4 minute read time.
CLF Sept '23 Debrief by 3sides: The impact of AI on (online) communities

AI is here! All of a sudden. It was rumbling already in the tech scene for some time but Open AI’s ChatGPT brought its capabilities to the masses with a bang. We are now living in the AI era and new Generation AI will put Gen Z into past tense. Similar to the first atomic blast on July 16, 1945, and its application 3 weeks later on August 6, 1945, in Hiroshima, we are now living in a new era and we won’t ever go back. Ironically the inception of AI started in the same era with Alan Turing’s cracking the German WWII Enigma code. It just took a few more years for hardware to enable it.

The question on every community manager’s mind is: What does it mean? What are the threats and what are the opportunities? I’ll share with you the findings of our community of Community Managers who meet in a monthly meetup called Community Leaders Forum. Several of you are already a member. We do not record as we agreed but the residu or our meetings is something that can and should be shared. We are in our 3rd year of operation and the conversations are really helpful for all of us to develop insights and share experiences. We spent this year already 3 sessions on the topic and we’re definitely not done as we now see the first AI applications in online communities.

Anybody who claims to know it all at this point should be met with caution. I’m committed to this group of extraordinary people by being the pathfinder. Driving searches and looking for answers. Here is what we as a group found.

AI tools will give ‘digital nuclear power’ to individuals and organizations. Many will use it for good and some will use it for bad purposes (ever the optimist!). The first GPT usage in communities we saw were spammers crafting credible questions, to which real people responded. Even moderators were fooled. We currently see the internet exploding with modified or crafted identities, telling credible stories that were invented. People will think twice before they start subscribing to anything in the future including your community, similar to your current email practice where you only allow emails from people you know. There is a huge opportunity for community managers to provide this authentic and safe haven for members to feel at home and to trust the organization and the members.

The big question then is: What is trust? What does it look like? What are the ingredients? What are the levers a CM can pull on to increase trust? I have been doing communities for close to 2 decades and nowhere did I ever find any expert talking about this topic and giving evidence of how it works. So, I set out on a quest some years ago and went into scientific research of trust and also books written about it. I won’t bore you with all the findings but the residu of that research is captured a few conclusions:

  1. Fear is nature, Trust is nurture

  2. Ease of Use, Usefulness and Trust in Organization lead to higher Trust in Members

  3. Responsiveness and Authenticity lead to higher trust

  4. Trust comes when it is given

We reached some serious unified nodding in the team on these findings. It leads to a discussion about ambition. Moving from a utilitarian “Q and A” model to a hedonic “I belong here” model.

A second opportunity is also related to a threat. While it is desirable for search engines to crawl your site, it is not desirable for AI engines to scrape your shared IP in your community into their Large Language Model and provide answers with your content without the members ever visiting your community again while the AI company make a buck on your hard work and investments. It will be imperative to put a digital “No Trespassing” on your site as many newspapers already are doing. The opportunity is of course to build your own Large Language Model and have your own version of Outbound Chat GPT generate an answer based on your content on YOUR community. We have been helping to implement the first customer with such a setup. It is a great alternative to federated search. Also, we see opportunities to write assisted responses that lead to positive higher sentiments simply by the choice of words. This too adds to a higher sense of belonging.

We talked about combined GPT where an Inbound GPT question gets answered by an Outbound GPT engine of the company. We even talked about the doom scenario where every community disappears and some giant all-knowing LLM answers anything. No so very likely. We found more opportunities than threats.

 The GPT mechanisms discussed

We will continue our quest to find answers, no matter what topic you bring to the table. If you want to join that conversation, it is free for all. In return, we ask you to tell your story. Every story has proven to be important to the team. Don’t think you have nothing to say. Come join us.