Skip to main content

Three drivers for establishing and maintaining value with AI

artificial intelligence
(Image credit: Image source: Shutterstock/PHOTOCREO Michal Bednarek)

Working out how to determine and deliver value to the enterprise using AI is no mean feat. For the most part, success boils down to organizations embedding AI methodology into the core of their business process — integrating it into their data strategy, and also their holistic business model and processes. 

Naturally, driving value with AI requires planning — even after investing in the right technology, it takes a thoughtful and systematic approach. Here are three essential drivers that organizations must take into account to evolve the maturity of their AI systems and deliver business value.

Vision: Align AI efforts to strategic objectives

Even the greatest, largest, and most successful AI projects had to start somewhere - and so does a business when it comes to planning its overall AI objectives. A detailed evaluation on the current state of your organization’s AI implementations and overall maturity is crucial, as it will allow for realistic planning. Based on this evaluation, executives can start to determine the organization’s short-, medium-, and long-term visions for AI at scale.

Short-term AI goals might revolve around low-hanging fruit use cases across different lines of business. These initial projects serve to build trust in AI, test operational frameworks, and ensure the proper processes as well as technology are in place for loftier goals. Longer-term goals might include democratizing AI more widely, striving for Everyday AI where data access and impact aren’t limited just to data professionals, but infused into the day-to-day of the business. The bottom line is you need both the short-term wins and the longer-term vision to succeed in driving sustained value.

Of course, being successful with both short- and long-term AI vision requires buy-in from the top. Every leader in the organization should invest time in understanding AI and its potential on the business, asking questions like:

  • Does pursuing AI or advanced analytics pose any threats for the company?
  • What are the opportunities to use these technologies to improve existing processes?
  • How can they be used to generate new business opportunities?
  • What is the risk of not leveraging AI within the company or for some specific function? 

Thinking about these crucial points will help guide the company’s AI vision at all levels.

Talent: Building through hiring & upskilling

Arguably, the value of AI is only as great as the talent that underpins it, and leveraging a cross-profile workforce will offer a wide range of thoughts and perspectives needed to make some of the most complex AI implementations a success. Defining how new talent will make an impact within the business should be clearly mapped out from the outset, and for existing staff, individual plans to upskill around AI should always be in place. 

That said, having the best talent in place will only prove successful if placed within a cross-profile, collaborative environment (e.g., data scientists can work hand-in-hand with experts on the business side). That means ensuring people in all job roles across all levels have an active role in building business-impacting AI systems.

Ensuring that talent is well engaged with AI implementations is key to achieving Everyday AI and to AI value. This includes empowering non-technical colleagues to both understand and get hands-on with sandbox AI tools, to develop and upskill their own AI knowledge and take charge of their own confidence. Likewise, unifying the talent within the organization to create an AI ‘community of practice’ group can help all colleagues collaborate and create a diverse, AI-native workforce.

Systems: Invest in the right technology

As data and AI projects become ever more complex with increasingly more people involved in their development, choosing the right tools has become paramount. Over the years, companies may have experimented with various technologies for different stages of the data or AI project lifecycle, which has now led to disparate systems that make it difficult to track data lineage end to end. 

Overall, organizations run the risk of not having a total, holistic view over all AI processes, which in turn means difficulty achieving model explainability. Not only can this prove a problem for those internally who are using data modeling tools, but a business may find itself in the bad books of the AI auditors. For some, these auditors are internal only, but many countries are now placing certain regulations and standards in place that need to be adhered to.

Handoffs between different tools and platforms ultimately take time and also increase the risk of error. Likewise, many are missing out on the value that is driven via automation processes conducted between different steps in the data and AI lifecycle. Without clear orchestration, data errors, queries around fairness and quality of models, and more, can soon come to light. It’s a problem best avoided.

Many organizations — in their desire to experiment and onboard many AI and data tools out of curiosity — are actually bleeding value. Through every disparate system and process, more insights are lost along the way and true value is often missed through the most minor of exchanges. Another issue is that systems built in this way often hold no scalable potential, which of course will limit the insights and value generated as the organization outgrows its existing AI implementations.

To ensure value is maximized, businesses must operationalize their AI projects in line with all other infrastructure transformation projects currently underway. AI tools must be reviewed, implemented and supported like all others, and their entire lifecycle considered. Most importantly, adopting AI tools with a single ‘one-platform’ mindset will help in terms of scalability of resources in a responsible and effective way, closing the gap between disparate systems and minimizing value loss.

Closing comments

What’s clear is that the effectiveness of AI systems — and the value they create — begins way before any initial implementation. It all begins with the right planning, ensuring that people across the organization agree on short- and long-term goals (as well as the use cases to tackle at each level of maturity), and that talent is best positioned to support it via a culture that promotes contributions from all areas of the business. There must be clarity on who defines, validates, and approves AI models before they are implemented and scaled across a business, favoring collaboration at every stage.

Claire Gubian, VP Business Transformation, Dataiku

Claire Gubian is VP Business Transformation at Dataiku, the centralized data platform that moves businesses along their data journey from analytics at scale to enterprise AI.