It’s been a while since I’ve written anyone on Natural Language Processing (NLP) so I figured I might as well do an article now. In a nutshell, NLP is the set of algorithms that deal with natural language data (e.g. English sentences), usually based on text data. Since it’s a fairly advanced part of data science, it is considered to be more AI-related than anything else. Nevertheless, NLP employs a large number of data science techniques and models under the hood, so it’s usually part of a data scientist’s toolbox.

What about modern NLP systems then? This kind of system deals with rough data (containing typos, strange characters like emojis, etc.) and delivers a humanly understandable result, e.g. in the form of a prediction (categorization of the text or new text). However, other modern NLP systems go a step further and deliver original text based on a corpus they have learned and a prompt. The latter is a sentence or two that help outline the topic of the text that follows. NLP systems geared towards that task pick up this topic and compose a novel piece of text that's similar to the original sentence and relevant to them topic-wise. NLP systems like these are made available through OpenAI, primarily.

All this is great, but this article is about the ethics of NLP. So, where do ethics enter the picture? Well, when you have an A.I. system writing a piece of text for you at practically no cost and no responsibility, the amount of information out there is going to skyrocket. The worst part is that much of that information would be seemingly real even if it wouldn't bear any relation to reality. In the case where a text like that produced by an A.I. is useful enough, it can be abused for school projects and such. So, any sense of merit when it comes to essay writing and content creation, in general, is bound to disappear gradually.

Besides, this whole situation affects our morals since it’s becoming increasingly acceptable to cheat using A.I. From the mediocre AI-written articles you see published on various social media, to promote this or the other product or service, to students cheating in their courses, the NLP matter is clearly a manifested problem rather than some imminent threat. However, this can also be viewed as a test for our morality and a chance to be ethical even if others are not.

So what can you do to get a hold on NLP and other AI-related matters? It may seem a daunting task but it’s doable, given enough understanding of the subject and practice. On useful and empowering resource on this topic is the AI for Data Science book I’ve co-authored with another data scientist a couple of years ago. Although its focus is A.I. in general, it provides the fundamentals for doing all sorts of tasks. Feel free to check it out when you have the chance. Cheers!

PS - you can get a 20% discount on this and any other book you buy from this publisher. Do you remember the coupon code I've mentioned in previous posts?