A flawless essay. Working code. Neither proves a student has learned anything.
That is the problem generative AI has created for every university. In an interview with BW Education, Dr Praphul Chandra, Dean - Research and Director of the Atria Center of Excellence in AI & Decentralised Technologies, shares Atria’s answer: stop grading the product, start grading the thinking.
Don’t ban AI, be deliberate
AU’s approach includes:
Pen-and-paper tasks done in class, supervised by the professor
AI-allowed assignments where students defend their work in an oral viva
AI-prepared handwritten cheat sheets that make open-book exams real preparation
Submitted AI conversations so faculty can see how students questioned the tool
AI detectors are too unreliable to anchor integrity. Instead, Atria designs work where reasoning is visible, with disclosure of what tool was used, for what, and how it was verified. Accountability stays with the human author.
LEAP-AI: literacy beyond prompting
First-year students across all majors study critical thinking, algorithmic thinking and critiquing data. AI sessions are followed by faculty-led debates where students challenge responses and question hidden assumptions. Ethics isn’t an elective. It sticks when students have to defend their choices.
Skills that outlast the tools
With code increasingly AI-generated, graduates need to:
Decide which problems are worth solving
Judge answers and know when not to trust a machine
Learn new fields fast (what you study at 19 may be outdated at 25)
Collaborate, explain and persuade
Ask “why” even where questioning isn’t welcomed
The takeaway
Integrity in the AI era is less about policing and more about building the judgement to use these tools well.
Read the full interview: Grade Reasoning, Not Output on BW Education
