"The near-instant content-generating powers of generative AI tools, such as ChatGPT and Gemini, might lead some people to think that making the most of emerging technology is as easy as writing a good prompt and clicking a button. However, with MIT recently reporting 95% failure rates, the evidence suggests that rolling out a successful AI initiative is much tougher than you'd think. Too many projects get stuck at the starting gate due to poor integration, prioritization, and cultural concerns."
"Most crucially, success is related to cadence. Push out too many updates rapidly, and the people who rely on your services will get left behind. "We learned that once we'd got to about the two-week rollout stage, our customers couldn't cope with changes too often. So, there's a limit to that speed of rollout, right?" Roth told ZDNET that her organization has an experiment-based approach to AI project development. After exploring 200 AI use cases, her organization has pushed 70 products live."
The near-instant content-generating powers of generative AI tools can create expectations that deployment is simple, but evidence shows high failure rates. Many AI projects stall because of poor integration, weak prioritization, and cultural resistance. Successful rollouts depend on timing, managed change, and linking investments to outcomes. Cadence matters: too-frequent updates overwhelm users, with a two-week rollout pace proving too rapid for some customers. An experiment-based development approach can help: exploring numerous use cases and pushing selected products live enables practical learning. Thomson Reuters developed legal AI products grounded in proprietary research databases to support professional workflows.
Read at ZDNET
Unable to calculate read time
Collection
[
|
...
]