Learning the lessons from problematic AI implementations

AI is full of the promise of revolutionising industries and redefining business operations. But the rush to adopt the latest AI trends, without a strategic approach, has led to significant financial losses and reputational damage for several organisations.  

These are the hidden costs of chasing AI trends.  

So, in this post, we explore recent AI initiatives that didn’t quite meet the mark, discussing the fallout and how following a pragmatic AI approach might have yielded different outcomes. 

The delays and underwhelming launch were caused mostly as a result of an ‘all for one’ approach being taken for feature-rich and very complicated product. A fragmented organisational structure and leadership hesitancy compounded things further.  

Google lost market share in the emerging Generative AI space, diminished consumer trust in Gemini versus ChatGPT, and internal frustration amongst its employees. And the traditional forerunner of innovation has been caught playing catch up ever since.  

Pragmatic AI would have encouraged clear leadership, focused development, and thorough testing before launch. Prioritising quality over speed would have enabled Google to preserve market confidence, its position and consumer trust. 

Health setbacks for IBM Watson 

IBM's Watson Health division faced challenges in delivering on its promises. The collaboration with the MD Anderson Cancer Center, aimed at eradicating cancer, ended after spending $62 million without meeting its goals. The project failed due to inaccurate cancer treatment recommendations and integration challenges. 

Additionally, a  partnership with Manipal Hospitals in India was discontinued in December 2018 after difficulties in localising AI models to India's healthcare context, resulting in high error rates and the inability to scale. 

These setbacks raised wider questions about the feasibility of AI in healthcare and led to a re-evaluation of AI's role in clinical decision-making.  

Additionally, the Watson division of IBM faced significant financial losses and reputational damage, leading to a strategic shift and the sale of assets. 

Pragmatic AI would have started with a clear, focused use case, such as diagnosing a specific type of cancer rather than aiming for a day one catch-all. This would have enabled the AI to be trained and tested on a specific dataset, ensuring more accurate outcomes. 

By defining a manageable use case, agreeing and measuring key metrics, and incorporating real-time feedback loops, IBM could have refined Watson’s capabilities before scaling it, avoiding the costly setbacks it experienced. 

What ties McDonalds, Google and Watson together? 

Chasing AI trends without a strategic, pragmatic approach can lead to significant financial losses and reputational damage. By focusing on sustainable growth, clear objectives, and iterative development, organisations can harness AI's potential while mitigating risks. The lessons from these cases underscore the importance of a measured approach to AI adoption, prioritising long-term success over short-term hype.  

This is where Pragmatic AI comes in. The approach, which has been used by large government and regulated industries, helps organisations avoid the pitfalls of overambition and costly failures. Instead of chasing big trends, it ensures AI solutions are Simple, Affordable, Enabling and Measuring. It means value and refinement before scaling, saving time, money, and reputation. To learn how it can guide your AI journey and set you up for greater success, download our Pragmatic AI white paper.