The Leaky Talent Pipeline In Academia

The loss of PhD students to industry has reached something of a fever pitch in the past year, especially in high demand areas such as machine learning.  It’s created a high degree of concern about the health of the talent pipeline in academia.

A recent study from Cornell set out to explore the issue in more depth and assess some of the reasons why people are leaving the sector.  The authors tracked several hundred doctoral students across science and engineering disciplines at 38 different American research universities.  They were quizzed throughout their graduate training to explore any changes in their career preferences during this time.

It emerged that 80% of students set out on their PhD with the intention of going on to a career in academia, but this plummets to around 50% by the time they graduate.  This certainly wasn’t down to any lack of confidence in finding work or in obtaining grants for projects, with few highlighting this as a possible reason.

Whilst the vast majority of doctoral students have already undertaken research in their undergraduate careers, they are still often unsure of exactly what a career in the field would entail.  It’s when they gain this greater insight into a career in academia that a divergence begins to emerge.

The authors believe that it would be valuable to provide students with more information on possible career paths, and even to help prepare them for lives outside of academia.

They cite the BEST (Broadening Experiences in Scientific Training) program at Cornell, which aims to provide doctoral students with training in careers outside of academia.

Whilst most of the emphasis of the work was on how this might affect the academic world, they also believe it has important insights for industries looking to recruit PhD students.  They suggest that rather than trying to emulate the academic environment, it’s important to understand that many students are looking to break from this, and instead want to focus on commercializing new technologies.