Chat GPT passes university exam in 20 minutes
Bristol University graduate Pieter Snepvengers has used the AI software, Chat GPT, to achieve a 2:2 in a 2000-word essay.
Chat GPT, the three-month old AI chatbot has enjoyed a problematic infancy, and has caused
(another) stir recently after it was proven to be able to produce degree-level essays that
pass inspection.
In a recent experiment, ex-student turned writer, Pieter Snepvengers, used the software to
generate a response to a social policy assignment. It wasn’t quite as simple as providing the
bot with the question and awaiting response – Pieter had to ask ten related questions and
stitch together a hybrid from the answers he received. Even so, it took less than twenty
minutes to create a 2:2 level submission.
The lecturer who agreed to mark it, whilst admitting that the surface-level piece was slightly
“fishy”, decided that it could pass as the work of a “lazy” student. In news that will worry
university faculties across the country, the lecturer also determined that if Pieter had simply
added references or made a few tiny adjustments, the submission “might even have hit the
high 50s”.
The lecturer who agreed to mark it, whilst admitting that the surface-level piece was slightly “fishy”, decided that it could pass as the work of a “lazy” student
Given that the currently free-to-use bot learns and evolves from every interaction, it is
possible that within a matter of months, the software, or ones similar may be able to
produce 2:1s or even Firsts, undermining the entire higher education grading system.
Before students begin the celebrations, it is worth noting that similar concerns in the US have led to blanket bans on AI use in education, and whilst UK universities are yet to put out a concerted response, it is likely that experiments like Pieter’s will apply the pressure needed for a crackdown.
