GenAI edges closer to enterprise maturity

While many organisations are piloting generative AI applications, few have mature production roll-outs, according to a survey of Computer Weekly readers worldwide

IT leaders are prioritising generative artificial intelligence (GenAI) projects that improve productivity or operational efficiency as the key benefits for early deployments of the technology, according to research by analyst Enterprise Strategy Group (ESG).

Although the majority (70%) of organisations have increased their investment in GenAI over the past 12 months, less than 10% of the IT decision-makers polled in a recent survey say they have deployed GenAI applications in a live working environment. Of the 832 IT leaders surveyed worldwide by ESG for its State of the generative AI market report, only 8% consider their GenAI deployments as being in "mature production".

However, while this figure is low, it has doubled in the last year and the proportion of organisations saying they are in early production has jumped by over a third (36%), which means that over a quarter of the IT executives surveyed (27%) work in organisations where AI is either in early or mature production. Another big increase came among those organisations running pilots and proof of concepts, where ESG found 22% growth in the number of firms piloting GenAI applications compared to 2023.

ESG is a subsidiary of Computer Weekly publisher, TechTarget - and as such, the survey included IT leaders from Computer Weekly's global readership.

GenAI empowers IT

The IT department itself appears to be the biggest beneficiary of AI deployments to date.

On average, among the IT decision-makers polled, GenAI is being used in 3.5 application areas. The top application for AI, according to Enterprise Strategy Group, is in software development, where 41% of respondents say GenAI is being used. This represents a 7% increase over 2023.

PwC has separately identified a number of use cases for GenAI in software development, which include automatic generation of test scripts, granular troubleshooting, code reviews and code completion plus automatic documentation generation. PwC believes skilled users will soon be able to instruct GenAI to generate high-quality artefacts for user stories, acceptance criteria, test cases, documentation and generate APIs automatically too. It believes GenAI will eventually augment work at every stage of the agile software development lifecycle.

Beyond software development, other areas that saw the largest increase in GenAI usage since 2023 in ESG's survey, include research, IT operations (7% growth compared to 2023)  and cyber security (6% growth compared to 2023). 

The broader use of AI to automate manual-intensive tasks in IT operations - known as AIops - is seen by many as an approach to tackling an evermore complex IT environment. For instance, energy provider EDF is deploying Dynatrace’s AI-driven automation technology to help its IT team optimise cloud operations and deliver reliable and secure customer experiences. The company sees Dynatrace’s AI monitoring offering as a way to identify potential inefficiencies in its tech infrastructure, to help the IT team remediate downtime quickly and improve customer self-service.

Looking at the role GenAI plays in cyber security, according to the authors of a paper published by the Alan Turing Institute’s Centre for Emerging Technologies, the technology poses new cyber risks and opportunities. Principal research engineer Sarah Mercer, and Tim Watson who is the Institute’s director for science and innovation, defence and national security, said there are various ways GenAI could disrupt the cyber security landscape.

“While GenAI can exacerbate existing risks with respect to the speed and scale of reconnaissance, social engineering, and spearphishing, the current impact of its code-generation abilities demonstrates a lesser effect on the attack landscape,” they said.

However, Mercer and Watson believe that current GenAI systems offer unique strengths, particularly in pattern recognition and natural language processing, where they are able to draw on extensive training data and offer multimodal capabilities. “Targeted application of these abilities to enhance state-of-the-art systems could significantly elevate existing technologies, for both cyber threat and cyber defence,” they added.

Enterprise GenAI maturity 

Those organisations that say they have mature AI deployments are charging ahead in terms of the proportion using GenAI for software development (10%), IT operations (15%), and finance (11%), when compared against the results for all organisations globally. Mature AI organisations also appear to be more likely to deploy generative AI to help manage cloud infrastructure costs.

The fact that GenAI can be trained on public internet data lowers the cost of deploying a useful AI application quickly. But this means everyone has access to the same large language model (LLM), which is trained on the same data. This quickly erodes any competitive advantage that may be gained from a GenAI deployment.

The survey results from Enterprise Strategy Group suggest that IT decision-makers not only recognise the limitations of deploying public GenAI models, but also why it is important to augment such models with their own data, or even use internal data to train their own LLM that can meet their specific business requirements.

While vast public datasets have helped to establish GenAI, giving people immediate access to the power of the technology, ESG reported that 84% of the IT leaders polled believe it is important their organisation uses its own data to support GenAI initiatives.

Over three-quarters (77%) of respondents say it is important to train their own GenAI models. This is reflected in the proportion of organisations using more than one LLM for generative AI deployment. In fact, two-thirds of IT leaders surveyed by Enterprise Strategy Group say they use two or more LLMs and almost one in 10 (9%) are using five or more.

Read more about GenAI

  • Many organisations are testing out uses for generative AI, but how are they getting on? We speak to five early adopters to find out the lessons learned so far.
  • Employees are using GenAI primarily for spelling and grammar checking, while business chiefs are using it for data analysis.
  • Deloitte survey shows business and IT leaders are optimistic about GenAI, but academic researchers warn of AI training time bom.

Read more on Artificial intelligence, automation and robotics

CIO
Security
Networking
Data Center
Data Management
Close