Adoption through better experiences, not through capabilities
In Ethan Mollick’s post What Can be Done in 59 Seconds: An Opportunity (and a Crisis), he touches on a subject that is so important that it’s easy to overlook it as a “of course that’s what’s happening” type of thing.
He highlights 2 recent developments that are not as model-capability-focused as most of what has gotten people excited over the last year.
Integrating LLMs into existing tools that a billion people use: Microsoft’s Copilot is now integrated into their Office suite
Creating and sharing purpose-built GPTs: OpenAI’s GPT Store and teams
His articulation of challenges in getting the most value out of ChatGPT is revealing:
It certainly didn’t help that the ubiquitous chatbot approach to AI hid a lot of its power, which was only revealed after hours of experimentation.
This is a much better way of articulating what I oversimplified (and ranted about) as a usability problem. It is a usability and a usefulness problem, but not simply a “user didn’t know how to complete a task” style of usability. Open-ended chat is amazing if you know how to have the conversation and ask the right questions. I think we can all agree that people are generally not great at doing exactly that in real life, so why would we collectively get better at asking great questions when we’re talking to a bot?
This feels like the next big step in the overall trend of how GenAI becomes more and more mainstream—usable and useful.
The technology came first: LLMs existed for a good while, but only a few people used them as they grew in capability.
The first good UX came next: ChatGPT was a usability breakthrough—for all of chat’s usability shortcomings, it is also the thing that made the technology accessible to billions.
Now comes thoughtful products, experiences, and ecosystems: the overall ease of use is increasing as the tech is being integrated (a la Microsoft), democratized (a la GPT Store), and normalized—not just a select few will be using this in the background of their work, but it will be part of their collaboration in teams and more.
And this growth phase as I’ve described it is completely ignoring how fast the technologies underneath are advancing. That will also increase adoption in more and more scenarios, but will have to follow the same paths to success as any other technology: to meet the fullest potential, everything must be thoughtfully and deliberately designed into peoples’ lives.
