From the advent of GPT-4 to the launch of Wenxin, to the debut of Sora, to the recent explosion of domestic Kimi, many people are amazed every time a phenomenal AI product comes out. AI is about to usher in the “iPhone moment” that will disrupt everything.

After a year and a half of catalysis, AI, dubbed the “Fourth Industrial Revolution,” actually has a big impact on ordinary people?

On April 2, a team led by Goldman Sachs economist Jan Hatzius released a report entitled “One Year After AI Transformation: Progress Is Smooth But the Impact Will Still Take Years”. The report said that at present, the actual application of AI is still limited. Less than 5% of companies use generative AI in production, and the current level of application is not enough to significantly improve productivity.

Goldman Sachs pointed out that the low application rate of AI has limited its impact on the labor market. Preliminary data shows that the development of AI has slightly increased demand in the labor market, and the impact on the unemployment rate is negligible, thus having a slight positive impact on the labor market.

For early users of AI, Goldman Sachs said that judging from data, AI can greatly improve production efficiency. Academic research and business cases have shown an average increase of 25% (median of 16%) after using AI.

The large-scale impact of AI will take time

Goldman Sachs pointed out that in the past year, through a series of articles, they have demonstrated that generative AI can increase labor productivity and promote global growth, but the emergence of these increases largely depends on the widespread application of AI technology and AI-related investments, and both of these may take years to fully achieve:

We have previously pointed out that the AI wave will become an important driver of global productivity. Over the next 10 years, global productivity will increase by more than 1.5% every year, driving economic growth of 7 trillion US dollars.

However, achieving a boost to macroeconomic growth largely depends on the large-scale application of technology and the scale of AI-related investments. We previously indicated that both may take years to fully achieve.

Goldman Sachs pointed out that the financial market also sees AI as a landmark technology, driving investment first and increasing productivity thereafter:

Judging from the gains over the past six months, US stock market earnings are mainly concentrated on AI hardware vendors (such as semiconductor companies) and software vendors (such as cloud service providers). They will benefit from the early growth of AI-related investments, while potential beneficiaries of AI productivity have only received more moderate increases.

As emphasized by our US portfolio strategy team, these benefits (hardware companies in particular) mainly reflect improvements in fundamentals driven by demand related to GPUs and data centers, rather than higher valuations driven by improved investors' expectations of the company's prospects.

Goldman Sachs believes that technology companies were among the first to benefit from AI applications, but judging from the AI mentioned in various companies' earnings conference calls, more and more industries have incorporated AI into future strategies, including IT and communications, industry, consumption, finance, and healthcare. This indicates that AI applications may bring efficiency improvements are receiving widespread attention:

Although AI applications are currently mainly concentrated in the technology industry, widespread attention from other industries and the rapid growth in AI-related investments mean that the widespread application of AI and the resulting increase in productivity may gradually be realized in the next few years

In 2025, additional investment in AI hardware could reach 250 billion US dollars, equivalent to 9% of US companies' investment or 1% of GDP. This is in line with previous predictions. Among the various AI hardware fields, the main focus is on the increase in demand for semiconductors (Nvidia accounts for more than 75%) and cloud services, servers, and network devices

Goldman Sachs believes that the stock market provides an optimistic signal that the AI investment cycle may have begun, but there is no significant increase in AI-related investment in official national accounts data (the main basis for GDP accounting). This suggests that factors other than AI, including demand for non-AI technology and cyclical factors, may currently play a more important role in driving overall capital expenditure.

AI is not widely used

According to data published by the well-known web market analysis platform SimilarWeb, at the beginning of 2023, the number of global visits to ChatGPT exceeded 1 billion, and in February of this year, the number of global visits to ChatGPT reached 1.6 billion. Goldman Sachs believes that this data shows that more than a quarter of Americans use artificial intelligence tools informally at least every week.

Goldman Sachs pointed out that despite the surge in visits to AI applications, with the exception of a few specific high-tech industries, the proportion of companies officially adopting AI is still very low. According to the Census Bureau's newly launched “Business Trends and Prospects Survey” AI supplement report, less than 5% of companies officially use AI generation technology to launch products and services, while this ratio has reached 10-15% among information, professional services, and financial companies:

There is a greater difference in AI usage in more segmented sub-industries. In the technology industry, more than 20% of companies use artificial intelligence generation tools in production, while in other digital fields such as film and sound production, current and anticipated AI usage is higher.

Goldman Sachs pointed out that companies expect most industries to adopt AI faster in the next six months, and many companies are investing in AI applications they will use in the next few years:

According to the IT spending survey, although currently only 12% of CIOs plan to spend more than 5% of their IT budget on generative AI, the proportion is expected to exceed half within the next three years, and AI's share of the IT budget will also double.

Early AI users improved their efficiency significantly

Goldman Sachs pointed out that although the adoption of AI is still in its early stages, and there is still a high degree of uncertainty about the final effects of the widespread use of AI, evidence from early users shows that AI may bring huge efficiency improvements:

We assess the impact of AI usage on productivity from academic research and company-level reports. There are a few things to keep in mind.

First, pay attention to selection bias. Early adopters are usually the group that can benefit the most from the new technology, so the effect may be higher than that of ordinary users. Most of the current cases focus on occupations with high repetitive tasks and suitable for AI automation, which may make the estimation results optimistic.

Second, publication bias. Academia and businesses may be more inclined to publish and report positive effects while ignoring less significant or negative results, which may also lead to overestimation. Furthermore, there is a possibility of underestimation. The current case only considers the automation of some tasks. If AI applications expand to more tasks, there may be more room for improvement.

According to Goldman Sachs, academic research estimates that AI can increase labor productivity by 9-56%, with a median of 16% and an average of 25%; corporate reports expect a median increase in productivity of 25%, with an average of 26%. Some studies have found that AI not only increases output, but also improves the quality of work.

At the same time, Goldman Sachs said that the productivity improvement effect of AI for newcomers is more significant than that of experienced employees. AI may be more helpful in speeding up learning and improving the efficiency of less experienced employees, while the help for highly skilled workers is relatively limited.

Editor/Somer