for voice/persona matching, auditability of data sources and decisions, robustness against manipulation, and building guardrails (privacy-preserving training, transparency and regulation) to prevent exploitative personalization.