From Clippy to ChatGPT: Is our content AI ready?
Remember the 2010's when chatbots were the future, how many sales pitches were suffered with reps telling us “this bot uses AI” and that this was essentially the saviour of our channel shift transformation program.
Beware of false prophets, this tech turned out to be depressingly dumb once you started to develop on it. Yes the language comprehension was incredible, but that was where the AI ended the rest was soul destroying manual configuration.
In reality this meant trying to think of every different way a user could ask the same question, just so you could cover every base and successfully point them at the correct answer 😫.
Where was the AI?
Very little about these systems appeared intelligent. This was faux AI, attempting to trick the user into thinking it was dealing with something intelligent.
So what did the user get, usually just a frustratingly elaborate website search/decision tree. Forcing people to pretend they were interacting with a bot with a zany personality (“Hi my names Clippy…”) that either did not having the answers you wanted or simply spat out links to pages on the website.
In some instances a more immersive experience was offered, but the dev work to achieve this meant the bot was narrowly scoped to a very specific topic.
Most rarely had the answers or transactions you wanted and as a safety net passed you across to customer services, this wasn’t the future of customer contact.
Are things different this time?
Could these new LLMs (chatGPT, Bard etc) be the AI future we dreamed of? They seem to be able to offer a human level of interaction for the user, and can interrogate and answer questions on the information you put into it.
No manual configuration/intents/painstaking levels of configuration and testing. It looks like we can simply scrape our website content, feed it to chatGPT and the bot can give you answers. Now this is a pilot project worth exploring.
But we now have a new problem, how accurate/up to date is our content (I’ll be honest, I’m not confident). If the technology works as expected, it lives and dies on accurate information in the correct format.
Future of digital teams
I think we’ve caught a glimpse of the future for digital service teams. We have to become the owners, creators, protectors and maintainers of service information and APIs that anything can plug into, from chatbots to websites. This sounds logical but is no simple task.
Information has to be accurate, up to date, correctly structured, regularly reviewed and maintained (and did I say accurate and up to date??). This is a job that can’t be offloaded to service delivery teams, this is a new specialist skill.
We will essentially be creating a council service database, the Wikipedia of council services that will become the businesses most precious customer service resource. It will be the core of all digital customer platforms, a single source of truth to win the customer self service battle.
This is the model we must design and prepare for, out of date or inaccurate information must be attacked and we must find a way to ensure we capture layers of detail not previously required by the web, enabling AI to handle the full enquiry without passing you to customer services.
Easy 😀 🚀
