The Time Of Our Lives

If I were to ask you to think of a trending topic that will change the course of our lives irreversibly, then there is a good chance that you might say Donald Trump becoming president again. Though it is a possibility, with fair certainty I can assume most of you might say ChatGPT. That's right, no one has been spared of it unless one has been spending their time gazing at the sky looking for balloons.

By now most of you might already be aware of the ever-growing capabilities of ChatGPT. It can answer practically any question that you could ask (sensibly, the majority of the time), generate well-written content, write contracts, debug programs (and provide a line-by-line explanation), be a caring psychologist, and even pass law exams. Let me stop here on enlisting the countless prowess of ChatGPT.

It won't be an overstatement to say Google search engine opened the world to this generation. It was able to find results for anything that one searched for. It used a ranking system to decide the relevancy or order of those results. One had to still use their judgment to gather the information that they were seeking. But ChatGPT has taken this to a different level, it provides a definitive answer to any question asked.

The reason for the buzz surrounding ChatGPT is that it has made people both excited and dreaded in equal measure. More so the latter. On one hand, people are awed by the endless possibilities it offers and the potential increase in productivity. On the other hand, they are scared of being replaced by an intelligent machine.

It is quite understandable that speech writers are losing sleep when a US Congressman uses a speech generated by this latest AI marvel. I say the latest because it is by far not the first one of its kind. Agreed, none of the previous ones made such a wave. What has changed this time around is the wide range of applicability that everyone is able to connect to.

About a year and a half back, I wrote an article about GitHub's Copilot. It is based on an AI model named Codex which again is from the same creators of ChatGPT i.e., OpenAI. The only difference being Copilot is a specific solution for programming whereas ChatGPT is a superset that can also help in programming.

At the time of writing the Copilot article, I had used it for only a short period of time so my appreciation of that was quite limited. I realized its true potential only when I started using it in a real implementation. As a pet project, I wrote a Pharmacy management SaaS product. My main objective was to create a software that could be used by doctors who have a small pharmacy set up in their own clinics, as well as learn a new frontend framework Angular.

From my college days, when I started my programming journey in Fortran (yes I am old), what I have seen over the years with so many different languages is that syntax changes but the basic language constructs remain the same. The purpose of specific languages differs but not the fundamental concepts.

So when I started the product implementation, my interest was more to learn the design of Angular and not so much in writing boilerplate code. I ended up using Copilot to create mainly utility methods and repeatable code. What impressed me was that while it was suggesting a code snippet it was always context aware i.e., though it pulled the code from a Git repo, it was intelligent enough to use the variables that I had already defined in the code. Copilot easily saved me days of effort.

AI as a whole has unlimited potential. Now, talking about the white elephant in the room, is it going to take away human jobs? The short answer is yes but then it is going to be gradual. What we have to remember is, for all its flaws, AI learns continuously and gets better. There is no point lamenting about a probabilistic doomsday. Instead of fighting this the best we can do is to embrace it. How so?

As I mentioned earlier, though ChatGPT has opened our eyes, there have been several AI solutions that have been there for a good amount of time. To me identifying a cat in any image or a specific object in an hour-long video or deducing the speakers' sentiment from audio in itself is as awesome as anything that ChatGPT does. Why not leverage and localize such solutions to make our own offerings better? In other words, create a domain-specific model.

For instance, in my organization, we are working on at least a couple of products at the moment. One is Freshflows, an Application Lifecycle Management product, used by product development teams. And the other is a media SaaS product for digital businesses like our own technology magazine Unboxed. I can straight away see a few AI solutions we can leverage in these products. For example, Freshflows can help product managers generate user stories by taking a short prompt (a set of words describing the requirement) thereby reducing repeatable tasks.

In the Unboxed magazine, today our experienced graphic designers spend a lot of time creating complex visuals. This can be speeded up by providing them with AI tools to create these in no time. There are multiple models and APIs out there that, if utilized well, can make wonders. Following is a sample where I wanted to transform myself into an astronaut. What would've taken a good amount of expertise in Photoshop and hours of effort was done in less than 3 seconds, thanks to Stable Diffusion. It is a deep learning, text-to-image model which did this magic just by taking a simple prompt.

My point is, we can imagine umpteen use cases, leveraging AI, to deliver value propositions in our own work. It is futile to dwell on the thought of what detrimental effect AI could have on our future. Make the best of what is available. We are at a time in our lives where things may not be the same going forward but we should not underestimate human potential come what may.

If this article has in any way got you into a gloomy mood, I would like to brighten it by urging you to listen to this song by Ahmed Chawki bearing the same title as this article.?