AI Ascendant
Whether it’s using automated fraud detection at a bank, using pattern recognition to develop new treatments for disease at a pharmaceutical, or using Generative Adversarial Networks (GANs) to explain credit denial to applicants, Artificial Intelligence (AI) is on the rise throughout the professional world. The consensus is that this technology will reshape the way we do business, changing roles and responsibilities on every level. It’s true that AI has the power to reshape many of the tools and processes organisations rely on, but this transformation does not come without risk.
Risky Road Ahead
The source of most business risk within AI solutions comes from the data that both builds an AI’s capability (training data) and supports its ability to make current decisions (business data).
It’s estimated that about 75% of resumes in the USA are now run through an AI screening process. This process uses AI to filter out candidates based on selected characteristics, often trained on historical data to build a picture of what is, and what isn’t, suitable in a candidate. As you may guess, if a company traditionally hires white men in executive positions, then there is a danger that the AI could filter out candidates that don’t fit this description.
In fact, the Information Commissioners Office (ICO) in the UK has launched an investigation into these solutions, noting that discrimination of this kind is a violation of GDPR, making the regulatory issue very real for thousands of businesses.
Predictive AI is another area where bad data can cause critical issues within businesses. Zillow’s iBuying AI model, used to predict the prices of houses, suffered from a condition called “concept drift” in which the information feeding the AI Model doesn’t have correct data or doesn’t have up to date data. In this case, it’s estimated that two thirds of homes sold on Zillow within the program were undervalued by the model, causing Zillow to suffer significant financial losses.
Maybe the biggest risk to organisations is the great unknown. We are only at the start of our journey and yet AI is already a part of many applications, and operating systems such as Windows 12 are purported to be designed from the ground-up with Artificial Intelligence in mind. This means increased presence of AI in business will get harder and harder to avoid and, without appropriate planning and action, even harder to control.
Getting ready for the AI (R)evolution
There are several ways to get ready for the oncoming AI evolution in business. Unsurprisingly, most of them centre on improving the diet of data feeding AI. Begin to ready your organisation for AI opportunities by looking after the following:
Data Management Foundations
Ensure the way you gather, process, consume and destroy data is fit for purpose. This means starting with the basics and ensuring data is visible, its meaning is understood, its source is trustworthy, and its quality is protected. One way to ensure your data management foundations are well prepared is to engage a Framework. The Enterprise Data Management Association (EDMA), for example, leads the way with its Data Management Capability Assessment Model (DCAM). This model enables businesses to baseline their data management capabilities and provides valuable insights on how to achieve and refine your data management space to satisfy business objectives.
Data Standardisation
AI consumes data best when it is standardised. This gives AI algorithms an easy one-to-one understanding of the language of the business and can accelerate automation of services and mitigate hallucinations and other failures of generative AI tools, such as Microsoft Copilot. To standardise your data, it must first be catalogued and defined. This lets you know what data you have and what it means. Next, map the data to understand context, and to identify its source and any transformations. This enables you to separate same-named items that should be exclusive, while combining differently named items that mean the same thing.
This process also has the benefit of enabling savings through the simplification of your data landscape. Why pay twice for data you’re already getting?
Data Ethics Framework
As AI begins to take a hold of core business processes, the need for a Data Ethics Framework is more crucial than ever. A solid foundation in Data Ethics will help your organisation avoid preferential treatment due to race or gender and stop your AI from developing bad habits and attitudes by ensuring data is analysed for inappropriate bias before being fed into AI for training or analysis.
Awareness and Culture
Of course, all this ties back to people. At the centre of any business, the people who work there are going to have questions and fears about AI and may have some bad habits or perceptions that will hamper responsible implementation of AI solutions. Help your people and your AI solutions by ensuring AI implementations are tied to real business outcomes, by building an awareness of the AI solutions in use at your organisation and by developing a data culture of fluency and understanding.
Don't go at it alone
Artificial Intelligence views the world without history or understanding. It doesn’t understand “how we’ve always done it” or “this thing about working here”. Due to this, when it’s trained by internal parties, it accepts all the information it receives as being true, without a deeper analysis. Using a third party when preparing for AI can ensure you aren’t transferring biases or bad information into your AI and put controls and processes in place to ensure these dangers are identified and addressed before they cost millions to fix. Engaging with a trusted partner to help ready your organisation for AI before implementation is essential.
Conclusion
While the use of generative AI tools brings about significant benefits, it also poses certain risks. However, with the right measures in place, these risks can be effectively managed, enabling businesses to take advantage of everything the bright future of AI has to offer.
Need help readying your business for that bright future? Talk to us at Ortecha.
With years of experience helping businesses address data and analytics issues, Ortecha are the best partners on your journey toward AI readiness.
Stephen Gatchell
Partner & Head of AI Strategy at Ortecha
Sean Russell
Managing Principal & Head of AI Enablement at Ortecha
