Industry
Manufacturing
Future-Ready Manufacturing: Partnering in your journey to a sustainable manufacturing enterprise
Highlights
- Gen-AI adoption has been slower due to multiple factors such as concerns around data security, non-availability of benefit realization framework or a ROI calculator, technology complexity, etc.
- A proper use of selection framework during use case prioritization that considers all aspects like kind of data to be used, impact of use case, risk associated with adoption, and laws related to use cases could help move towards faster realization of outcomes.
- A structured approach to adopting GenAI, with stakeholder involvement and alignment on the goals and benefits, is critical for faster outcomes.
On this page
The GenAI Saga
Generative AI has the potential to revolutionize the way manufacturing organizations operate businesses.
Improved accuracy, enhanced customer experiences and increased employee productivity are some of the benefits that Gen AI offers. However, the adoption of such a powerful technology has been slower due to multiple factors like concerns around data security, non-availability of benefit realization management framework or ROI calculator, technology complexity, etc.
The adoption of GenAI projects has been a top priority for many organizations, but the journey to production has been challenging. According to the latest survey by Gartner, 30% of Gen AI projects could be dropped by the end of 2025. In this blog, we will discuss the reasons behind the tardiness of GenAI projects and propose solutions to overcome these challenges.
GenAI Adoption
Many CXOs and directors have mandated the adoption of GenAI, and it is a part of their goal sheet.
However, finding the right use case, quantifying the ROI, and getting buy-in from all stakeholders have been impeding factors for a top-down approach. On the other hand, a bottom-up approach where different units work on creating Proof of Concepts (POCs) for their areas has worked well until productionization. During productionization, different teams, such as application teams, security teams, and product owners, need to approve the solution, thus leading to slow clearances.
We have found that most customers do not have an AI and GenAI Governance Office to streamline the process. Having an AI Governance Office could help in streamlining the process from selection to productionization, regardless of the approach. The Governance Office will also ensure synergy between different business units and leverage each other's use cases as much as possible. Immediate advantages include getting business approvals, security clearance, data from different business units to train models, and minimizing redundancy of work.
Fastest Evolving Technology
GenAI has been the fastest evolving technology, with new models, players, modalities, and evolving laws.
This makes it difficult for an enterprise to ascertain that the current investment may not become obsolete, and there may be better options available. For example, we completed tagging the 30K apps of a telecom major using traditional AI, which took a year. Now, with the advent of GenAI, this task could have been completed in a couple of months. To mitigate the ‘loss’, understanding the reusability components like data, test results, algorithms while designing use cases itself is crucial.
Apart from cutting-edge technology, evolving laws can cause slowness in adopting GenAI. While Europe has laid out a risk-based approach towards AI productionization, there are AI laws from other geographies which are continuously evolving.
This could also lead to a risk averse approach by providers likes of OpenAI, Meta, Google etc. For e.g. Meta recently announced that it will withhold all its multimodal models to be deployed in the European Union (EU) due to unpredictable nature of European Regulatory Environment.
A proper use of selection framework during use case prioritization that considers all aspects like kind of data to be used, impact of use case, risk associated with adoption, and laws related to use case could help move towards faster realization of outcomes.
Upskilling New Roles
Involving people with AI skills has always been a challenge. GenAI brings a new dimension in complexity as there are roles that did not exist earlier.
Creating a new role, such as a CXO for AI or GenAI, to drive AI or GenAI within the organization has been one of the trends off late. However, according to a latest LinkedIn survey, only 13% of organizations have an AI executive role defined since 2022. Defining a charter for such a role requires technology to be mature and stable.
Other roles like prompt engineer, LLMOps engineer at execution level also require a thorough understanding of their roles and responsibilities.
For example, for one of our customers – a leading manufacturer of high-speed trains, we were trying to build use cases around legal, HR, and application development using GenAI over a year ago. While we understood that a prompt engineer is required, we were unclear of their role in application development and who aptly qualifies to be a prompt engineer - an AI Engineer, a Test Engineer, or a BA.
Many of the organizations have also started upskilling their existing workforce to take up AI and Generative AI roles such as GenAI developer which have high demands. Moreover, even after upskilling, there could be issues like the customer's interest in seeing results faster than a beginner in the technology could provide. This could result in frustration and a lack of confidence in the technology, which could hinder its adoption in production.
To address this challenge, companies must have a well-planned and structured upskilling program that addresses the specific needs of their workforce. They should identify the skills gaps and provide training programs that are tailored to the specific needs of their employees.
Finally, companies must also manage customer expectations by communicating the capabilities and limitations of the technology clearly. They should be transparent about the level of expertise of their employees and provide realistic timelines for delivering results. By doing so, they can build trust with their customers and ensure the successful adoption of Generative AI in production.
In conclusion, addressing these challenges can help move GenAI projects to production and realize the benefits of this technology. A structured approach to adopting GenAI, with all stakeholders involved and aligned on the goals and benefits, is critical for faster outcome.
References
Meta pulls plug on release of advanced AI model in EU | Meta | The Guardian
