.svg)
“The generative AI process is a black box”, you say. “There’s no way to know how the machines think, what they get or don’t.”
Untrue. Together we can take steps to see how a LLM may have reached its conclusions.
First, we can identify every possible digital data source the chatbot may have scanned and ingested, even if we don’t know exactly which elements proved most important.
Second, we know how machines generally read and score that data, e.g., Natural Language Processing (NLP) Sentiment Analysis, a well-established approach used by search engines drawing off a common data foundation.
So, each time a LLM delivers a summary, there are several elements every brand and business can analyze and many it can even control. That means we can act. We aren’t just passive bystanders, helplessly watching machine-learning powered search takeover.
What we can do is Artificial Intelligence Optimization. AIO, for short.
It starts with reviewing all the content about your business or brand that’s visible on the web and then assessing how a Chatbot or search algorithm would read, score, and summarize it. The results can be relatively rapid if done right.
You start by breaking your content into two categories:
--Your Platforms: your website, your content and social media pages, your meta tags and keywords across these platforms
--Everyone Else’s Platforms: Wikipedia, digital media (trades, top tier), influencers, etc.
Then your run a series of tests to see just what the machines see:
--Interaction: error analysis, A/B testing, LLM input testing
--Language: NLP testing
Finally, you optimize across all your content, (i.e. everything the chatbots can read or see which you also control or own). Steps include:
--Timely, regular publishing around key themes, including Thought Leadership (i.e., great editorial and video content that actually adds knowledge, not promotional dreck) from key executives
--Keyword optimization (based on sentiment analysis)
--Data presentation: infographics and other visuals
In one of the cases mentioned in part one , our approach delivered real change driven by an updated website, improved, more relevant content using the right keywords in the right sequence, and regular publishing. See below (edited to protect client confidentiality):
“These areas highlight [Company’s] multifaceted role in the…sector, underlying its position as a leading full-service…firm with a global reach and deep expertise across all sectors and regions.”
It can be that simple. If you take the right steps and stick with them. Even if the outcome isn’t rapid or only an incremental improvement, you will be doing everything under your control and in your power to take on the biggest Generative AI Challenge Companies Face.
What if the chatbots get it wrong?
