Business owners and managers should gear up quickly to adopt artificial intelligence, as the early users will get the lion’s share of benefits, according to an expert who studies the impact of technology on business productivity and strategy.

Erik Brynjolfsson, director of the Stanford Digital Economy Lab and a senior fellow at the Stanford Institute for Human-Centered AI, is one of the main speakers at this year’s Super Session, an annual conference hosted by the University of Toronto’s Creative Destruction Lab.

“This revolution is going to happen very quickly. This is much bigger than the technologies of the past decade. I think it’s ultimately going to be seen as a bigger effect on work and employment than the pandemic, which had a huge effect, but this is going to be much more lasting,” Brynjolfsson told BNN Bloomberg.

Brynjolfsson noted that the diffusion of AI technology can happen quickly as it can leverage existing platforms like the internet or office suites. Earlier this year, a UBS study estimated that the ChatGPT chatbot reached 100 million monthly active users just two months after its launch, making it the fastest growing consumer application in history.   

Generative AI’s large language model or LLM technology is expected to impact the work of up to 80 per cent of the U.S. work force in some form, Brynjolfsson said, referencing a research paper.

“Knowledge and information workers in fields like law, advertising, medicine and finance are going to be in the bullseye of the transformation,” said Brynjolfsson, adding, “This will affect white collar information workers a lot more than blue collar workers.”

In a recent paper he co-authored, Brynjolfsson said their study showed a 30-35 per cent increase in productivity of call centre workers, and over 50 per cent in software coding, with the help of AI. He advises businesses to prioritize which areas of their operations could benefit most from AI.

Workhelix, a company he co-founded, analyzes tasks in variously-sized companies and highlights the ones most likely to be affected, especially those involving language and white collar work.

The biggest cost to a business is often not the technology. In the call centres that Brynjolfsson studied, he explains, “They added a tool to help the workers answer questions more effectively. The humans were still answering questions from the customers, but it was a new set of software and screens that helped them. There was some retraining cost, some reorganization cost but the software itself is not that expensive.”

As evident from the rapid uptake of ChatGPT, organizations don’t require workers with special technical or programming skills to start using off-the-shelf AI tools.

“If you want to train your own LLM, that of course requires more skills. But for most people I think the place to start is from some of the off-the-shelf tools like ChatGPT or Bard or Inflection AI. What I recommend is that CEOs of every company ask their teams to spend a day just having a hackathon and talk about how the existing tools can change their own work,” Brynjolfsson said.

If LLMs are estimated to affect a vast majority of the knowledge sectors, what should workers expect? Mass disruption is a definite risk, rather than mass unemployment, according to Brynjolfsson.

“I think it’s unlikely we’ll have a big loss of jobs any time soon. But there will be a lot of disruption and shifting around of work as demand increases for some jobs, decreases for others,” he said, adding that humans simply cannot be replaced in many functions.

“There’s no question there are a lot of jobs that only humans can do whether it’s in creative work, child care, health care, environmental work, and so we’re going to have demand for people to do those kinds of jobs, even as some other jobs, like routine legal analysis, can be done more easily by machines.”

Several experts have recently warned of the risks of the misuse of AI. Last month, the Center for AI Safety released a statement signed by over 350 executives and scientists working with AI, that read, “Mitigating the risk of extinction from AI should be a global priority alongside other societal-scale risks such as pandemics and nuclear war.”
The threat is real, Brynjolfsson agreed.
“I don’t want to be alarmist, but I do want people to have their eyes open.” That includes businesses stepping up their vigilance and updating their security infrastructure, he said.
Asked whether the spread of AI would benefit businesses or consumers the most, Brynjolfsson said in the end, over 95 per cent of a technology’s benefits go to consumers. But in the short run, he said, early movers among businesses could be the winners that take all, scoring on productivity.
“We’re in the early stages of a revolution in work, employment and productivity. The coming decade is going to see much higher productivity growth than the official projections,” Brynjolfsson said.