Artificial intelligence gets into auditing, what’s next?

Earlier this week, Big Four accounting firm KPMG LLP announced that it will be teaming up with IBM’s Watson AI (artificial intelligence) unit to automate some auditing functions. While KPMG certainly isn’t the only accounting firm making a big investment in new technology, it is the first to test AI or machine learning as the brains behind auditing tools.

Google and Amazon, among others, have made massive AI investments in recent years, which has fueled excitement about its commercial applications. And AI as a concept has been around for at least a century in the form of science fiction (and non-fiction).

But it’s not yet clear which industry will be the next to embrace this old/new technology. In my opinion, machine learning will fuel innovation in every industry, and will likely be adopted first by companies that are responsible for dealing with large data sets. In order to properly adopt AI, however, and full reap the benefits, companies have to turn to good old-fashioned coding.

That’s right: software development will be crucial for achieving success with AI. Programmers, rejoice — and start sharpening your skills.

Three major areas of software development will be key in ensuring the success of AI integration at any company.

Data integration: Integrating key data from multiple sources will play an important role in preparing for the integration of AI into applications and systems within the enterprise.Having the necessary resources in one central location will significantly eliminate the risk of errors, and establish a clear path for machine learning. For example, who will provide the electronic transactions KPMG auditors are going to need to feed into IBM’s Watson models? Software developers. This means software developers will be probably as important as bookkeepers to Finance Departments.

Application modernization: Regardless of how current your company’s software programs are, they will require some type of update in order to enable the integration of machine learning functionality into existing products. Rather than starting a gut renovation and slowing down existing operations (or bringing them to a complete halt), companies will have to look for less intensive updates that modernize existing software.

The best way to tackle this challenge is to make tweaks and updates on a regular basis to prevent more work needed down the road. Think of it as routine maintenance and the occasional replacement of parts that take wear and tear in a vehicle. Small changes over time add up to a much healthier machine over time. Microservices architecture has showed us that we do not need to create massive software systems to deploy our business solutions; we can create various microservices that together provide the business with the needed functionality. Are your current systems ready to interact with microservices? Which current functionality can be modernized into a microservice?

Employee education: Even the most talented and future-forward technology staff needs to understand machine learning and its implications on every aspect of the technology stack. This is particularly true for software developers and project managers who oversee sprints and other development cycles that can be interrupted by a transition to a new technology, or even enhanced by the addition of new systems. Tackling questions early, and providing resources often, will be key to helping your teams get on board effectively. Online courses by Coursera, Udemy, or Stanford Online can provided the needed education and training.

Across the board, machine learning is an exciting new technology that can have far-reaching benefits for companies in virtually any industry, but the key to harnessing the potential of artificial intelligence will be to ensure that your systems — and your people — are adequately prepared.