- Since OpenAI launched ChatGPT, it seems like the default for all LLMs and artificial intelligence tools is some variation on a chatbot. This is *not* good for users. Amelia Wattenberger [[Wattenberger, Amelia - Why Chatbots Are Not the Future|points out]] several reasons we need to move away from the chatbot interface: - It doesn’t tell users *how* it should be used, it’s just simple text input. - You have to continuously provide [[{6.3a} ai prompts are just context|context for your prompt]]. - Prompts and responses aren’t adequately connected; you have to keep referencing one to understand the other. - Users are forced to switch between **implementation** (asking a question) and **evaluation** (reading the response) without any sense of flow between the two. - It’s pushing users toward shorter, lower-quality content instead of being a tool for deeper critical thinking. - [[Henley, Austin|Austin Henley]] puts all of this more simply in [‘Natural Language is the lazy interface’]([Natural language is the lazy user interface - Austin Z. Henley (austinhenley.com)](https://austinhenley.com/blog/naturallanguageui.html)): > Expecting users to primarily interact with software in natural language is _lazy_. > > <mark style="background: #D2B3FFA6;">It puts all the burden on the user to articulate good questions.</mark> What to ask, when to ask it, how to ask it, to make sense of the response, and then to repeat that many times. > > **But a user may not know what they don't know.** > > A good user interface let's me iteratively and incrementally explore the problem and solution space in a variety of ways. - [[Appleton, Maggie|Maggie Appleton]] seems to agree and provides ideas for some amazing alternatives in her [Language Model Sketchbook]([Language Model Sketchbook, or Why I Hate Chatbots (maggieappleton.com)](https://maggieappleton.com/lm-sketchbook)). - [[2024-06-22]] — **ChatGPT is becoming *harder*, not *easier*, to use.** - This has been on my mind a lot since the launch of GPT-4o. I’ve always had a hard time with CGPT giving a ton of extraneous detail, repeating my prompts back to me, and not following my suggestions for tone and brevity, but it’s gotten significantly worse since this last launch. - The software itself has had a lot of QOL updates that I appreciate, like custom GPTs, better memory, custom instructions, and so on, but the responses are *sooo looonggg*. Makes it difficult to follow the conversation.