Key findings:
-
AI has the potential to boost productivity, but the actual benefits have not fully materialised
-
Like an employee, AI needs trust-building, clear expectations and small projects to demonstrate benefits
-
Accurate and transparent data is crucial for AI performance – poor data inputs can lead to high ‘hallucination’ rates
Artificial intelligence (AI) has been heralded as the saviour of productivity, yet it appears the promised benefits have stalled.
On the face of it, AI is the ideal colleague – available on demand, always insightful, never tires. It can free up time usually spent on menial tasks, and allow more time for strategic and creative thinking that is engaging for human employees and, importantly, more productive.
The promise of AI in boosting productivity
To say that AI has been embraced is an understatement. The technology has been enthusiastically taken up in the last two years, as ChatGPT, Co-Pilot and similar tools have come into the mainstream, with reports the global AI industry will be valued at US$1.81 trillion (AU$2.7 trillion) by 2030.
The potential is gobsmacking. Research group Nielsen Norman analysed studies that demonstrated business professionals drew on AI to write 59% more business documents per hour, while programmers could code 126% more projects per week.
Yet that number can be deceiving. AI’s potential workmates have decidedly mixed feelings about its arrival into their workplaces, and employers haven’t seen the promised productivity rewards materialise.
Quite remarkably, many business leaders are alarmed about productivity trends.
The latest Business Radar report released by Pitcher Partners, capturing the sentiments of Australia’s middle market business leaders, shows that 52% are very or extremely concerned about productivity.
In Australia, labour productivity in 2022-23 fell by 3.7% as output growth failed to keep pace with a record 6.9% surge in hours, according to the Productivity Commission.
It begs the question: Is something amiss with the way the technology is being implemented?
Onboarding AI: treating it like a new employee
In many ways, AI is almost human in the way it responds. So maybe we think of AI as a person, a new employee.
The first step for any new employee is onboarding. Now, your newly engaged AI colleague has a huge technical advantage over the flesh and blood colleague, but there are aspects of normal onboarding that should be considered to ensure your new work colleague can merge seamlessly into the business and its workflow.
Trust in its purpose is undoubtedly a concern. Employees may well be cynical or at best sceptical of its insights and purpose, especially with reports that AI systems are already learning how to deceive humans.
And rather than thinking of AI as a useful pair of hands to ease heavy workloads and produce better outcomes, the technology is seen as the first step in reducing human contribution both in numbers and direct engagement.
And nobody wants to train their own replacement.
An unhappy and suspicious workplace is not conducive to good business and improved productivity.
So, business leaders, with an eye to the future, need to do something about it.
Employee engagement is essential. Seeking input from employees can, in one fell swoop, tailor effective and creative solutions that meet their needs and address their concerns.
It’s an opportunity to educate employees about AI’s capabilities and limitations, reinforce its value and set ground rules to ensure the technology delivers its productivity promise.
And, as with any new employee, don’t overwhelm AI on the first day. Start small with pilot projects and use case studies to demonstrate to the rest of the employee group how change could benefit them and the company.
Trust and transparency prevents ‘hallucinations’
Accurate and transparent access to data can be enlightening.
Sharing even basic automated reporting creates visibility of your company’s data for everyone, which can lead to accountability for improving the quality of data inputs.
Poor data inputs lead to a high hallucination rate for AI models – delivering insights that are useless or inaccurate – a quick way to eliminate trust.
Like any employee, AI must answer to someone. A sound governance strategy is to establish clear accountability frameworks that define who is responsible for AI outcomes and demonstrate human oversight is being maintained to monitor AI performance.
Business leaders who take the time to onboard AI properly, and bring their people along for the journey, have the best chance of realising those much-promised productivity gains.
AI’s arrival is something to celebrate. So let’s lay out the welcome mat for a valuable new team member.
A jump of just 1% in productivity can build competitive advantage and operational efficiency – in other words, it can make a business more resilient.
But that doesn’t mean there won’t be teething issues or blockers to overcome.
