AI will create more jobs, not fewer jobs
getty
Amazon founder Jeff Bezos just predicted in an interview with CNBC that AI will lead to an impending “labor shortage” rather than job losses. People predicting jobs are wrong, he says. “What’s really going to happen is to uplift people.” It’s like a software engineer who was digging a basement with a shovel and was handed a bulldozer. “We humans will never run out of problems and the need for solutions, but that work will be done at a higher level with bulldozers.”
Dan Shipper, CEO of Every, believes this applies not just within his company, but across the industry. AI speeds things up and solves problems, but humans work non-stop. AI “is creating more jobs for humans, not fewer. The more automation there is, the more specialized human jobs need to be.”
In his latest letter, Schipper explains how his company has “automated everything that can be done using AI agents, but humans have to do much more work than ever before. Since GPT-3, we’ve gone from having four human employees to more than 30.”
The paradox of AI is that replacing some of the work of experts may only serve to emphasize the need for human experts. “There are more situations where expert judgment is needed,” Schipper explained. “If an operations person uses AI to submit a pull request, they need an engineer to review it. If a marketer creates a YouTube thumbnail, they need a designer to sharpen it. If an engineer writes, they need writers and editors to make the draft just right.”
Schipper explains that there are two forces at play. First, AI still requires human oversight at the beginning and end of the process. Second, as AI churns out homogenized sameness outcomes, the value of human expertise increases to the point where it becomes a kind of status.
Human expertise is required to maintain the tasks handled by AI. “We have a team of AI engineers who are responsible for making sure our agents function properly, and we will need them for the foreseeable future,” he said. Humans are required at the beginning and end of all processes involving agents.
- At the beginning of the process, Humans “set the framework of what are we going to do? What is considered good?”
- During the process, The AI agent “drafts, searches, summarizes, and compares tasks.”
- At the end of the process, Humans judge and expand, “Is it a good thing? Where does it belong? What should happen next?”
Yes, there are roles that an AI agent could easily take on, such as “code, prose, images, support tickets, product specs, etc.,” Schipper says. “They take everything, every successfully completed task, and package it into a format that’s affordable to everyone. As a result, previously rare skills like coding pull requests, creating YouTube thumbnails, and writing newsletters are now widely available to almost everyone.”
At the same time, such richness creates sameness and increases the demand for expertise that can provide differentiation and diversity. Widely available models result in “visible sameness, nauseating repetition.”
That sameness creates a demand for differentiation that only humans can achieve. “If you look too much, you start to smell like a rat,” Schipper said.
For some aspects of business, such blandness and sameness is fine. For example, financial reports or customer surveys. But to achieve innovation and strategic thinking leadership, progressive companies need deliverables that “feel vivid and specific, not cheap and generic,” Schipper said. “I want something with status.”
As a result, demand for specialists will increase, Schipper noted. “Scarce and valuable work must be done by humans. The models of the current generation only know about the work done. Humans know what they have to do in this moment. If a situation is reduced to a text and made into a corpus, it is a corpse.”
