Natural Language Programming
Are we there yet?
Talk to anyone who has used ‘google home’, ‘siri’, ‘alexa’, etc, and without doubt you would think the era of true Natural Language programming has arrived.
Next came large language models like lambda, chatGPT, and DeepSeek, and even one software engineer who swears that Google’s AI is sentient.
Natural Language simply is the ability for a machine to understand human language. Traditionally computers are ‘programmed’ using a small subset of language known as a ‘programming language’. Rooted in math and logic, developers use programming languages to define discrete steps for the machine to perform (process a set of inputs, into a set of outputs).
Natural Language should free the ‘programmer’ from the ‘program’, allowing the user to interact in a natural way with the machine, and the machine again convert inputs into outputs without any specific program to fulfill the task.
And again, on the surface at least, it appears we have arrived. It is only when we dig down into the nitty, gritty details do we begin to see that not all is as we have thought.
Did you know? - there are specific courses you can take today on the art of prompting. That is how to specifically craft your sentences to systems like chatGPT and DeepSeek to improve the results you wish to obtain. The Ultimate ChatGPT Prompts Cheatsheet:
And of course, this is the great irony, is the ‘prompting’ of specific phrases to improve your ability to get results at all different from the programming writing code in a programming language? - I submit to you, no it is not. Perhaps it could be considered some type of new ‘6th generation programming language’ where the ‘AI’ can do some (if not most) of the work for you, but you still need to find the ‘right’ way to ask the question to get the deep answers you seek.
A true General Artificial Intelligence model would not require any special type of prompting, nor should there really be any bias at all to how a particular questions is stated or ‘framed’. Except perhaps as a human might do to empathize with the other person, or follow a particular moral belief system.
In conclusion, while there can be no doubt that large strides have been made in the field of Natural Language Processing, we have yet to fully succeed, and will not succeed until the idea of prompting is no longer something we need to learn as humans, but instead the AI must adapt to our particular way of speaking - naturally!
In order to make AI ‘stronger’ we need more advanced simulations of the real world environment.