Research from American software company Cloudera shows that more than eight out of ten American companies are concerned about sharing data with third parties when using generative AI. This is primarily about training and refining the models.

The American software company Cloudera conducted a survey on generative AI among five hundred American companies. It found that 84 percent of respondents have concerns about sharing data with third parties when training and refining models. It is noted that it is mainly perception that promotes this feeling.

Applications

This percentage can also be put into perspective. Of the five hundred respondents, just over half are currently working with generative AI. Another 36 percent will not be in the early stages of implementation until next year.

The five highest percentages in terms of applications were:

• Customer communication (chatbots and co.) – 55 percent
• Product development – ​​44 percent
• Concept development – ​​44 percent
• Data analysis – 34 percent
• Software development – ​​32 percent
• Automation – 28 percent

According to Cloudera, ChatGPT has certainly accelerated AI innovation.

stumbling blocks

Organizations are particularly concerned about models that are not trained in the business context. This can lead to incorrect answers. In this context, the threat to training models posed by irrelevant public data is also a problem.

Other pain points that concerned the 84 percent were:

• Trust
• observance
• permit
• Intellectual property

According to the study, companies benefit greatly from successfully creating trustworthy and secure data sources. This contributes to more reliable results in their AI applications.

Cloudera announced a partnership with AWS in September to collaborate on data management. Moving large amounts of data has been one of the company’s main focuses for some time.