- Some workers are worried artificial intelligence will come for their jobs.
- Yet AI might sometimes be more like an intern — taking on tasks that otherwise wouldn't get done.
Like most people, there were things on Heather Holding's to-do list she knew she'd never get to.
That led Holding, who oversees risk management at fintech company Best Egg, to wonder whether an artificial intelligence tool could draft policy documents she and her colleagues couldn't find time to pull together.
"One day, I was like, 'You know what, let's just try to put these through,'" Holding told Business Insider. So, she wrote prompts for an AI tool built into a corporate reporting platform made by Workiva. The chatbot drafted documents that needed some tweaking, but Holding and her team got through the backlog of 10 policies within a day.
"We operate very lean. So something like this is really a game changer for us," she said.
Holding's experience points to the sometimes subtle ways AI is changing work. Many workers have expressed fears that the technology will show up at the office like a young gun coming for their job. Yet, in some cases, AI might be more like the temp that gets brought in when work piles up — or to do the jobs nobody wanted to do.
An AI bot could act as your intern.
One reason chatbots like the one Holding used might not take over tons of jobs right away is there are limits to what AI can do — for now. Generative AI tools often still require oversight, much like an inexperienced worker might. Holding has to review and sometimes adjust what a bot kicks out, but even with that, she saves time.
To write a policy doc from scratch, she said, might take from eight to 30 hours, depending on the type and how much research it requires. As long as she does the upfront legwork to draft the proper instructions for the bot, the results are compelling, she said.
"You can literally have the sections mirror exactly what you want in an actual policy," Holding said.
Tackling those jobs that people otherwise might not even start is an important part of how AI will transform the workplace, according to Ron Williams, CEO and founder of Kindo, which helps companies adopt and use AI tools securely. In part, it's because the technology lets companies do things they couldn't do easily or cost-effectively before.
"There are jobs not getting done today that aren't threatened at all — that AI is just going to be able to start doing," he told BI.
Williams expects some of the biggest gains from AI won't center on what companies and workers are doing now. Instead, he sees major wins coming from what workers aren't doing but that AI could. Accomplishing those things could change the trajectory of a business, Williams said.
"As the AIs get better, that's just going to accelerate and expand the reach of what they're going to be able to impact," he said.
Williams pointed to work such as content moderation or aggregating reviews in e-commerce. He said AI can sift through thousands of posts and summarize customers' sentiments about a product. Amazon is trying this. Hot takes from AI could boost sales, Williams said, if shoppers have a better understanding of what people think about a product. It's the kind of service that wasn't feasible without technology.
"Before, that just wasn't economical to have a human go through and try to aggregate that stuff," he said.
Filling in when help is scarce
One reason jobs sometimes aren't getting done is because there aren't always enough workers. So, some leaders are looking to AI for help. In a Gigged.AI survey of about 250 UK execs in mid-2023, about half said GenAI could address skills shortages in tech.
There are also efforts to train workers in fields like accounting that are facing worker shortages about the benefits of AI. A startup called Docyt recently introduced a professional development platform to help teach accountants how to use AI to complete their jobs more efficiently and accurately.
Jérôme Pesenti, founder and CEO of an AI-powered learning app called Sizzle and former VP of AI at Facebook parent Meta, told BI that sometimes AI takes on parts of jobs that might be unexpected. He pointed to research labs where machines excel at repeating experiments with high precision. But tasks that might seem menial — like cleaning machines and moving samples — are actually often harder to leave to robots.
"We are literally decades away from automating this," he said. "Yet some jobs that we thought were, you know, maybe like higher level, which was like doing precise experiments — it actually turns out machines are better at doing."
Humans still come out ahead in areas like behaving in a complex environment and dealing with multiple systems, Pesenti said. But that doesn't mean people won't turn to AI as a hired hand in the workplace, he noted. "I'm still betting on the human, but I do think the human needs to leverage the machine to be a lot better," Pesenti said.
At Best Egg, Holding hopes AI might one day take on even more work, like scooping up thousands of rows of data and allowing auditors to ask a chatbot questions about that information or hunt down anomalies. Right now, she might spend up to 40 hours conducting transactional tests. If AI could be deployed to help with this work, the time the task takes might get cut in half, Holding said.
"We're all kind of leaning into this technology to help us do more with less," she said.