Microsoft AI and Productivity Report - huge amounts of AI on the sly!

We know that Generative AI is already helping people be measurably more productive in their day-to-day jobs but the report gives real detail on how real-world workflows show a more complex productivity story compared to the now familiar lab studies.

In short, productivity gains vary by role, function, and organization. These adoption and utilisation differences influence AI's impact. Early studies suggest generative AI may affect the cognitive effort required for task completion.

Real-World Contexts 

We are starting to see its impact in Real-World Contexts and the good news is that the gains predicted by lab studies translate into significant impacts when AI is used for real work. Contextual factors, such as skill level or task type, are all important for AI's impact and AI's presence may change, not just productivity, but also the tasks people choose to do.

Sly use of AI

Survey of 31,000 knowledge workers showed widespread use of unsanctioned AI tools to save time.

One key finding is the widespread use of ‘unsanctioned’ AI tools among employees. 

• 78% used at least some AI tools NOT provided by their organisation 

• many employees therefore turn to AI to meet their needs

• 29% of respondents were classified as AI power users

o using it at least several times a week

o saving more than 30 minutes a day

• Power users had noticeably lower use of unsanctioned AI 

o 66% power user

o the non-power user average of 83%, p<.05)

This suggests that, if organisations want to reap productivity gains, they should sanction the sue of AI.

Copilot Usage in the Workplace Survey

Extended Copilot use correlated with perceived benefits, such as fewer meetings and increased job satisfaction. But differences in benefits were observed across roles, with customer service and sales reporting higher productivity and fulfilment.

• Developers see AI as helpful for automating routine tasks and writing documentation.

• Security Professionals - accuracy and speed for security tasks.

• Sales - licensing chatbot improved speed and accuracy in answering customer questions.

• Multilingual Contexts - helped non-native speakers achieve high accuracy in understanding meeting content.

Conclusion

Generative AI shows positive productivity effects in real-world settings, with variations by context. AI can reduce cognitive load and make tasks feel easier and less stressful and there is a huge amount of unsactioned ‘AI on the Sly’ use.