Blog |

Platform Updates - Commenting Redesign & AI Image Moderation

Our development team have been working on the latest updates to LEGO Ideas and we're happy to share 2 new things with you all. 


Commenting has received a makeover

For a long time, commenting on LEGO Ideas has not been entirely ideal in terms of following along in a conversation and other aspects. That changes today, as we introduce a revamped commenting module. 

So what's new to comments?

  • You can now easily see what and who someone has replied to. This makes it much easier to follow a particular conversation or sub-conversation.
  • Comments can now more easily spin-off into manageable sub-conversations and even sub-sub-conversations. If too many sub-conversations are created then it will spin off the thread to its own separate thread (although you can always easily go back to the main conversation). 
  • The 'Like' button has been replaced by an 'Upvote' button, but it works pretty much the same.
  • We've removed the option to edit the formatting of text comments (for instance, bold, italics, etc). This is part of our overall visual redesign and desire to create a cleaner, minimalistic look and feel. 



AI Image Moderation - BETA

This sounds like something straight out of a sci-fi film, but don't worry Skynet won't take over LEGO Ideas anytime soon (we hope). We're still lots of human beings running things behind it all. 

One of the most common pieces of feedback we give to members who upload submissions is that their photos need to be re-taken as they are too blurry, grainy, dark or otherwise difficult to see. For this reason, we are implementing a test of AI (Artificial Intelligence) driven software that will be able to detect particularly low quality photos and automatically send the member feedback if there is a problem with their images.

Why are we doing this?

  • It will help reduce our own workload a bit so that we can focus on bringing value to you all in other ways that AI/automation can't. 
  • It will help provide quicker feedback to project submitters on one particular criterion (low-quality images) so that they can make changes quickly and resubmit (they may still need to make other changes after a human moderator has reviewed their submission).

How will this work?

For now, this will work in a way that you go through the regular submission flow and submit your creation. The system will then crunch the data behind the scenes and if the system determines that your images are blurry, then you will receive an email within approximately 30 minutes informing you of this.

If you don't receive an email immediately, that means your submission has passed through to our regular moderation queue where one of our moderators will take a look at your submission including reviewing it for the many other criteria that we look at before approving submissions.

The AI's algorithm has been trained with images we've disallowed in the past and will continually be optimised to improve itself. 

We look forward to testing this new system out and to hear your responses to it as well.

  • development update

Opens in a new window