- Many AI models are not as effective as they are marketed to be, the report states
- 95% of the companies surveyed have seen very little impact of their LLMS
- Specialization is the key to the successful adoption of AI
A new research carried out by the Nanda of MIT initiative has affirmed that the vast majority of Genai initiatives trying to promote rapid income growth are “decreasing.”
Of the samples, 95% of the companies that implement generative are stagnating, “offering little or no measurable impact” on profits and losses.
It seems to be a game of all or nothing, since 5% of the companies that benefit from generative AI are exceptional: these are mainly, says the main author, the new companies led by young people aged 19 or 20, who have seen a revenue leap from zero to $ 20 million in a year ‘.
It seems that the key to success with AI models is specialization. The successful implementation is about choosing ‘a pain point’ and executing this well, and carefully associated with companies that use tools.
Specialized suppliers succeed around 67% of the time, but the models built internally succeed only around a third with the frequency. The highly regulated sectors such as the financial industry see that many organizations build their own AI systems, but research suggests that companies are much more prone to failure when they do.
When line managers are empowered to boost adoption, they see success because they can choose tools that can adapt over time.
The allocation is also important, since most Genai’s budgets are dedicated to sales and marketing, but the largest ROI was seen in Back -Office automation.
This is not the first time that research suggested that AI models do not work as they should. A significant number of companies has introduced layoffs of lower level workers and brought to AI systems, but more than half of the United Kingdom companies that replaced workers with AI regret their decision.
The tangible benefits of these models are increasingly difficult to find, and safety risks linked to the models are related to organizations, as well as AI models that make Esg objectives much more difficult to achieve.
