Follow Datanami:
January 25, 2024

Key Insights from cnvrg.io GenAI and ML Report

via Shutterstock

It seems like every industry is racing toward AI, however, data shows that the adoption rates don’t match the GenAI hype. The results of the recently released ML Insider Survey by cnvrg.io, an Intel company, reveal that the majority of organizations are still in the research and testing phase for GenAI. 

The ML Insider survey by cnvrg.io, now in its third year, is an annual report on the latest trends and greatest challenges, and the latest strategies for building successful AI use cases.  This year’s survey included over 400 participants from a diverse set of organizations from various sectors and functions. 

The report shows that a high percentage (44 percent) of respondents believed GenAI to be extremely important or very important, however, only 10 percent have launched GenAI solutions to production in 2023. 

As GenAI is still relatively new for many organizations, they are still researching use cases (29 percent), building internal demos for the technology (23 percent), or developing pilot projects for select use cases (25 percent). 

“While still in early development, generative AI has been one of the most talked-about technologies of 2023. The survey suggests organizations may be hesitant to adopt GenAI due to the barriers they face when implementing LLMs,” said Markus Flierl, corporate vice president and general manager of Intel Cloud Services. 

Nearly half of the respondents found infrastructure the largest barrier to productionizing LLMs. However, Flierl is confident that with greater access to cost-effective infrastructure and services, such as those provided by Intel Developer Cloud and cnvrgo.io, GenAI adoption will rise. Other barriers to productionizing LLMs include monitoring (20 percent) and updating (17 percent). 

Deriving value from LLMs requires a specific set of skills and expertise, and the ML Survey respondents are aware of this. Eighty-four percent admitted that skills need improvement due to increased demand for GenAI adoption. 

A key insight from the ML survey is that the larger the company, the more difficult it is to execute a successful AI project. More than two-thirds of AI projects admit that they find it difficult to execute a successful AI project. 

“The 2023 ML Insider Survey shows that a majority of AI developers say lack of technical skills is slowing down their organization’s adoption of ML and Large Language Models, which creates pressure in a business world racing to implement GenAI capabilities. As an industry, we need to do everything we can to remove complexity and simplify tasks to make it easier for developers.” – Tony Mongkolsmai, Software Architect and Technical Evangelist, Intel.

(EstherQueen999/Shutterstock)

The ML Insider Survey results echo the findings of several other studies. A report by Deloitte last year showed that 42 percent of companies are experimenting with GenAI with only 15 percent actively deploying the technology into their business strategy. One in four respondents admitted to reading and talking about GenAI but said it is too early for them to make a decision on using GenAI in their companies. 

While the adoption rate might still be low, according to a KPMG study, executives expect GenAI to have a big impact in the near future. Nearly two-thirds of U.S. executives in the survey say that GenAI will have a high or extremely high impact in the next three to five years. 

A Predibase report highlighted the reluctance of enterprises to use commercial LLMs citing data privacy as chief concern. Enterprises are turning to customized LLMs for more accurate results. 

While GenAI technology continues to shift the industry, there is evidence that enterprises are slow to adopt it. However, as organizations move from experimenting to production by overcoming some of the key challenges, AI adoption is on the path to rise significantly in 2024.

Related Items  

Aporia Survey Indicates High Frequency of Malfunctions in AI Production Models

2024 GenAI Predictions: Part Deux

How Large Language Models and Humans Can Make Strategic Decisions Together

Datanami