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 america arsenic a preferred source on Google.

ZDNET's cardinal takeaways

• Many tasks are ripe for automation, but others autumn into a grey area. 
• Speed and productivity are not the extremity of the story.
• AI decisions are absorption 101, not method decisions. 

You whitethorn person already heard the presumption that AI agents service arsenic "co-workers" to quality counterparts, functioning arsenic de facto extensions of the workforce. The situation is decoding what enactment they are champion suited to execute -- and it's not an casual question.

There are tasks ripe for automation and others that are amended handled manually. But galore are successful a grey area, successful which automation makes sense, but is it worthy the investment? The clip has travel to presumption AI agents arsenic an adjunct workforce, and negociate them accordingly. 

Also: True agentic AI is years distant - here's wherefore and however we get there

There is adjacent speculation that IT, which manages agents, is taking connected the relation of a quality resources department. 

Management 101 skills vs. method skills 

That calls for absorption 101 skills versus purely method skills, urges Ethan Mollick, prof astatine the University of Pennsylvania and starring AI expert. In "Management arsenic AI superpower," helium describes the thought process for deciding if AI should grip a task. 

From an economical and productivity standpoint, determination is nary question that AI is accelerated and cheap, Mollick says. "It produces enactment successful minutes that would instrumentality galore hours for a quality to do, and it doesn't caput if you make aggregate versions and propulsion astir of them away." 

Consider the emergence of chatbots successful lawsuit interaction systems. Automated responses -- boosted by AI -- present grip ample volumes of regular inquiries that slowed down reps' productivity to a crawl astatine 1 time. Now those reps are elevated to grip much analyzable lawsuit issues.  

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

However, velocity and productivity are not the extremity of the story. Here's different origin to consider: "You don't reliably cognize what the AI volition beryllium bully oregon atrocious astatine connected analyzable tasks," says Mollick. Doing the incorrect happening faster has ever been 1 of the top challenges of management. 

The champion attack to making specified decisions is not to presumption AI deployment arsenic purely a exertion decision, but to presumption it from a absorption 101 perspective. This whitethorn yet thrust the AI and AI cause question to the adjacent realm -- to beryllium handled arsenic an adjunct workforce. 

To get the astir retired of AI agents, for example, you request to beryllium capable to "explain what you need, springiness effectual feedback, and plan ways of evaluating work," Mollick says.

How to find if delegation makes sense

Again, absorption 101 comes into play here. Do it, ditch it, oregon delegate it. "You delegate due to the fact that you can't bash everything yourself, and due to the fact that endowment is constricted and expensive," Mollick explains. "AI changes the equation. Now the endowment is abundant and cheap. What's scarce is knowing what to inquire for."

Also: Deploying AI agents is not your emblematic bundle motorboat - 7 lessons from the trenches

Successful delegation, then, depends connected being capable to determine what to delegate a task to an AI agent. Mollick suggests 3 measurements to find if AI is champion for the task:

• Human baseline time: "How agelong the task would instrumentality you to bash yourself."
• Probability of success: "How apt the AI is to nutrient an output that meets your barroom connected a fixed attempt."
• AI process time: "How agelong it takes you to request, hold for, and measure an AI output."   

These determination points interact with each other, and determination are trade-offs that request to beryllium weighed. And successful figuring retired however to springiness these instructions to the AI, it turns retired you are fundamentally reinventing management. "Consider a task that takes you an hr to do, but the AI tin bash it successful minutes, though checking the reply takes 30 minutes," Mollick illustrates. "In that case, you should lone springiness the enactment to the AI if 'probability of success' is precise high, different you'll walk much clip generating and checking drafts than conscionable doing it yourself."

"If the 'human baseline clip is 10 hours," helium continued, "it could beryllium worthy respective hours of moving with the AI, assuming that the AI tin beryllium made to bash a competent job."

Also: 4 caller roles volition pb the agentic AI gyration - here's what they require

A squad of AI agents needs to conscionable the aforesaid goals arsenic a quality team, by asking basal absorption 101 questions, specified arsenic "what are we trying to accomplish, and why?" and however to way and measurement an agent's progress.  

Management itself, arsenic good arsenic the workplace, whitethorn germinate "when everyone is simply a manager with an service of tireless agents," Mollick predicts. "The radical who thrive volition beryllium the ones who cognize what bully looks similar -- and tin explicate it intelligibly capable that adjacent an AI tin present it."