Why is the ROI of AI out of alignment with reality?

Why are so many AI projects failing? Different analyst firms, consultancies, and integration firms cite over 70% of AI projects are failing to meet their objectives. Many of these projects are halted or otherwise terminated. But what are the reasons for these failures? Is the AI technology bad? Are the technologists inexperienced? Is there some systemic issue with artificial intelligence? When you have well-known AI experts such as Andrew Ng, Rodney Brooks, Yann Lecun, and others citing project failures certainly it can’t be due to lack of experience or unavailability of resources or tools. Something else must be going on. And it seems that something else is going on.

In our upcoming Lessons Learned from AI Project failures webinar, we cite many of these reasons, ranging from treating AI projects as application development projects instead of the data projects that they are, issues around insufficient data and poor quality data, focusing on high-risk “proof of concept” projects instead of lower-risk pilot projects and a number of other surprisingly mundane reasons that have nothing to do with the technology, the technologists, or the concept of AI itself.

We also dive deep into many of these reasons in a number of recent AI Today podcasts:

Despite the high visibility of AI project failures, there are countless number of project successes that manage to match AI capabilities to identified, practical business cases, have a strong grasp on the availability of data and manage scope to address data quality and real-world issues. Indeed, it’s possible to get to an 80% success rate instead of these failures we’re seeing.

Why is the ROI of AI out of alignment with reality? It doesn’t have to be this way. Check out the upcoming webinar and podcasts to gain insights on how to not be an AI failure statistic.