Reviewing AI/bot and low-value posting patterns

Our core forum stats can always be found at the bottom of the forums page. We run a deeper dive into the data behind those stats in our weekly reviews on the platform and we’ve adjusted that review to address concerns about AI, bots, and low-value content.

As with any engaged and popular forum, there are all sorts of accounts. We have those who are truly engaged, users who are just browsing before their next big project and of course those who are here to try and game the system.

We put significant resources into addressing the issue of community quality already. For example:

• Marketplace bump reduction – ref: Change log: behind the scenes at BHW (page 4)
• AI scanner – ref: Beta testing AI detection for unwanted AI content
• Custom Home page – if you simply don’t want to see a specific type of thread then there is always the custom homepage option

There are many other configurations we have on the backend that we use to reduce poor use of AI, bots and increase the quality of posts.

Moderation highlights over the past 12 months​

Whilst some of this is automated, much still requires manual review – the use of AI and humanity in balance, as it should be.

We have performed 650,603 moderation actions, an average of 1,782 actions per day, with a median of 1,103, based on the high volume of actions around specific dates as we develop our processes.

Some specific highlights:

• Deleted 200,304 posts.
• Locked 199,115 threads.
• Given 45,435 warnings.
• Spam cleaned 38,481 accounts.
• Moved 17,470 threads into the correct forums.
• And approved over 11,000 accounts, threads and posts.

In the last 6 months our BHW AI processes have: (AI detection feature)

• Identified 95,377 posts requiring scanning
• Deleted 22,524 posts
• Sent 6,403 further posts for moderation

Remember this isn’t simply a delete event. The BHW team must read the content, consider the action and submit the change. This includes consideration of decision quality and the impact of the member in question.

How we will address this going forward?​

We’re always looking to improve the quality of the content posted on BHW and what reaches your screen. From the stats above, you can see how much you never get to see. There's moderation actions to marketplace bumps that don’t make the homepage to AI spam that’s reviewed and stopped before it reaches the feed. As you can see we already do a lot. Of course, we can and need to do more.

We can improve how we use the existing AI content checker to draw that balance between the use of AI in a positive way. We want members to be able to use true assistive AI such as translation tools and writing assistants like Grammarly but stop copy-paste from a question asked to an LLM delivering low-value content. In my book, original thought reasoned through LLM's is fine - copy-paste low value content isn't.

We need to Improve onboarding reviews of new accounts using verification tools and more frequent manual account reviews.

There's the revamped Exec program that actively encourages quality threads, focused on successful members and engagement.

When will the next update come?​

Going forward, we'll share updates like the ones listed above more frequently and clearly, without giving away the secret sauce.

We shall provide stats to back up why we’re considering the changes we make and look forward to hearing from you about what you think of them and how they impact your experience on BHW. BEFORE we set them live.

In the next 30 days we will update you on the changes we will make to the AI content checker and the onboarding process behind the new accounts.

Encouraging more quality content through ideas like the revamped Exec program will take longer, but we aim to have it live by the end of Q1 2027.