🤖 Common GenAI Usage Patterns
AI tools are slowly but surely creeping into in my daily work, and I notice 3 main usage patterns emerging (in order of usefulness to me): autocomplete, explain, and generate.
Autocomplete
While the most basic of the three, autocomplete is most frequently used for me. I use Apple Notes and Dictation often to help me transcribe and jot down quick notes throughout the day. Because those autocomplete features are so seamlessly integrated into the user experience, they save me a ton of effort in typing and really enhance my workflows in noticeable ways.
Explain
I find the this pattern slowly replacing common searches I normally do through search engines. It’s just much easier to ask LLMs to answer to my specific questions in structured and concise responses, than to have to filter through pages of links and read snippets to find what I want. I find this non-time sensitive “fact retrieval” type search covers 60-70% of my cases cover. The other 30-40% that still requires manual search involves finding the latest information, like news, events and home listings, etc.
Generate
So far I rarely use the generate functionality, but sometimes I am just in awe of what it can do. The creativity that generative models exhibit is so interesting, whether it’s image/video, code, or text generation, the generative usage pattern has so much potential to enhance our lives in fun and exciting ways. It’s definitely a usage pattern I’m looking to explore more.
Each usage pattern above has its rightful place and suitable use case. A genAI product may be hard to use because there’s a mismatch between the use case and usage pattern. For instance, I don’t think generating emails with LLMs work well. Instead, email autocompletion feels much more intuitive and useful. Conversely, autocompleting the writing of a function can be hard to do, but describing what input/output you want in the function and let a model generate and explain it may be more productive.
All I’m saying is, usage patterns matter when it comes to developing useful genAI features.