If you had listened to some of the rhetoric from the AI industry, you might get the impression that there are no limits to what AI can help an organization achieve. This might well be true – at Squirro, we certainly believe that AI can be transformative - but it isn't borne out in the number of organizations using AI in a meaningful way.
Many studies and research papers over the past few years have suggested that the volume of organizations actually using AI is in the low double digits. Furthermore, the use cases that such organizations put forward are rarely that concrete. Instead of AI being used to powerful effect in some aspect of day-to-day operations, it usually appears to be used by a small team of data scientists.
What is behind this disconnect? If even half the AI hype would be material then surely every organization would want to adopt this technology rather sooner than later? For Dorian, it is a matter of practical readiness of the technology, and trust.
While AI as a field of research is not novel, its adoption in the enterprise is still in its infancy. While many AI models work marvelously in a lab or test environment, when put into production there’s room for improvement.
It’s not too dissimilar to when computers hit the business world back in the late 70s and early 80s. Economist Robert Solow famously said in 1987 that the computer age was everywhere except for productivity statistics.
The reason was twofold: Many of the then clunky machines were simply not ready to take over even such mundane tasks as payroll processing. There were too many errors and too many issues. The result was that the organization could not reduce the manual payroll work. Not only did it not make any cost savings, but it also had to spend significantly to iron out the kinks in these early implementations.
This rather practical problem was overshadowed by a second issue: The organization had to get used to the new process and had to confer an additional element into this new setup: Trust. Trust that the computed payroll really corresponds to what should be paid.
The same is happening today.
When AI does work in an organization, it is usually invisible. People can’t see it, but they must trust the outcomes that it delivers. This might be improved deal flow, deeper understanding of customers and their markets, finding information quicker and in context, or something else entirely. But this trust is vital to the increased adoption of AI.
In his keynote address, Dorian will be talking through these issues and offering pointers on how an organization can begin building trust in AI. Much of the program at AI Week 2021 is also focused on this. The use cases will demonstrate the value of AI and the tangible business benefits it provides. Presenters will also provide guidance as to how best that value can be communicated to the business.
The executive masterclasses will also feature ways in which to build trust in AI. The business masterclasses explain AI and unstructured data using a guided demonstration aligned with specific business challenges. The IT masterclasses will go through the steps of setting up an ML solution with a hands-on, no-code approach to data preparation, model creation, and dashboard configuration.
For AI to deliver on its vast potential, trust is essential.