Hovhannes Avoyan - the co-founder and CEO of PicsArt, published an article about Artificial Intelligence (AI) on VentureBeat.com - the leading source for news, events, groundbreaking research and perspective on technology innovation.

The rapid development of AI is like the birth of the universe right after the Big Bang: incredibly hot and expanding rapidly. AI is so dynamic that no one can predict what the field will be like in five years. Based on its current development, we believe that we will witness the standardization of models and frameworks and potentially the creation of a marketplace for AI models.

Here’s why: Machine learning (ML) models have become the core of most modern software because of their ability to adapt in numerous ways and their high efficiency, both in terms of product functions and implementation costs and the overall business itself. We can compare this stage of AI development with the early days of the internet, when everyone acknowledged its importance but didn’t see its full potential. Of course, it presents the same risks as the dot-com bubble, but ultimately that crisis proved to be beneficial in terms of pushing the further development of the tech industry overall.

As the AI industry is brand new, most companies leading the field have open-sourced their ML models to protect their dominance and set the rules for the market. Such an open source initiative provides an efficient way to get the most out of hardware capabilities and allows developers to collaborate and share their machine learning models in a unified fashion.

If the trend develops further in this way, new standards and platforms will emerge, like popular deep learning frameworks TensorFlow (Google), Caffe2(Facebook), and PyTorch (community, Facebook), along with higher level platforms like parl.ai (Facebook) for conversational AI.

This approach helps integrate ML models/learning algorithms into software to reduce time spent on routine coding. Developers using these frameworks can do high-level coding for their products directly, which would solve many problems with deployment and hardware compatibility. It’s important to add that collaborative development of AI frameworks is beneficial for many companies because the intellectual property is not in the software itself, but in the models that they are built on. More engineers building better software will allow better infrastructure and better models to be built — and those models are where the value for software companies lies.

So, an open source framework approach will bring industry standardization. Today, software engineering involves integrating a lot of third-party logic blocks that work together to produce a solution for a given problem. Making ML models composable with hard-coded logic blocks into one differentiable stack is a very possible outcome of what our future software might look like it terms of unification of the market. We will see less hard-coded logic and more models in apps, like conversational interfaces for users. We will also see frameworks integrating different models into new apps. For example, an object segmentation model can be connected with an image transformation model to provide unique image filters. Or a voice recognition model can be connected with a bot model to manipulate it.

This world of new models being developed every day creates a demand for an efficient way to share, collaborate on, and distribute these models to empower many new and existing businesses using AI. New ML model marketplaces, where developers can share ML models and quickly tailor them to their needs, will emerge to create value for many kinds of businesses. Traditional software businesses will radically change, providing even more jobs for data scientists than the industry has now. Bearing all this in mind, it is clear that AI is the future of software engineering, as well as a new marketplace and an opportunity for various consulting services.

Read the full article on venturebeat.com.