From Conversation to Collaboration
The Evolving UX of Generative AI
Happy New Year, everyone! I hope you all had a wonderful holiday and are feeling excited and recharged for what is sure to be a wild and exciting 2024!
At the end of each year, I like to set goals for the upcoming one. I do this both in my personal life as well as at work.
While forming my thoughts for the year ahead, I first take a moment to reflect back on the previous year. I think about what happened, what I learned, and (when relevant) what others in the world are thinking to help me view things through a different lens. This time, more than ever, I decided to spend some time digging into various perspectives and predictions for Generative AI.
I read posts from VCs investing in the space, notes and key takeaways from folks who attended NeurIPS, and thoughts from prominent Generative AI people on LinkedIn, Twitter, and blogs. I also read through dozens of “prediction posts” - ending the year with my own fun spin on this, a BINGO card for Generative AI in 2024.
Lately I’ve been spending a lot of time thinking about a move from conversational interfaces to collaborative ones.
Right now, ChatGPT is the most popular generative AI application, with over 100M weekly active users; but it’s just a chat agent. Granted, a much more successful one than our favourite relic of technology’s past - Clippy.
We took the most powerful technology we’ve seen since, well, the smartphone, and turned it into a chatbot. Surely there has to be something more….
It does make sense that we started here. It’s also not too surprising that it got so popular (at least in retrospect). The approachability and intuitiveness of the interface (having a conversation!) combined with how it can fit into existing workflows (pause what you’re doing and ask a question) makes for an easy combination.
That said, just because something “works” doesn’t make it the best solution (faster horses, mail-order DVDs, encyclopedias). For example, I find myself often interrupting my current workflows to exit out of them and consult with various generative AI tools, whether it’s to ask a question to Bard or ChatGPT, to generate an image in Firefly, or a video on Runway. Even writing these blog posts requires jumping between many different tools.
It’s interesting to see how the conversational UX is also evolving - take ChatGPT for example - you can now highlight certain parts of the response and use that in a quote to reply directly to…
I suspect that in 2024 we’re going to see a lot more natural and “baked in” ways that generative AI fits into different workflows. I also suspect that we will see a lot more products that change the way work is performed, making it more efficient, and powered by the new and emergent capabilities of generative AI.
So what new interaction patterns will we see in 2024? Well, more on that to come later - but I will say that this is one of the big reasons I’m excited to be enabling developers through Google AI Studio and the Gemini API - to see what new and interesting things people will build!