Dario Sarmiento
IT Service Delivery Director | Managed Services | Complex IT Program Management | IT Architecture | Cloud Computing | Cloud Migration | Service Delivery Management.
The growing frustration with the AI hype
Last month of March, S&P Global Market Intelligence, released a report about current use of Generative AI Technologies in the enterprise. This same report was used by The Economist last week, as base of an analysis that concludes that there is a feeling of disappointment among CEOs and business leaders on the pace and actual value realization of the investments that they have made in this. Numbers shows that 60% of companies has implemented some GenAI technology, but 42% of the initiatives are abandoned before reaching production. The report clearly points to Cost, Data Privacy and Security concerns, as the major roadblocks, while the interpretation from The Economist, is that there are difficulties getting the right expertise and hardship to implement on top of legacy IT systems, so all this fueling risk aversion that is curbing the enthusiasm of executive adopters.
One consequence of this current state of things, is that there is not much trustable Market Intelligence data about the actual realization of value from the deployment of the solutions, even for the companies claiming they have already fully integrated AI workflows in their organizations, as they still are calibrating / assessing the long term impact on any of their organizational objectives coming from the use of the Technology. As most of the current successful cases of AI implementation are heavily advertised/achieved by the Hyperscalers, or their backed-up companies and partners, any claim from vendors about the ROI of AI and concreate measurements of impact should be taken with caution, and not to be the only criteria of adoption.
The correct approach to the AI adoption
AI should not be implemented as a Technology-focused initiative only. While the enthusiasm from the progress that is happening in the field on a daily basis, create a compelling reason and urgency on the business leaders to implement it before their competitors, we should have those considerations below in place.
- Start small with POCs for specific areas in which the potential to solution a problem with use of AI is clearly visible and relevant, to identify and prioritize the initiatives that would deserve to develop a formal and measurable business case.
- Once decided on the projects to invest in, and a proper business case is done, deployment of AI in production is a Digital Transformation program that is sharing the characteristics and challenges of similar Big Enterprise IT implementations. ( Like Cloud migration, ERP implementation, IT Service Outsourcing, Shared Services, etc). How the deployment affects the organization and people, (OCM), the processes and workflows, and the technology dependencies, are things that should be managed with a governance layer of a formal Program management practice.
- There is no AI initiative that will work if the data needed to make it functional is not fully available and accurate, and if the security and compliance requirements are not addressed. Having an updated Enterprise Architecture according with the size and maturity of the organization is a must to have baselines for sizing, evaluation and measurement of progress and success of the AI initiatives, and to document changes needed on Applications, Data, Infrastructure and business processes.
- In our view, no AI business case should be measured in terms of reduction of labor and H/C, but in empower and assist the human worker, so they can make better use of their time and talent and improve efficiency and efficacy of the processes they run, and as a result, they are enablers for growth.
AI can and should work, but executives should look beyond the “hype”, and look for the right group of people , vendors and methods that are needed to help them on creating enterprise-grade solutions that brings measurable benefits.
Reference: Welcome to the AI trough of disillusionment
