Is your AI agent up to the task? 3 ways to determine when to delegate

MF3d/iStock/Getty Images Plus via Getty Images

Follow ZDNET: Add us as a preferred source on Google.

ZDNET's key takeaways

• Many tasks are ripe for automation, but others fall into a gray area. 
• Speed and productivity are not the end of the story.
• AI decisions are management 101, not technical decisions. 

You may have already heard the view that AI agents serve as "co-workers" to human counterparts, functioning as de facto extensions of the workforce. The challenge is decoding what work they are best suited to perform -- and it's not an easy question.

There are tasks ripe for automation and others that are better handled manually. But many are in a gray area, in which automation makes sense, but is it worth the investment? The time has come to view AI agents as an adjunct workforce, and manage them accordingly. 

Also: True agentic AI is years away - here's why and how we get there

There is even speculation that IT, which manages agents, is taking on the role of a human resources department. 

Management 101 skills vs. technical skills 

That calls for management 101 skills versus purely technical skills, urges Ethan Mollick, professor at the University of Pennsylvania and leading AI expert. In "Management as AI superpower," he describes the thought process for deciding if AI should handle a task. 

From an economic and productivity standpoint, there is no question that AI is fast and cheap, Mollick says. "It produces work in minutes that would take many hours for a human to do, and it doesn't mind if you generate multiple versions and throw most of them away." 

Consider the rise of chatbots in customer contact systems. Automated responses -- boosted by AI -- now handle large volumes of routine inquiries that slowed down reps' productivity to a crawl at one time. Now those reps are elevated to handle more complex customer issues.  

Also: As AI agents multiply, IT becomes the new HR department

However, speed and productivity are not the end of the story. Here's another factor to consider: "You don't reliably know what the AI will be good or bad at on complex tasks," says Mollick. Doing the wrong thing faster has always been one of the greatest challenges of management. 

The best approach to making such decisions is not to view AI deployment as purely a technology decision, but to view it from a management 101 perspective. This may ultimately drive the AI and AI agent movement to the next realm -- to be handled as an adjunct workforce. 

To get the most out of AI agents, for example, you need to be able to "explain what you need, give effective feedback, and design ways of evaluating work," Mollick says.

How to determine if delegation makes sense

Again, management 101 comes into play here. Do it, ditch it, or delegate it. "You delegate because you can't do everything yourself, and because talent is limited and expensive," Mollick explains. "AI changes the equation. Now the talent is abundant and cheap. What's scarce is knowing what to ask for."

Also: Deploying AI agents is not your typical software launch - 7 lessons from the trenches

Successful delegation, then, depends on being able to decide what to delegate a task to an AI agent. Mollick suggests three measurements to determine if AI is best for the task:

• Human baseline time: "How long the task would take you to do yourself."
• Probability of success: "How likely the AI is to produce an output that meets your bar on a given attempt."
• AI process time: "How long it takes you to request, wait for, and evaluate an AI output."   

These decision points interact with each other, and there are trade-offs that need to be weighed. And in figuring out how to give these instructions to the AI, it turns out you are basically reinventing management. "Consider a task that takes you an hour to do, but the AI can do it in minutes, though checking the answer takes 30 minutes," Mollick illustrates. "In that case, you should only give the work to the AI if 'probability of success' is very high, otherwise you'll spend more time generating and checking drafts than just doing it yourself."

"If the 'human baseline time is 10 hours," he continued, "it could be worth several hours of working with the AI, assuming that the AI can be made to do a competent job."

Also: 4 new roles will lead the agentic AI revolution - here's what they require

A team of AI agents needs to meet the same goals as a human team, by asking basic management 101 questions, such as "what are we trying to accomplish, and why?" and how to track and measure an agent's progress.  

Management itself, as well as the workplace, may evolve "when everyone is a manager with an army of tireless agents," Mollick predicts. "The people who thrive will be the ones who know what good looks like -- and can explain it clearly enough that even an AI can deliver it."