Is ChatGPT autocomplete bad UX/UI?
I get it. I am not the only user the world revolves around. When an app does not behave the way I would prefer, it’s probably because most people have different preferences, and the app is optimized for them.
But sometimes, I encounter UI/UX decisions that puzzle me. Not because I dislike them but because I have trouble imagining anyone likes them.
One of those puzzling UI/UX decisions is ChatGPT autocomplete.
I was able to come up with only two possible explanations for its existence:
- UX/UI experts at OpenAI (and other multi-billion companies) have a flawed understanding of how people chat.
- I am weird and chat differently than most people.
Before we conclude that solution #2 is the correct one, let me make my case for #1.
Autocomplete is helpful in search bars because the potential response space is small and the likelihood of success big - two words can be enough to guess a four-word search query.
Chat is different. I write long-ish posts that are a lot more complex than a typical search query - the likelihood of ChatGPT getting what I want after the first characters is minuscule. My experience corresponds with this - ChatGPT never guesses correctly what I want to write. Literally not once.
This explains why ChatGPT autocomplete is useless to me. But why does it feel so annoying and frustrating? Because the mental process I use when chatting differs from the one I use when searching.
When using a search, I already know what I am searching for. Using chat is a lot more like regular talking… and when I talk, I do not know exactly what I will say before I say it. I am thinking aloud. And when someone is talking over me, it is distracting. My train of thought gets derailed when I start typing and ChatGPT inserts something completely unrelated.
Imagine if ChatGPT were a colleague. You want to discuss how to implement a new feature in your app…
Should we…
…go to the library? Exercise more? Make compound chicken for lunch?
No, I want to talk about…
…the latest tech trends? Healthy lifestyle tips? My favorite books and why I love them?
For Christ’s sake, could you shut…
…the front door? All the windows in the building? Our business?
A colleague like that would drive me crazy. And I would be puzzled if other people were like “Oh, he finished my sentence again. He is awesome”. What am I missing?
UPDATE: There are some interesting comments on Hacker News. One of them points out that the autocomplete could be more of a hint on how to use ChatGPT than a feature to help you finish writing as quickly as possible.
Probably, at one point, they did some user testing or similar and figured out that most people don’t even know what to ask these chatbots. So they add some “quickstarts” for people to “get inspired” by.
When I was writing the article, I quickly considered discoverability as a reason they added this feature and discarded it immediately because everyone knows how to use chat, right? Well, maybe not this kind of chat. Chatting with artificial intelligence is new.
Other commenters do not find those autocomplete suggestions distracting because they never look at them. How simple. Simple, yet surprising to me, because I literally cannot NOT look at them. Perhaps I underestimated my distractability. Maybe it is my perception quirk. I find it very difficult to read a page where something is moving, even if it’s just a little GIF somewhere in the corner. Am I the weird one? Possibly.