Brian T. O'Neill’s Post

View profile for Brian T. O'Neill, graphic

I help midsize B2B AI & analytics companies unlock lost revenue through UX design. Host of 🎧 Experiencing Data podcast & founder of the DPLC community. For free tips, subscribe to my weekly email list.

#LLMs are full of #UX challenges. These magical things....also frustrate us...because they're so "smart"....and so stupid at the same time. Should we be yelling at the product teams? A notable user experience element I've been thinking about when working with these is whether the LLM understands when I am frustrated - does it realize I'm twisting, turning, and morphing my prompts in a desperate attempt to squeeze out something better? Something more useful? Does it know that I'm struggling? Does it know that I don't know if the problem is that there is no good answer, insight, or better version to be had -- or that it's "user error" because I'm too stupid to speak "prompt" fluently, and don't feel like downloading a bazillion LinkedIN PDF hack sheets on how to game the system? Does it know I'm not a gamer, and having to learn cheat-codes is a tax and not a benefit? This feels like an area that could be improved - understanding intent, rephrasing of questions/prompts, and seeing patterns that suggest "this user is trying to get something out of the system that doesn't exist or cannot be improved upon." It seems LLM #UX could be improved here -- with AI -- because the training data seems to be right there in the prompts. How do you most often feel when you end up down one of these holes? Do you feel like "that's the cost of using the tech"? ~~~~~ ⬆️ Join 2k readers: tap "Visit my website” to get my latest thinking in your inbox 🚫 Don’t like LinkedIn’s algo deciding what you read? Tap the bell 🔔 under my profile banner to get all my posts ♻️ Please repost if this was useful

To view or add a comment, sign in

Explore topics