In the era of superintelligence, will people no longer need to work?

Science fiction and technology are overlapping at an accelerating pace.

On September 24th, when Wu Yongming, the CEO of Alibaba, proposed at the Yunqi Conference that the ultimate goal of AI is ASI (Artificial Superintelligence), and everyone will have hundreds of Agents serving them, the first question that popped into my mind was: "What will people be doing then?"

Elon Musk once said in an interview, "If AI can do your job better than you, how can we find meaning in life?" The same question appears in Nick Bostrom's book The Future of Humanity: The Goals and Meanings of Humans in the Age of Superintelligence. In a "world where all problems are solved," what will give meaning and purpose to our lives?

The second, more realistic question is, when ASI arrives, will people really not need to work?

John Maynard Keynes once predicted that by 2030, the total production value of human society would increase by 4 - 8 times compared to his era, and the average working hours would be reduced to 15 hours per week. However, the reality is that with five years left until 2030, the per capita GDP has increased by more than 7 times, but the mainstream working hours around the world are still 35 - 40 hours per week. Of course, this figure excludes many large Chinese Internet companies, as "996" is their norm.

The third, more realistic question is, if large models become the next - generation operating system and a single sentence can create an Agent, what will happen to traditional commercial software companies? Wu Yongming said that ultimately, large models will engulf software, and there will no longer be commercial software that requires professional development by engineers.

In 2023, when large models showed "intelligent emergence" in programming, I asked more than 10 software companies what they would do if in the future, large models could directly turn users' requirements expressed in natural language into executable functions. Most business leaders thought it was a wild idea.

Two years later, this is a future that can be seen.

Image source: unsplash

The order of answering these three questions is exactly the opposite.

If we start from the most specific and realistic question of "the fate of software companies" and finally reach the most abstract and fundamental question of "the meaning of human existence," this process of reverse questioning may be the decades during which we gradually figure out these three questions.

First, will commercial software companies disappear?

They won't disappear but "merge."

The core logic of traditional commercial software is "function encapsulation": solidifying business processes into code and allowing users to operate through interface interaction. However, when large models can understand natural language, call APIs, generate code, and execute tasks, the "shell" of software becomes redundant. Users no longer need to click the "Create Order" button. They just need to say, "Help me place an order for 5,000 pieces with Customer A and use the optimal logistics solution." - The Agent will automatically complete the entire process.

This means that software will no longer exist in the form of a "product" but will run in the form of a "service flow" or an "intention agent." The role of software companies will undergo a fundamental transformation.

We can also see this from the latest actions of some multinational commercial software companies. SAP's strategy this year is "AI first, suite first." Through AI + suites, data silos are broken, and business processes achieve smooth end - to - end flow. In this way, Agents can be used out of the box in various fields such as finance, supply chain, human resources, and IT development. As it develops further, in the future, software companies may become "Agent operators." Although it seems that users can create an Agent with a single sentence, behind it is industry semantics learned from countless "best practices."

Therefore, commercial software companies will not "disappear" but will integrate into the large - model ecosystem. Their moat will no longer be code but domain knowledge, data assets, and ethical governance capabilities.

Second, will people really not need to work?

No need to work, but must keep learning.

Less than three years after ChatGPT entered the scene, the world has changed dramatically. Programming, design, legal documents, market analysis, and even writing... These jobs once considered "high - skill" can now be done, or partially done, by AI.

At this rate, in five years, the replacement of certain human jobs by AI will become an obvious fact. Even jobs that we previously considered to involve creative cognition may be replaced by Agents.

Image source: unsplash

However, "not needing to work" does not mean "having no tasks." In the real world, humans still need to set goals, evaluate results, and handle ambiguity and value conflicts. For example, a marketing Agent can generate 100 sets of plans, but humans still need to judge "which one is more in line with the brand's values."

If making judgments and decisions on AI work will become a new form of work, then where does this decision - making ability come from?

Not long ago, iQiyi announced that it would launch an "AI theater" where AI would completely create TV dramas, and invited Peter Pau, the cinematographer of Crouching Tiger, Hidden Dragon, as the general consultant. Why choose a cinematographer instead of a director? My personal guess is that aesthetic sense will be the most important ability for future AI content creation.

Similarly, if humans want to become important decision - makers responsible for the results of AI work, then you must have the ability to judge. Just as Peter Pau is an Oscar - winning cinematographer.

So, if you want to become a new - era worker such as an "Agent ethics auditor," an "intention clarifier," or a "digital empathy coach," you may not need to repeat mental and physical labor, but you cannot stop learning. In fact, you may need to learn throughout your life.

Third, when all problems are solved, what will people be doing then? (PS: There is no need to learn either. Through a brain - computer interface, you can instantly obtain data intelligence.)

I don't have an answer.

In the medieval legend of Cockaigne, Utopia is a wonderful world where "people can live leisurely without working. Regardless of age, strength, or weakness, everyone can enjoy life without worrying about material shortages, and even the walls are made of sausages." In Wu Yongming's description, it is a world where hundreds of Agents work for you 24 hours a day. Just as Kevin Kelly believes that in the future, smart glasses will access all the data of the world and humans, so it will be a transparent "mirror world," and humans can obtain fully personalized services, making life extremely convenient.

But before reaching the ASI era where humans can "lie back" and all problems are solved, how can we overcome the "Dark Forest" rule formed in hundreds of thousands of years of human evolution?

In Isaac Asimov's Robot series of short stories, he depicts a Earth controlled by a supercomputer. It knows everything about the Earth and can arrange the most appropriate economic activities according to Earth's resources. It is the real "overlord" on Earth. In The Future of Humanity: The Goals and Meanings of Humans in the Age of Superintelligence, in the letter that the fox Fyodor wrote to his uncle, he tells the story of how he and a philosopher wanted to establish a harmonious society for animals but were eventually attacked by wolves. This seems to imply that Utopia will ultimately be unable to resist the real world where the strong prey on the weak.

From this perspective, Kevin Kelly may be more realistic. He wrote in The Inevitable: "Protopia creates almost as many new problems as it does new benefits. Today's problems come from yesterday's success. And the technological solutions to today's problems will plant hidden dangers for tomorrow. As time passes, real benefits gradually accumulate behind this cycle of problems and solutions expanding simultaneously."

However, whether we expect it or not, ASI will eventually arrive.

Images / unsplash   Doubao AI

This article is from the WeChat official account "IT Times," author: Hao Junhui. Republished by 36Kr with permission.