… and then there was prompt

Everybody was (and still is) excited about ChatGPT (text, language) and Midjourney (images, visual). They are being used, discussed, debated, scorned, embraced, and misunderstood. Like it or not, people are using it. But, they are beasts that need to be tamed.

I have tried to use both of them but the results are still not to my liking. The reason is that I don’t understand the proper “words” or “keywords” to tell them what to do. To instruct them, we need to type in words as commands. We are facing with prompt! I guess, I still need to learn how to give commands properly. I need to learn how “to prompt.” “Prompt” has become a verb.

Many said that we need to teach students how to prompt. Not in my opinion.

You see, “prompting” is similar to “command line interface” (CLI) in the old computer days. If you are old enough to remember DOS (MS-DOS) or UNIX (or older operating system), then you know what a CLI is. We give commands to a computer by typing commands. Prompting, that is. This is the only user interface at that time.

Computers became more powerful and we have other options. Starting from CLI, we moved to menu-based, then not long ago we started to have GUI — the Graphical User Interface. This is what we have right now. But what about the near future. Voice is the natural path. The reason we use voice is there is not enough space for us to use GUI in many devices. In my smartwatch — Mi-Band — I cannot use GUI because the device has a limited surface. I can select operation based on (bad) menus. That’s the only option available. But, why not voice? Computation — the need for higher computation platform — is what restricting us to use voice.

Moore’s law is still in action; computing power doubles every 18 months (or 2 years). While we are sleeping, progress in computing is still going. Now, it is the time we invest on voice command. The computation to deal with Large Language Model (LLM) is going to get better in terms of resource usage.

In summary instead of going through the old path — CLI, menu, GUI, then voice — we should jump ahead from prompting to voice command.

Any takers?