The rise of AI in enterprise software

As large language models become embedded in enterprise software, CIOs will likely face a brick wall of resistance from internal stakeholders concerned about sharing confidential data and data compliance regulations like GDPR.

Generative AI systems and large language models like ChatGPT, rely on processing vast swathes of training data, often collected from the public cloud.

Many heads of IT and IT security are assessing the privacy and security implications for their organisations.

A new BlackBerry’s study, conducted in June and July by OnePoll,  based on a survey of 2,000 IT decision-makers, found that two-third (66%) of respondents have banned or are considering banning ChatGPT.

Computer Weekly previously discussed how smaller businesses are at a disadvantage in the world of large language models, as their datasets may not span a wide enough range to provide robust training for an AI system. There will clearly be opportunities to purchase commercial LLMs that have been trained on much larger datasets. The challenge is then how to combine multiple data sources in a way that offers decision-makers meaningful, auditable and accurate insights.

Larger organisations may well have diverse enough datasets to build their own data models internally and the expertise and resources to enhance these with controlled access to public data stores. But some recent changes at SAP may mark the beginning of a trend in enterprise software, taking the build or buy debate into the AI era.