Where is AI going?
Every day I open my email inbox and I find dozens of emails with enlightening articles on Artificial Intelligence. I am always stunned by the sureness of the predictions envisioned in those articles. It seems that they know exactly where and when the next transformation will happen. Trends, sectors, methodologies, industries, clients needs. It seems that everything has been visualized and this is like planning for the next development phase.
Is it like this? I don’t think so. I believe that no one has a clue and yet many people have a clear view of the future.
And why is that? Because we are in an exploration phase. Apart fo some well-exploited sector, AI and ML are still trying to find their place in the world. And this is not easy. There is the resistance, of course, the difficulty of convincing the enterprise that AI is the right solution for many problems or needs. And there is a difficulty in scaling the PoCs or MVPs in production. This means that executives might change their mind and, as a side effect, the course of AI development.
Is it all about choices?
It is not. Like many exploratory fields, it is clear — quite clear — where you start but it is not possible to predict where you go, and how many directions you take. It’s the beauty of it. And it’s the horror of it :)
However, there is something we can try to predict.
- AI are voracious of data and despite all the stories about how many tons of data we generate, there are never enough. There are many cases where there are some data but not sufficient to train a model. And what do we do in those case? We can pray, of course, although I am afraid it won’t work. you can drop your model? That’s a way usually taken. However, a rising topic is to generate synthetic data, based on real-world data, which helps to train models and to do it in real-time too.
- AI is usually trained offline using large datasets. However, it is often consumed on peripheral devices. Sometimes, the latency can be a huge problem, especially in healthcare and autonomous driving. Moreover, the exchange of data can cause bandwidth saturation. Therefore, edge Machine Learning could be another trend for the coming years.
- As I am writing this article, there are 7,841,018,837 people living on this planet. Resources are limited and the need for healthy food is on the rise. This means that new techniques for agriculture are not an option. The farmers of the future will be more on the tech side. And AI will play a predominant role.
- What follows represents, more than the others, my point of view. We have passed from sharing things to our friends, relatives and colleagues to everyone in the world. This is the social era. At a certain point, this will end and the privacy will be a primary need. The same will happen within the e-commerce space. AI will not focus on bombarding people with useless proposals but target them on real needs, on time and on the right device in the right location. Everything else will be perceived as annoying.
Conclusion: should this mean that I know where we are going? Absolutely not! I can only guess it and, as every guess, it could be wrong. My approach is to keep my eyes and ears wide open. This is a changing world and technology is moving fast and in many different directions.
Thanks for reading.
