Hello World, Happy New year :)

I would like to begin by wishing you a very Happy new year, 2024 😊

It’s been over two years since I last wrote a blog. With the rapid pace of technological advancements and the shift towards Generative Artificial Intelligence (Gen AI), I felt inspired to return to my old policy of "Learn by doing.." and start blogging again. Although I’ve been away from this space for a while, I pushed myself to write this very generic blog - to get started - to motivate and push myself - and to hope I gather the time and courage needed to write some good material. 

I’m sure many of you are familiar with FOMO, or the Fear of Missing Out. With so many new skills to learn and technologies to explore, I’ve been feeling FOMO for quite some time. However, I’ve recently realized that I should rather embrace JOMO, or the Joy of Missing Out. Instead of worrying about all the details, I’m going to focus on the one topic that I can really sink my teeth into, and now that Microsoft has released a co-pilot for almost everything to do half of my job. I plan to joyfully explore the one topic on at a time. And hope to share while I do so - Wish me luck😅

Just like most of you, I have been exploring/utilizing Generative AI quite a bit lately. So, why not we talk about the same today. 

So, what is Generative AI? 

Why don't we ask one of such tools available - Bing Chat. 

Ok - when you post a question. Bing Chat immediately triggered the chatGPT-4 engine owned by OpenAI and did a search through the internet and wrote a response back. Not only that, Bing Chat also provides the reference links which it used to generate the response. 

So, in simple terms - Generative AI is an AI capable of generating new and original content based on its learning from various sources. The generated content could by just plain text, as shown above (or) could be images and videos too - also a sample is shown in the above screenshot. Not only this, GenAI is widely used to generate programming code as well (Co-pilot for GitHub) is a prominent example. 

OpenAI says this is only a minor step in their journey to achieve Artificial General Intelligence (AGI). Otherwise known as Full AI. OpenAI and many other similar tech companies want to create Artificial Intelligence that can act like a human can do, it should be able to take up any mental or intellectual task which a human can do.  

Ok before I forget, In the above screen grab, the question I raised to Bing Chat (the blue bubble - top right) is called a prompt. There are much better ways to ask a question to GenAI model, the better you engineer your prompt, the better response will be coming your way. We even have a term for the skill of asking the right question to a GenAI model, it is called Prompt Engineering😁 

So, now let's try to provide some context to Bing Chat and ask about Artificial General Intelligence and see what it has to share. 

As you can notice, the model changes its tone, doesn't write too long sentences and even adds a smiley automatically. This is where the aspect of human touch has been added directly into the model. This is all taken care already by Microsoft while training the model in the first place - so all you have to do is set the scene and wait. 

There are multiple ways to engineer your prompt, some key approaches are: 

1. Specifying the Audience - like I did above
   Example: But Imagine you are speaking to a 15 year old kid. 
   
   
2. You can also set the Tone in advance
   Example: Be caring and empathetic with your response. 
   
   
3. Also the most commonly used (by me, atleast) is Personas.
   Example: Act like a X++ Developer and suggest how to.... 

So far so good.. right!! 

However, not so much in reality. The biggest risk with Generative AI Models as of today is Hallucinations. A term used to describe a defect in GenAI models - it means that the model could create responses that are might look very convincing (might even suggest them to you as facts) but are entirely fictional and invented by GenAI to prove its point and respond to your question. This is a bit scary, ain't it? 

This is the reason why, GenAI models are primarily suggested to be used as co-pilots / supporting partners. Meaning someone knowledgeable has to review the responses and take support from co-pilots in documenting, drafting a template, creating new ideas / brain storming and other mundane tasks you can think of & get it done by GenAI models.

Another concern could be people misusing GenAI models, even training GenAI models to create Deep fakes - manipulate existing audio, images and videos to create pretty realistic clones of them with added corruptive content. So, I would say we need to learn more in order to be better prepared while striking the balance to use co-pilots and GenAI models to reduce the work effort on our jobs and daily tasks. 

I summarize today with a prompt illustrating various components (Persona to use while responding, Length recommended for response, the tone to use, and setting the possible audience for the response)

You see, there is a minor touch of rhyming words but not using too complex English literature and also shared what it thinks would be the most prominent outcome of the year. Probably has hallucinated based on the previous conversations - never know, but let's raise the glass nevertheless 🥂🥂🥂