What generative AI can learn from the failures of IoT
If you can’t set a good example, at least be a horrible warning.
In the radiant glow of generative AI, it’s easy to forget that just over a decade ago, another technology – IoT – was similarly hyped. Combined with the burgeoning concept of ‘Platform as a Service’ (PaaS) architecture, IoT was, for example, poised to catapult GE into the league of top global software companies. By 2019, however, many digital assets related to Predix – the ambitious project that was to make this leap possible – were sold off by GE.
Numerous articles have detailed the Predix project’s failure, so we will not reiterate the tale here. Instead, we wish to draw attention to the lessons learned from its implementation.
The story of GE’s failed IoT attempt serves as a stark reminder of the challenges that can surface during the implementation of transformative technologies – a lesson that is especially pertinent to the burgeoning field of generative AI.
In 2014, GE launched Predix, the world’s first industrial cloud-based platform designed to store and analyze machine data on a large scale. It was designed to interpret signals from and automate industrial equipment, regardless of the manufacturer.
Yet the use of proprietary and outdated technology proved to be a roadblock. This issue of interoperability, particularly concerning data, can also impact AI projects. Even when data exists, is it truly accessible? Transformation may be necessary to format the data correctly. Once obtained, the question remains: is the data platform reliable enough to function consistently?
The world seems swept up in the promise of generative AI today, much like the IoT revolution before it.
While concerns about the cultural implications of AI are frequently discussed, there is likely a silent majority genuinely fearful of AI’s potential to replace human roles across industries. The challenging task of organizational change management cannot be underestimated. Successful projects will have to demonstrate the advantages of AI, incorporating enhancements methodically into everyday workflows.
The technical challenge of retrofitting legacy industrial machinery for Predix is echoed in the complexities of integrating AI into existing IT infrastructures. Generative AI projects often demand considerable computational resources, and specific environments may be required to run algorithms effectively.
Existing systems may not be equipped to handle such complexity, necessitating significant IT upgrades. This scenario can add to both the timeline and the budget of the AI project.
If your generative AI project alters how customers interact with your products or services, their readiness and ability to adapt need consideration. This becomes particularly challenging if customers are accustomed to legacy systems.
It’s crucial to design the AI experience with the customer at heart, offer ample support during the transition, and clearly communicate the benefits the new AI system will bring to their interaction with your products or services.
The journey of successfully introducing any large-scale technology, such as generative AI, requires a balance of technical, human, and business considerations. Inspired by the lessons from GE’s experience, here are a set of practical strategies for your business:
- Start with an interoperability audit and a cultural readiness assessment to identify potential conflicts or gaps. Invest in comprehensive training to help employees navigate the new technology landscape.
- Evaluate the capacity of your existing IT infrastructure to meet the demands of AI and consider an iterative approach to implementation, allowing for manageable testing and gradual scale-up.
- Always put the customer at the heart of your design process, involving them through surveys, beta testing, or user experience research. Maintain transparent communication about the benefits of the AI system and the changes it will bring, providing ample support during the transition phase.
- Ensure you learn from the cautionary tales of the past, using them as stepping stones rather than stumbling blocks on your path to technological transformation.
If you’re able to achieve all of these things, generative AI projects can become clear examples of successful integration rather than stark warnings of failure.