GLM has been the actuarial standard for technical rating models of retail GI business since the 1990s. But that was 30 years ago – computing power has exploded since then, data science techniques have become available that can significantly outperform GLM, and the data itself is becoming richer and more complex. Isn’t it time for the profession to embrace the 21st century and move on from GLM? Or would we just be locking ourselves into a machine learning “black box”? Is the loss of transparency and agency over our models too high a price to pay? Hear from a panel of expert practitioners, who will explore both sides of the argument – and the space in between – to plot the way forward for rating in the 2020s and beyond.