con bonus alla fine: la risposta alla seconda del per niente permaloso Sergei Guriev, MS, PhD.
Da: Giacomo DA ROS
Data: 10 apr 2023, 18:41 +0200
A: Sergei GURIEV
Cc: MATHIAS VICHERAT
Oggetto: Re: on ChatGPT
Dear Mr Guriev,
I sincerely thank you for your answer. I had not seen your recent comments and I am glad to see that your perspective is much more nuanced and informed compared to the initial communication by Sciences Po.
I was especially delighted to read "we encourage the use of ChatGPT”. This contrasts with the very negative messaging that accompanied Sciences Po’s decision, which I would say is best embodied by this comment, but is also the defining trait of your original email. This attitude is at the core of why I believe this policy prevents an honest discussion of such tools. The very first point the director makes in his comments is about factual inaccuracy: why, then, would anyone admit to using such tools? One might fear that his/her work might be considered less serious or be subjected to additional scrutiny.
This is, fundamentally, what I was referring to when using the word “ban”. The policy that was laid out in the original email is deeply flawed as it is impossible to verify whether a student has, for instance, used AI tools to research a paper. This makes it completely risk-free to not reference the tool. Conversely, as previously noted, one might imagine potential downsides to actually doing so. Logically, this should result in very few students abiding by the policy. This is unfortunate from an ethical standpoint first. Secondly, it also ensures no discussion can be had around the actual strengths and weaknesses of such tools, or how to integrate them in research work. This amounts to a de facto ban on meaningful reflection around the technology and its implications.
I have taken the liberty to attach to this message two documents related to TU Delft’s policy on the matter. This is a case I know reasonably well for personal reasons, and it seems to be consistent with other constructive takes from academic institutions, such as this one. While - as noted in slide 19 - they also mention possible academic integrity issues, their approach differs from Sciences Po's in three key ways.
- TU Delft took the care to reflect on the topic and provide (through its professors, in this case) the students with an accessible explanation of its workings.
- It has developed an effective method for students to transparently detail their research process - as is also testified by the other attached document.
- Most importantly, the implicit message conveyed by the document is profoundly different. Their overall message is positive, demonstrating an understanding of AI tools, expressing optimism for their use, encouraging students to utilise them, and conveying institutional curiosity about students' achievements with these tools.
I thank you again for your answer and wish you a pleasant week. Kind regards,
Giacomo Da Ros
------
Da: Sergei GURIEV
Data: 7 apr 2023, 18:38 +0200
A: Giacomo DA ROS
Cc: MATHIAS VICHERAT
Oggetto: Re: on ChatGPT
Dear Mr. Da Ros,
I have indeed not responded to your email. I assumed that you would subsequently reread my message and figure out that Sciences Po has not “banned ChatGPT". We have just asked students not to use ChatGPT without transparent reference. This is fully consistent with our anti-plagiarism policy. If you use ChatGPT and acknowledge it, it is perfectly allowed. Here is a detailed discussion of my thoughts on this issue: link
On a personal note, I would like to add that I have a summa cum laude master's degree in computer science and a PhD in math, I have extensive experience in mathematical modeling and coding and have used text analysis and machine learning in my recent research. Actually, in my speech in Reims Campus of Sciences in August 2022, I did use AI (in this case, DALL-E) myself - but of course fully acknowledged it. link
I take this opportunity to wish you a good weekend.
Sincerely,
Sergei Guriev
Provost
Sciences Po, Paris