Director of Product
Roisi is Director of Product at Altmetric, creating products and services that help the academic research community understand the ecosystem of attention around their work. Prior to this Roisi has worked across domains as varied as the film industry, e-commerce, adtech and gaming. Throughout her career, Roisi has prioritised understanding both the benefits of what she is building, but also the risks the work presents - she believes that understanding how bias can manifest in unexpected ways is a crucial facet of responsible product development.
Debunking the "magic" of algorithms - facing up to the realities of modern data science
As Arthur C Clarke once famously wrote - “Any sufficiently advanced technology is indistinguishable from magic”. From an outside perspective, AI, ML and Data Science has reached this point already, with algorithms ready and waiting to decide what we buy, who we socialise with and what our interests are - doing so in a pristine environment free from the pressures and complexities of the average human existence. Problem is, it’s more marketing than magic. The products and services that underpin so much of modern life have inherited the bias of their creators, and unless you have either a relentless stubborn streak or a PHd they are mostly impenetrable. Because it’s so daunting to consider the risks, we’ve pulled back from doing so, instead choosing to “focus on the good”, often at the expense of the people who need the most help. So join me in walking through some of the traps we’ve already fallen into, and some that we still have the opportunity to push back against - including some practical tips to reframe product decision making in away that allows you to still focus on the good, while acknowledging and mitigating against the risks of pushing computer-based answers to human problems.