Random Musings - Feb 2025
This is a series to share my thoughts on various product/program management articles for the month. Mind you, that these are just my thoughts only!
Prime dates to remember for technology
In 1911, we had IBM and in 1975, Microsoft originated. It’s products were bundled with IBM in 1980. The progress on technology was slow with the WWW2 until 1945 and then slowly, we had some progress. In 1989, WWW and HTTP originated so that information could be shared beyond IT specialists! We then also had folks come up with the idea of open source. We had LINUX in 1991! And then you can see how technology started changing the landscape and with sharing of information technology is changing almost everyday. From decades, it shrunk to years and then months and now days for developers to come up with new technology because there is sharing!
In Feb 2025, there were so many articles on how AI can solve everything and take away a lot of jobs! Let’s think about a few things.
Should AI solve everything?
There are many simple problems to solve for especially small businesses. For instance, even today, there are businesses that need automated actions like say time-based appointment reminders, which could be handled by simple code. As of today, many organizations have their employees manually email such reminders.
So then what should AI solve? Some applications that I could see are as below - 1. Healthcare (Pharma, med-tech, dentistry, life-sciences) - There are several diseases (Especially end-stasge ones), that don’t have a cure today. Using AI, we can possibly come up with solutions for the same.
Supply chain - We have certainly had great strides in this industry by introducing “home delivery” but we can do better.
Financial services - Everyone wants to ensure that they have enough saved for retirement with inflation etc. It would be great if we knew accurately if we truly had enough saved!
Few things to remember about AI
You need to define the goal of the AI model. Is the problem worth solving with AI? What is the utility of the AI solution and how safe is it? Is it commercially viable? You still have skeptics and early adopters alike!
Coding is required for AI as well!
Data forms the basis of AI. We still need different types of data - training, validation to train the AI model. You would also need real-world data, which would be entered with applications that are available today. Data cleaning is one of the most complicated tasks in the process for the AI model to use it!
You still need buy-in from stakeholders for the AI solution and determine the workflow with end-users.
Final Thoughts
You ’d still need data for your AI solution. You need automated solutions for simple problems and don’t need AI. So, it is important to determine which problems need AI vs not. For the ones that need AI, you also need domain experts (eg: for healthcare, you might need clinicians) from the beginning so that they can verify the accuracy all along. Traditional jobs will evolve to new ones!