Fien De Waele Tongeren, Belgium 1H AGO

I use ChatGPT to help clean up computers that have accumulated multiple backups, copied folders, and files that may or may not be duplicates. The main problem is uncertainty. When several backup generations exist, it can be difficult to tell which files are genuinely unique, which are copies, and what can safely be removed. Instead of deleting anything immediately, I first generate a local folder tree showing the computer’s structure and filenames. I give that overview to ChatGPT, which analyzes it and creates a shortlist of folders and files that look like likely duplicates, redundant backups, or otherwise worth checking. From there, I use a small script to extract the relevant filenames into an Excel file. This turns a messy filesystem into a much easier review list. I then manually search for those candidates in Windows Explorer and make the final decision myself. In one cleanup session, this reduced the problem to around 150 files that I could inspect manually. The entire process—including generating the folder structure, analyzing likely duplicates, reviewing the shortlist, and deleting confirmed redundant files—took about two hours. By the end, the drive was roughly half as full. The useful part is the division of labor: ChatGPT handles pattern spotting, comparison, and shortlist creation, while I retain control over anything destructive. That makes large-scale computer cleanup much faster without turning it into a risky “delete duplicates automatically” exercise. Step-by-step: 1. I generate a local folder tree showing the computer’s folder structure and filenames. 2. I give the overview to ChatGPT and ask it to identify likely duplicates, redundant backups, and other files worth checking. 3. I use a small script to extract the relevant filenames into an Excel file. 4. I manually search for the shortlisted candidates in Windows Explorer. 5. I decide which files can be removed and delete only the confirmed redundant files. 6. I retain human control over every destructive action rather than deleting duplicates automatically.