Fortunately for managers, building human capital takes a long time. At least in the past, artificial intelligence is, among other things, a technology that accelerates learning and improves performance. Millions of people are using large-scale language models today. Not all of them are flirting with chatbots. Instead, with the help of AI, we found ourselves able to quickly perform tasks we had never done before and learn about subjects that were previously inaccessible. What happens if the rate of human capital growth suddenly accelerates? This is one of the challenges that AI poses to the business world, which is struggling to understand the value of the technology.
For a variety of reasons, it feels strange to think of AI as a tool for increasing human capital. Perhaps its utility lies in intelligent automation that makes hard-earned human knowledge redundant? Leading AI companies talk about a future where their systems replace workers en masse. Large companies currently integrating AI into their businesses are almost certainly thinking along similar lines. We have to do that because AI is expensive. Microsoft charges per user for its enterprise chatbot Copilot. If a large company with thousands of employees wants to buy Copilot “seats” for their employees, they are looking to invest millions of dollars each year.
Will that “expenditure” lead to a reasonable return? The easiest way for companies to answer this question is to think in terms of new products or layoffs. This generates revenue and reduces costs, respectively. (Of course, you can also combine the two.) In a new report on “enterprise” AI released this week, OpenAI presents a number of examples focused on products that replace human labor. A classic example is an AI voice agent that helps with customer service calls. The company says one such agent is currently saving companies “hundreds of millions of dollars annually.”
Given all this, it seems as though employee replacement is the logical endpoint for enterprise AI, but it's important to note that large companies often have difficulty figuring out how to integrate new technologies, both conceptually and as an internal accounting matter. When IT departments were in their infancy in the 1980s and '90s, it was sometimes unclear how they could be justified within a company. IT departments can spend millions of dollars each year on new computers, networking hardware, or productivity software. Did all that spending produce a profit? How can we determine its value? If large companies adopt mainframes, some accountants may be replaced. If an IT manager wants to explain to her boss why computers are important, the simplest thing she can say might be that computers can replace typing pools.
However, over time, it became clear that the costs and benefits of information technology far exceeded what could be explained in this way. Modern companies have reorganized their organizations around computers. In this new world, the goal of IT departments was not to replace computer-dependent employees, but to increase their efficiency. Employees started demanding more from IT departments. In a development known as “consumerization,” the tools tech-savvy employees use at home (such as smartphones) have become more sophisticated than those provided at work. Employees who wanted more began requesting upgrades. As a result, when IT “spends” are proposed today, no one claims that the investments will have the crude effect of displacing workers. The key question is whether new investments will help existing employees meet their goals and keep up with other companies' competitors.
The idea that the best use, and perhaps the only profitable use, of AI is direct replacement of workers combines two ideas. One stems from speculation about the future of AI, and the other stems from the short-term balance sheet thinking that is perhaps inevitable as companies explore new technologies. On the other hand, this is very different from what many of us experience when actually using AI. A huge number of individuals are paying for accounts with OpenAI, Anthropic, and other companies because they believe AI will improve their abilities and productivity. From their perspective, it's a multiplication of human capital. If you know the details of what you want to accomplish, whether it's writing software, analyzing research, diagnosing a disease, or fixing something in your home, AI can help you do it faster and better. Today's companies spend a lot of money training their employees. Even highly qualified white-collar workers are being sent to online seminars or expensive training camps in hopes of coming back improved. Let's say AI improves some employees' knowledge and abilities by 5 to 10 percent. How much should companies pay for cognitive enhancement?
