Where’s the ROAI?

As the world becomes increasingly reliant on technology and driven by data, the excitement about artificial intelligence (AI) solutions continues to skyrocket. Corporate boardrooms are abuzz with discussions about exploring the possibilities of AI, and massive amounts of capital are being funneled into building AI infrastructure. Companies such Nvidia, Corestreet, and OpenAI are experiencing exponential revenue growth – and valuations, driven by this immense interest.

However, one can’t help but wonder: how long can this frenzy continue? The saying goes, “The market can remain irrational longer than you can remain solvent.” Sooner or later, the market will demand tangible financial returns, and real-world use cases with quantifiable Return on Investment on AI (ROAI) will need to emerge to sustain this level of investment.

I’m not arguing that AI is an unsustainable bubble. As someone who has spent a significant part of my career in data and analytics, I’m an optimist and a believer in the transformative potential that can be achieved. I’ve witnessed firsthand any number of incredible things that intelligent applications of data – all of which now get put under the general-purpose label of “AI” – can accomplish and am enthusiastic about the opportunities it presents in the near and distant future. I fully believe that AI offers a prime example of Amara’s Law in action—we may overestimate its short-term impact but profoundly underestimate its long-term significance.

However, the reality on the ground is that most corporate executives, after expressing excitement about AI’s possibilities, struggle to identify specific examples where those possibilities have been realized. The challenge lies in determining where and why to invest significant resources in AI without a clear understanding of the Return on AI – or “ROAI”.

Many – including myself – have pointed out that “AI” isn’t a new concept – but clearly the current AI frenzy was ignited by the introduction of ChatGPT. ChatGPT was a watershed moment for those of us who had long been excited about AI’s potential, and it was thrilling to see widespread enthusiasm ignite among business executives who, if asked previously about “AI” may have responded with a quizzical expression at best, or a list of underperforming initiatives that may have been conducted in the past.

While the excitement surrounding ChatGPT and AI is palpable, there are two critical aspects to consider:

1. ChatGPT – and chatbots like it – are still fundamentally demos: Despite their impressive capabilities to interact with people, the business utility and value of chatbots still aren’t always readily apparent. Many popular “demos” such as MidJourney and Sora face similar challenges in driving more “wow that’s cool” responses than “I can see how to make (or save) money with that”.
2. AI is Heavily Subsidized: The perception that AI is affordable is misleading. Tech giants like Microsoft, Amazon, Meta, Alphabet/Google, and OpenAI are investing colossal sums of capital in their infrastructures, making AI chatbots and services appear more accessible than they will be once their full costs are paid by the customers using them. Without these massive subsidies, AI chatbots would be both less impressive and significantly more expensive.

A recent article in The Washington Post aptly titled, “The AI hype bubble is deflating. Now comes the hard part,” highlights the challenge of turning AI hype into financial returns, not only for vendors but also for end users. But while we may be slipping into the “Trough of Disillusionment” for AI – I am a firm believer that the “Plateau of Productivity” lies ahead.

How do we get there? The reality is that the Return on AI still depends on traditional metrics: cost savings and revenue generation. While AI holds immense promise, it’s essential to recognize that widespread adoption of financially impactful AI use cases is still in its early stages. The key to success is building momentum internally – and for technology executives to partner with their business counterparts to build out and demonstrate business-specific use cases that drive ROAI.

That is starting to happen – applications from Customer Service to Content Marketing are becoming visible, and more industry-specific examples are beginning to emerge as well. As more use cases emerge where AI can make a tangible difference in saving costs and driving growth, the “hard part” of monetizing AI will gradually become easier – and the “ROAI” will become more apparent.