The funny and sinister sides of machine learning, how to make nanoparticles by the tonne - Physics World Weekly Podcast


Manage episode 269541692 series 2352990
Creato da Physics World, autore scoperto da Player FM e dalla nostra community - Il copyright è detenuto dall'editore, non da Player FM, e l'audio viene riprodotto direttamente dal suo server. Clicca sul pulsante Iscriviti per rimanere aggiornato su Player FM, o incolla l'URL del feed in un altra app per i podcast.

Would you use a pick-up line composed by a machine-learning algorithm? Perhaps something like You Look Like a Thing and I Love You, which is a machine’s attempt at an amorous icebreaker and the title of a book by Janelle Shane.

Shane is an optical engineer with a blog that chronicles the absurdities that are sometimes generated by artificial intelligence (AI) systems – and is interviewed in this episode by Physics World’s Margaret Harris. While pick-up lines and recipes generated by AI can be hilarious, Shane cautions that there is also a sinister side to AI because it can become very good at mimicking human prejudices.

Have you ever wondered how large quantities of nanoparticles can be manufactured to very high precision? Ed Lester, founder of UK-based Promethean Particles, explains how the firm’s continuous-flow hydrothermal synthesis process does the job using a much more sophisticated version of a method familiar to anyone who grew a “crystal garden” in science class.

Lester spun the company out of the University of Nottingham, where he developed the technology. He looks back at the challenges of developing and commercializing the process and talks about some exciting new nanoparticle applications such as anti-icing coatings for aeroplanes and artificial bone for medical applications.

124 episodi