Author: Andrew Ng

Translation: Deep Tide TechFlow

Deep Tide Reading: Andrew Ng's latest article directly counters the narrative that "AI will lead to mass unemployment." His core argument is: software engineering is the industry most affected by AI, but hiring demand remains strong; the unemployment rate in the United States remains at a healthy level of 4.3%. AI labs, SaaS companies, and companies undergoing layoffs each have their motives to exaggerate AI's replacement capability. The value of this article lies in the fact that it comes from someone who both drives the development of AI and has long been concerned with the popularization of AI education.

There will be no AI apocalypse of unemployment.

The narrative that "AI will cause mass unemployment" is creating unnecessary panic. AI, like any technology, will affect jobs, but blowing up the narrative of "mass unemployment" in this way is irresponsible and harmful. It needs to stop.

I have previously expressed skepticism about the "AI unemployment apocalypse." The good news is that mainstream media is now beginning to push back against this narrative. The image below collects some recent headlines.

Software engineering is the field most impacted by AI tools, with coding agents leading the way. However, the demand for software engineers remains strong. While there are indeed cases of AI replacing jobs, the trend clearly indicates that the net number of jobs created by AI far exceeds those it eliminates, just like every previous wave of technology. Moreover, despite exciting advancements in AI, the unemployment rate in the United States remains at a healthy 4.3%.

Who is promoting the "AI unemployment apocalypse"?

Why is the "AI unemployment apocalypse" narrative so popular?

Firstly, leading AI labs have strong motives to tell stories that make AI sound incredibly powerful. The most extreme version is the sci-fi scenario where AI "takes over the world" and leads to human extinction. The logic is simple: if a technology can replace a large number of employees, then that technology must be very valuable.

Secondly, many SaaS companies have pricing ranging from $100 to $1000 per user per year. However, if an AI company can replace an employee with a $100,000 annual salary, or increase their efficiency by 50%, then charging $10,000 a year seems reasonable. The anchor point shifts from SaaS pricing to employee salaries, allowing AI companies to charge much higher prices.

Additionally, companies themselves have motives to attribute layoffs to AI. After all, saying "we improved efficiency with AI, so we streamlined our workforce" sounds much better than "we hired too many people during the cheap capital period of the pandemic." The real reason behind the latter is over-hiring caused by low interest rates and massive government fiscal stimulus.

This is not the apocalypse of unemployment; it's a job carnival

I admit that AI is changing many people's jobs. This is challenging and makes people anxious. (For some, it may even be quite interesting.) I understand every person affected. However, changes in job content and predicting a collapse of the job market are completely different things.

Society has the capacity to tell itself a story for many years, even if this story has little basis in reality, and then make poor collective decisions based on it. For example: fear of nuclear power safety leads to severe underinvestment in nuclear power; panic over the "population bomb" in the 1960s led multiple countries to implement strict population control policies; and concerns over dietary fats led governments to promote unhealthy high-sugar diets for decades.

Mainstream media is now beginning to openly question the "AI unemployment apocalypse" narrative, and I hope such stories gradually lose influence, just as the panic of "AI leading to human extinction" slowly fades away.

It's not an apocalypse of unemployment; it's a job carnival

My prediction is exactly the opposite: there will be a job carnival driven by AI. AI will create a large number of high-quality AI engineering positions, and I am also very optimistic about the future of the overall job market. The work done by AI engineers will differ from traditional software engineering, and many new positions will appear in companies outside traditional large tech employers. The skills required for non-AI positions will also change due to AI. Now is a good time to encourage more people to learn AI and prepare for a future with different but abundant jobs.