Wordle, the sensational online word-game, has endeared the internet — and new takes on the simple format have cropped up for just about anything you can imagine. Now, there’s 3DPrintle: a Wordle-inspired minigame that challenges players to identify 3D print failures against an AI.

3DPrintle’s creator, 3DQue, is behind Quinly, a 3D printer automation system that uses the company’s proprietary software and VAAPR print bed to enable autonomous print removal and continuous printing. Right now, the company is working on a new cloud-based failure detection AI it calls QuinlyVision to detect 14 different print failures and, if possible, address them remotely; which spawned this new 3D printing quiz game.

“We were training the QuinlyVision AI internally and thought ‘hey, this is pretty fun!'” said Steph Sharp, co-founder and CEO of 3DQue. With the recent popularity of Wordle in mind, 3DQue’s engineers set on an ambitious side project to make their AI training program available to the public.

Though the game is rooted in AI training, 3DQue isn’t forcing users to give up their data to play. Players are free to choose between participating as a guest, where no data is saved, or as a challenger, where player input helps to train the AI. The company sees value in its game whether or not the results help with machine learning.

“The difference between an experienced operator and a newbie is the amount of failures they’ve seen,” Sharp said when discussing the value of 3Dprintle to those new to 3D printing. In 3DQue’s recent live stream about QuinlyVision Steven McCulloch, 3DQue’s Quinly Project Lead, said “I probably saw my first 100 fails in my first year and then saw my next 100 fails in my next four years.” 3DQue hopes the ability to see print failures in a gamified format will help new operators get up to speed fast to help save time and material.

To further the educational value of its new game, 3DQue has produced a number of videos featuring McCulloch that players can find on the game’s welcome page. The videos teach viewers about these failures: what they are, how they’re detected by AI, why they happen, and how to prevent them.

Gamification of AI training has been used for some time and has made its way into massively popular online games. EVE Online, a popular sandbox space MMO, and its Project Discovery minigame is one such example where players perform a real scientific practice — flow cytometry — to handle real-world scientific work. In the minigame’s first iteration, which aided The Human Protein Atlas, players helped train AI with their results — just like 3Dprintle. Successive versions of Project Discovery have continued to help with machine learning, but results are used at the receiving scientists’ discretion.

In 3DPrintle, challengers who have proven themselves skilled will have the opportunity to participate in training the QuinlyVision AI at a deeper level. Sharp says that experienced users will be able to submit their own images and videos to 3DPrintle and participate in more challenging quizzes.

Ultimately, the goal with QuinlyVision is to prevent material waste and save production time. Identified failures can be removed from the print bed and parameters can be set to attempt resolving production-halting problems (like nozzle blobs) remotely, reprint failed models a set number of times, or just move on in the queue. QuinlyVision’s overarching goals go far beyond that, with plans for root cause identification (model, G-code, or machine) and what Sharp called “predictive maintenance”.

3DQue hopes to roll out QuinlyVision with the ability to detect 14 print failures in May 2023 and to have its more ambitious features ready to launch by 2024. Though, Sharp says they’ll roll out features as they become reliable and expects what 3DQue refers to internally as “nozzle saver” detection (blobs, under extrusion, and no extrusion) to go live first — but beta testers will have access this year.

Until then, you can help get QuinlyVision ready faster and have some fun while you’re at it by playing 3DPrintle!