Pump Those Numbers Up, Those Are Rookie Numbers

This month has been all about productivity in Australia. Traffyk.ai's CEO Kate Abrahams and Director of Business Development Catriona Booth attended the Productivity Parliamentary Showcase and were involved in interesting discussions with MPs on the subject.

In this article, I want to explore how I see AI fits into the picture.

Productivity is mathematically defined as: (Output)/(Unit of input)

At a macroscale this is usually: (GDP)/(total labour hours)

At a microscale it could be:(# Linkedin posts)/(hours worked)

It is a no-brainer that microscale definitions of productivity can be boosted by AI. For example, studies of developers using coding assistants can be quoted to measure productivity increases of up to 40% for an individual.  There are caveats of course: The benefits of assistants diminish with increasing seniority/experience. It has even been noted that top tier developers experience a decrease in productivity when trying to use a coding assistant. This is to say nothing of the increase in debugging time and setbacks when an AI agent deletes your entire codebase (Replit).

Ignoring these caveats, how far does a 40% increase in individual productivity for mid-tier day-to-day task get you on a macroscale? So far, nowhere, everyone’s still waiting for that boom. I fear that purely relying on assistants or half-baked agents, productivity gains will be lost in the translation from micro to macro.

My feeling is that real productive power is going to be innovation rather than assisting. Fundamental shifts in ways of working whereby a hybrid AI and human workflow is created. What I’m seeing at the moment is that there is still a vast sea of untapped or fragmented data that exists in the business world. People simply do not have enough time in their day among the meetings, emails on top of their daily workload to expand past their siloed view and get a handle on the bigger picture. Leveraging AI to do what it’s good at and compress all this untapped information into simple human-readable formats will empower us to make better strategic business decisions. This in turn will better fulfill the needs of employees, customers, regulators and shareholders. This is where AI GDP gains are realised.

For this, pioneering work needs to be done. It is not a simple plug and play. Engagement with human experts in their domain, deep thought into the current problems, and critical analysis of blind spots/inefficiencies in current ways of working need to be done in all sectors. To unlock this, policies at the government level need to be made to promote a deeper look into AI adoption rather than a simple off-the-shelf purchase.