No longer a ‘dark art’, machine learning (ML) is now accessible, affordable, and changing the way we create. In this piece for ExchangeWire, Kai Henniges (pictured below), CEO and co-founder, video intelligence, explains why we must think creatively about where ML can be applied best and be resourceful about utilising its abilities. 

Naturally, the building of machine-learning processes is a highly skilled area and bringing it in-house is an investment in people, skills, and time. But we’re witnessing the barriers to entry come crumbling down before our eyes, thanks to a combination of open-access APIs, plug-in and play AIs, and ML-driven software.

Each time you read a corporate tweet, chances are that it’s been scheduled for a time optimised by ML. If you’ve ever refined a headline using CoSchedule’s legendary Headline Analyzer tool, you’ll also have seen ML in action. These are examples of basic ML at work in mundane, yet widespread, social media applications. It’s just the beginning.

Kai Henniges, CEO & Co-Founder, video intelligence

The world of e-commerce is also revolutionising the way it targets. If you visit on online store homepage, it’s highly likely that you’re seeing products selected based on what you’re likely to buy. This is not just determined by your purchase history, but by your projected purchase future, determined from learned behaviour (see this example from Glossy). And if you begin to dig down into Amazon’s dynamic pricing mechanisms you’ll see some very smart ML tech at work.

Beyond social and retail, advertising technology is rapidly absorbing ML. One place we’ve all seen machine learning, and arguably the initial driver of the whole ad-tech market, is programmatic. The smartest programmatic buying models employ ML to optimise buying methods. In many cases, this has made the adops function massively scalable; media buys that would have taken hundreds of people can now be done by a few.

But, as with all AI, the combination of humans and machines remains the most potent. IBM’s Dr John Kelly convincingly argues this point; and it’s a crucial one as we consider how we work alongside machines going forward. Machine learning becomes really exciting when you and I come in. When anyone can consider a problem, and apply a machine to help solve it, the possibilities open up.

There are a number of brilliant free tools available that mean that digitally literate teams of all sizes (and budgets) can use ML to improve their audience understanding, contextual accuracy, and overall efficiency.

Google, Amazon, and IBM have suites of APIs that are available for anyone to use. Google’s Vision API ‘image content analysis’ platform is particularly powerful – and a lot of fun to use. Drop any image onto it and it will run an analysis based on billions of data points. It will tell you, with a displayed degree of accuracy, what is in the image and the likelihood that it contains unsafe content.

At vi, we’ve created tools by combining these tools and APIs successfully with our own tech. But the applications are limitless, it’s sure to prove to be the basis for many new products.

Another useful AI tool is IBM’s Watson, perhaps the best-known AI on the market. Watson’s Natural Language Processing (NLP) function is a world-class tool that can analyse content, relationship, and sentiment. Applying a tool like Watson to niche, specific ad topics can yield incredible results, and you’ll see analysis improve before your eyes. Without resorting to clichés, at this point, the only thing holding us back is our own imaginations.

Ad tech will continue to do what it does best – adapt and absorb. ML technology allows us to take giant leaps forward as we scale our problem-solving abilities. We must think creatively about where it can be applied best, and be resourceful about utilising its abilities. If our sector is to continue its rapid transformation, the future is in our hands.