Prompts: The Stored Procedures of LLMs

Today I took the amazing course on prompt engineering, taught by Isa Fulford from OpenAI and Andrew Ng from DeepLearning.AI

I highly recommend to take this course, it took me about 2 hours. You can access it here. It doesn't require sophisticated technical skills.

The course is very practical, beautiful examples, Isa and Andrew build with you prompts on the fly. I learned a lot from this.

I belong to an older generation of Computer Scientists. I grew up with databases, and stored procedures. I used Ontologies for Semantic tagging and searches.

I remember when I went to the DBAs to ask them to write some special stored procedures, to get me some data. They knew how to do it, in a secure and efficient way. Then I went to a web designer to ask them to move the data to the web site to display some info for the users. It was very complex !!!! It took forever !

I realize that the prompts are the "Stored Procedures" of LLMs and Generative AI. The difference is that we don't have relational or object databases, we don't have SQL or SPARQL, you can specify how to get the result, (in what format), how to be formulated , formally or in slang, in what languages . You can also indicate to the LLMs how creative to be, by specifying something called "Temperature". If you use "0" the answer will be the same to the same prompt all the time, if not, the answer varies. You can look at a message entered by a client and find out how is the client, happy or pissed off about a product. You can indicate the AI to respond to a message according to the senders sentiment. And all these can be done in a single prompt written in almost plain English. It is fantastic !

And you can build a smart chat bot in no time !!!

Here is the summary of the course:

You can learn all these checking out this course and you will feel like me that we must be grateful to live during this time when the technology solves problems that created lots of headages before and some were unsolvable.

But in the same time we have to be careful, to make sure that we are not creating technologies that can bring harm to others and also to prevent others who are using our creation to harm others. This is called Responsible AI. This is the closing message from Isa and Andrew at the end of the course.