Generative AI models have firmly established their presence and show no signs of fading away. With their remarkable capacity for creativity and text generation, one might expect their user interfaces to reflect equally impressive capabilities. However, the reality, as of September 2023 and based on personal experiences with ChatGPT and Bard, reveals a different story. This article explores the existing limitations within the interfaces of these generative AI services and highlights the untapped potential that awaits us.
Autocompletion
One glaring omission in generative AI [1] services, is the absence of an autocompletion feature for prompts. This absence prompts the question: Why is there no autocompletion, even as basic as pressing the TAB key, when writing prompts for ChatGPT [2] or Bard [3]? It appears paradoxical because generative AI models essentially serve as advanced autocompletion tools themselves. Users, who are accustomed to the convenience of receiving suggestions as they type in various applications, find it perplexing that this feature is not integrated into generative AI platforms.
History and Keyboard shortcuts
As someone deeply familiar with the command line and a fan of keyboard shortcuts, I've noticed a quirky habit when interacting with generative AI models through prompts. I often catch myself instinctively reaching for the keyboard shortcuts I frequently use in the command line. However, to my disappointment, not all of these shortcuts work in the generative AI environment. This isn't an isolated issue specific to ChatGPT; rather, it's a universal challenge affecting all generative AI services. Given below is a snapshot of some keyboard shortcuts for ChatGPT, but this is indeed a generic problem for all the generative AI services.
| Action | Keyboard Shortcut |
|---|---|
| Open new chat | Ctrl + Shift + o |
| Focus chat input | Shift + Esc |
| Copy last code block | Ctrl + Shift + ; |
| Copy last response | Ctrl + Shift + c |
| Set custom instructions | Ctrl + Shift + i |
| Toggle sidebar | Ctrl + Shift + s |
| Delete chat | Ctrl + Shift + ⌫ |
| Show shortcuts | Ctrl + / |
One of the expectations many users, including those accustomed to command-line interfaces, bring to the table when working with generative AI services is the ability to efficiently navigate their conversation history. Specifically, they anticipate the convenience of using the up and down arrow keys to access previous prompts and responses. This feature allows for a more seamless and contextual interaction, akin to scrolling through a conversation log. However, this valuable functionality is often lacking in generative AI services, including ChatGPT, leaving users like myself wanting for a more intuitive and history-friendly experience.
Search and Filter
Efficiently retrieving previous conversations is a common need for users of generative AI services like ChatGPT and Bard, especially for those who engage in numerous interactions. Frustratingly, these platforms often lack a basic search and filter feature, forcing users to either wade through a sea of past chats or redundantly recreate conversations. The absence of such a seemingly fundamental feature raises questions about why, given the "advanced capabilities" of these models, the implementation of a straightforward search and filter function remains elusive.
Tags and Categories
While the development of basic or advanced search capabilities may not currently be a priority for generative AI services, there is an alternative approach that could significantly enhance user experience. By providing users with the ability to add tags to individual conversations and organize them into distinct categories, these services can offer a practical interim solution. This feature allows users to efficiently manage and retrieve their previous interactions, filling the gap until comprehensive search capabilities are eventually integrated.
Conclusion
In the ever-evolving landscape of generative AI, the enduring presence and expanding capabilities of these models are undeniable. Yet, as of September 2023, our exploration into generative AI services, including ChatGPT and Bard, has unveiled notable shortcomings in their user interfaces. These limitations encompass the absence of essential features for prompt writing such as autocompletion, keyboard shortcuts, efficient navigation of conversation history, and basic search and filter functions.
However, amidst these challenges, there is a glimmer of hope. The potential for enhancing user experience lies not only in the promise of future search capabilities but also in a practical interim solution – the introduction of tags and categories. By empowering users to organize their conversations with these tools, generative AI services can bridge the gap, offering improved conversation management and retrieval. As we move forward, we eagerly anticipate the evolution of generative AI services, as they work to unlock their full potential and provide users with richer, more user-friendly experiences in the exciting world of AI-driven text generation.