Tracking the AI ​​boom: Lessons from economic history

problem:

Artificial intelligence (AI) has experienced an extraordinary boom in recent years. This is because AI is widely recognized as a truly transformative technology that will drive economic growth for many years, comparable to the spread of railroads, electricity, automobiles, personal computers, and the internet over the past two centuries. But many skeptics warn that the boom has gone too far and that the economic benefits will be severely undermined. come lateand its AI is expanding bubble That could have dangerous consequences for the U.S. economy. The dramatic investment cycles in technology transformation to date may provide lessons for understanding the different ways in which AI may actually be a bubble, what the consequences of different forms of bubbles are, and what to look for to determine which path AI is actually on.

Previous waves of technological innovation have experienced failures along the way, even if the innovations eventually became foundational technologies.

fact:

• There are many aspects to the AI ​​boom. artificial intelligence It was my long-standing desire to pursue computer science. After years of slow progress, the use of AI has exploded since November 2022. introduction A user-friendly interactive chatbot developed from Advanced statistical analysis Large dataset. Chat GPT AI chatbots such as are rapidly being adopted by both individuals and businesses. Far surpassing previous waves of computing innovation. Ann Unusual boom in AI-related hardware US software (data centers, chip manufacturing plants) and software (new AI models and tools) account for a large share of recent US economic growth. 1% of GDP growth rate Some estimates say it will happen in 2025. Stock prices of the top seven tech companies rose far too much This has pushed the widely used Shiller P/E ratio, a measure of whether a stock is undervalued or overvalued, closer to its historical peak at the height of the dot-com bubble (see chart). And then something amazing happened. Proliferation of compensation packages For top AI innovators.
• The challenge is to predict the extent to which this AI boom will be justified by continuing tremendous technological advances commensurate with the economic benefits. Despite remarkable recent advances, widespread uncertainty remains regarding the path of continued AI innovation and its potential applications and economic impact. We need to recognize that AI is already being applied in a variety of fields. We provide powerful and user-friendly software for search, image creation, communication, learning, and more for individual consumers. For businesses, it provides tools to quickly perform many mid-level white-collar tasks, such as coding, data analysis, and reporting. AI tools are advancing with new versions that address previous flaws such as “AI hallucinations” (chatbots fabricating incorrect answers) and introduce more advanced inference capabilities. As AI continues to evolve, it could be used to transform many areas of economic activity, such as self-driving robocars, personalized education, supporting the development of new medicines and medical diagnostic tools tailored to individual health needs, or in ways that are difficult to predict. But there is diverse opinions How far ahead will AI capabilities go, or can AI evolve instead? Reaching the limits of technology.
• One important lesson from economic history is that previous waves of frenetic technological innovation Things rarely went smoothly And even if innovation eventually becomes the foundational technology, many people experience setbacks along the way. These booms and busts often take the form of bubbles. of classic bubble shape This occurs in financial markets, where investor expectations for an asset's price to rise attract additional capital, pushing the price far above the asset's true underlying value until sentiment eventually reverses, new capital dries up, and the price collapses. ”dot com bubbleThe late 1990s are a perfect example of the excitement surrounding a new technology, the Internet, that culminated in a financial bubble and a collapse in stock prices. Bubbles can also emerge in the real economy, where new technologies first lead to a surge in infrastructure investment by entrepreneurs looking to build network capacity and capture markets, and then burst when the total infrastructure build-out proves to be far outstripping needs, at least for the time being. american railroad construction in 19 yearsth Century is a well-known example. Railway companies borrowed large amounts of money to build new lines. When there were more railroad bonds for sale than investors were willing to buy, many railroads went bankrupt and one of New York City's largest banks went bankrupt. financial panic of 1873 Next. Note that financial bubbles and boom-bust cycles in the real economy usually go hand in hand, as the dot-com and railroad experiences show.
• Previous stories about the technology-driven investment boom suggest different scenarios for how the AI ​​boom will play out in the coming years. Technology booms often end. Shattered expectations While this technology brought significant benefits to specific sectors, it had limited impact on overall productivity and profitability, at least in the short to medium term (although the long-term benefits were ultimately significant). The pattern was early waves of innovation railways, internet, etc. Developing new skills and products to take full advantage of emerging new technologies often takes longer than expected. In some cases, the new technology was largely superseded by a further wave of innovation: the canals built in the first half of the 19th century.th The century was made redundant by the railroads. (Such a fate, Current large-scale language model-based AI technologiesIn any case, the end result was a widespread bankruptcy with investment collapsing to more normal levels, stock prices reversing sharply, and widespread financial distress for the overinvested companies.
• Alternatively, the technology-driven boom could end with: Collapse of profitabilityIn this case, investors find it difficult to obtain a true economic return in corporate profits that is commensurate with the abnormal level of investment. When the dot-com bubble burst, many of the companies that went bankrupt had high valuations but little in the way of profits. After the dot-com bust, successful software companies found ways to effectively monetize their large investments in the 2000s. For example, Google and Facebook were able to achieve this through sophisticated advertising algorithms developed for their search and social network applications, while Microsoft continued to thrive through fee-based structures (subscription or pay-to-play) for the use of its business productivity software. So far in the current AI cycle, manufacturers of advanced chips like Nvidia have been able to: exponential profitsrevenue from the use of AI software is still very limited. At best, future revenue sources seem likely to be constrained, except for a few successful companies that capture a large share of the market. And if AI becomes a near-standardized commodity, or if alternative, better technologies emerge, both chip makers and software designers could be vulnerable. In this scenario, there would be contraction and consolidation in the AI ​​software and infrastructure industry, but the broader economy could continue to thrive from AI applications. Stock markets will see a decline in the performance of high-tech companies that produce AI software and infrastructure, and a rebalance to companies that apply AI tools to improve their products and increase profitability.
• Even if the expected gains in overall productivity and profitability are achieved, a shakeout may occur in the midst of a sustained boom. As has befallen previous software market leaders such as Netscape and MySpace, the AI ​​sector itself is likely to have a few big winners and many losers given network effects and economies of scale. Even ChatGPT recentlycode red” That's all Threats to AI leadersit may fall by the wayside. Similarly, many companies in industries that leverage AI may fall behind others that are more successful in applying newly available AI tools, or be absorbed by leaders in AI adoption, a familiar pattern from past cycles. And workers more generally will need to adapt to the fact that many companies will no longer need many employees to perform routine tasks such as basic coding and reporting. This will require improving AI skills or finding alternative activities that are less likely to be replaced by AI, such as practical creative or manual skills.
• each one different cycleswhile it is important to recognize certain features of the current experience, economic history suggests some specific signals that are useful to watch for in order to predict how the current AI boom will end as it continues to evolve. Do the adoption of AI tools and the efficiency of AI in operations continue to dramatically improve? so far, The user base continues to grow rapidly and the efficiency of new models continues to improve exponentially. Will it continue like this? moore's law Has it been 50 years? If this is not the case, the likelihood of a “disappointment” scenario increases, constraining further adoption and profit flow in the AI ​​space. Similarly, are business users of AI tools achieving significant productivity gains and amazing new innovations to justify significant AI spending? Research by MIT Media Lab and McKinsey So far, most companies have yet to realize a meaningful return on their investments, but these efforts nonetheless signal a way forward. Are the market prices of “hot stocks” driven primarily by expectations for further short-term price momentum beyond rational expectations of future earnings? Such behavior will most likely lead to an inevitable reversal and a market crash similar to the one that occurred during the “dot com bubble.”