Machine learning is more than just an algorithm.

Machine learning is more than just an algorithm. Source: Shutterstock

How do companies actually get started with AI?

ARTIFICIAL INTELLIGENCE (AI) can lend a hand to many of today’s businesses.

Whether it’s a retailer who wants to automate how inventory is recorded into its e-commerce platform or a team of plumbers who want to know how best to allocate work amongst themselves in order to optimize resources and maximize revenues, AI has a solution.

However, in order to get started with AI, companies need data — to train the algorithms and ensure it’s able to perform the task accurately.

Imagine the retailer rushing into building the AI tool without enough ‘clean’ data at hand. The predictions that the resulting solution makes are bound to be faulty. The same will be the case for the team of plumbers. To be effective (and useful), AI needs good data to analyze and learn from.

Now, there are two ways to clean data: Automatically and manually. And there’s a use for both.

A lot of businesses need petabytes of structured data to be combed through, standardized, parsed, and managed. That’s what intelligent platforms like Datalogue, Data Robot, and even Tableau can help with.

Bot powered platforms are a cost-efficient way to get started with AI. For the team of plumbers, this is probably the best solution.

However, when your needs are more complicated, a manual approach is ideal. That’s where companies like Supahands, Clarifai, Google’s CloudML come into the picture.

Each, of course, has its own unique selling point, but they all essentially help clean non-standard data for businesses. Some, like Supahands, have a more hands-on approach, while others, like Google’s Cloud ML have a more algorithmic approach.

Getting complicated AI projects off the ground

Businesses are always in a rush to get things done. It’s the same with AI projects — and for good reason. Beating the competition to building and leveraging AI can provide a distinct competitive advantage.

To get started, data scientists need good data to work with.

Supahands, who have more than 2,200 contract workers available on demand, can provide clean data for a typical mid-sized project in as little as 2 months.

According to the company’s Sales Head Greg Meehan, Supahands is currently involved with projects in the retail industry and is expecting to grow rapidly over the next few years given the rising demand for AI.

Supahands’ Engineer Mohsen Saghafi told Tech Wire Asia that the company itself uses AI to optimize client projects. It’s system, built on Amazon Web Services to keeps costs low and maintenance issues at bay, helps identify the right people for the job from a pool of available contract workers.

Clarifai and Google’s CloudML too, can help ‘wash’ data, tag it appropriately, and make sure it’s ready to be fed to the machine learning or AI system.

Google’s Cloud AutoML technology, for example, is helping Disney build vision models to annotate its products with Disney characters, product categories, and colors.

The annotations are then integrated into its search engine to enhance its guest experience through more relevant search results, expedited discovery, and product recommendations on shopDisney.

AI is really exciting and useful, and the value is immediately apparent when you read an example or case study, but before getting started, businesses need to understand the importance of cleaning data.

Whether it is to improve the loyalty program, reduce customer churn, or inventory management for a retailer, cleaning data to train the algorithm is where it all starts.