If your professional conclusion about AI is based primarily on bad prompting, chatbot-style usage, or poorly designed agent workflows, you may be dramatically underestimating both the technology and the people learning to use it properly.
And ultimately, productivity is measurable under proper methods.
Opinions can disagree. The numbers eventually cannot. Most research measuring productivity are flawed, since they don’t count for engineering methodology and system architecture.