Research shows uptake of generative AI in the workplace, but a majority of workers say they lack the resources to use the technology safely.
Salesforce revealed this week that 38% of UK workers already use or plan to use generative AI in their jobs. However, 62% of these workers say they lack the skills necessary to use these technologies accurately and safely.
The Salesforce study, which was conducted in partnership with polling company YouGov, queried 1,384 full-time UK employees across sales, service, marketing, and commerce. The research found that workers rank better service for customers and the potential to save up to 3.5 hours per week on tasks among the benefits of using generative AI. But the respondents also said they are wary of trust and security risks.
Output quality is an issue
Workers expressed concerns about the quality of generative AI outputs. Of these, 58% cited “bias” and 46% named “inaccuracies” as potential problems. In addition, 70% said a lack of human contextual knowledge was a flaw. 66% cited a lack of human creativity as illustrating the need for human input and control.
“Generative AI is the most important technological breakthrough of our lifetime, revolutionising how businesses interact with customers. But its potential will only be realised if we put trust and safety at the centre of this technology”, said Zahra Bahrolouloumi, CEO, Salesforce UK/I.
Varying levels of confidence in AI
67% of UK workers are concerned that their teams don’t have the skills to effectively and safely use generative AI. In addition, 69% feel their employers do not know how to get the most value out of these technologies.
77% of UK C-suite and 67% managing director-level respondents say they are confident. However, only 35% of senior managers and 29% of junior managers share this sentiment. Clearly, the discrepancy points to an overall disconnect between these two management layers.
Salesforce’s Bahrolouloumi believes the respondents are right to be concerned. “AI has the power to transform how work gets done, but it is only as good as the data it is trained on. Without high-quality, trusted data, it becomes ‘garbage in, garbage out'”, he said.