search
HomeTechnology peripheralsAIAI-style 'technology diffusion”: 10 'lessons learned” from the past 100 years

AI-style 'technology diffusion”: 10 'lessons learned” from the past 100 years

Jun 05, 2023 am 11:49 AM
aitechnology diffusionLessons from the past

With the rise of AI technology, a new generation of technological revolution may be just around the corner. However, humans have left some lessons in past technological revolutions. What can we learn from these histories?

On June 2, Morgan Stanley equity strategist Edward Stanley used technology diffusion theory to analyze in his report. This theory studies how technology is first commercialized and applied, through the stages of vigorous promotion and widespread adoption. The process until finally being eliminated due to falling behind.

Technological diffusion occurs after technological progress and technological innovation. From the perspective of human history, technological diffusion plays a vital role in the process of technological progress. Because if a technological innovation has not been widely applied and promoted, it will not affect the economy in a material form.

Accordingly, Morgan Stanley believes that The diffusion rate of artificial intelligence technology has exceeded any previous epoch-making technology, which means more investment opportunities and more problems.

For example, there are regulatory issues, as well as those old technology companies that are disrupted (their stock prices may never recover), and there are deflationary issues caused by new technologies that increase productivity.

1#Development speed and regulatory issues

With the development of technology, each new generation of new technology accelerates its evolution based on the previous one, and the time required is continuously shortened.

Morgan Stanley compares the adoption curves of industries that harnessed electricity after 1885, the Internet after 2007, and artificial intelligence after 2022 since their respective "iPhone moments" (i.e. when new technologies are implemented at scale). the time required for application).

Data shows that from the "iPhone moment" of each of these three technologies to the penetration rate exceeding 10%, it took 20 years for electricity, 7 years for the Internet, and only 1 for generative artificial intelligence. Year.

To reach 30% penetration and radiate to adjacent fields, electricity will take 30 years, the Internet will take 15 years, and artificial intelligence may be less than half.

AI-style technology diffusion”: 10 lessons learned” from the past 100 years

Faced with such a fast speed, discussions on regulation by governments around the world are heating up. At present, the European Union, as a pioneer, is about to launch a regulatory bill.

2#Download speed>Uplink speed

Looking at the 80 structural positive and negative adoption curves over the past 50 years, Morgan Stanley pointed out that two things are clear:

  • Downside share losses in disrupted industries occur faster, with peak activity declining by approximately 20% in the first 5 years relative to emerging technologies that leverage the Internet as a distribution channel.
  • Share declines in disrupted industries are also more severe over longer periods of time, falling an average of 40% from their peak over 15 years, while new entrants’ market share increases by 30% over the same period.

AI-style technology diffusion”: 10 lessons learned” from the past 100 years

#3 Stock downside

The release of new technologies is a fatal blow to old manufacturers.

Morgan Stanley pointed out that after the iPhone was released, the stock prices of first-generation mobile phone manufacturers fell by an average of 50% within 2 years and by an average of 75% within 5 years.

These companies can shed the "challenged" label, but hybrid transformation and capital investment will take time.

For example, Motorola's stock price fell for 2.5 years after the iPhone was released before reaching its lowest point, and it took eleven years to recover to the level before the iPhone was released.

For other companies that are considered to be affected by new technological changes, their stock prices may fall rapidly in the short term, even if their financial data remains normal.

For example, HMV, a former phonograph manufacturer, saw a significant decline in profits only after its stock price fell for 7 years.

AI-style technology diffusion”: 10 lessons learned” from the past 100 years

4# Stock upside

Morgan Stanley noted that it is also common for companies considered to be "winners" to see share prices rise by 100%, but consistent sales and earnings upward revisions are usually required within 6 months to maintain the multiple. expansion.

Apple’s EV/Sales doubled within 6 months after the iPhone was released. In fact, more than half of this happens before launch, not after.

Meanwhile, the consensus stock price increased only 20% during this period and only increased 15% from the launch date to the end of 2007.

AI-style technology diffusion”: 10 lessons learned” from the past 100 years

The real rift between Apple and the established companies occurred over the next 5 years, with Apple's valuation falling back below where it was when the iPhone was launched, but with consensus sales forecasts 8x higher.

For established companies, although the market consensus is that their sales are stable, their average valuations are still above 90%.

#5 Hype cycles are now the norm, not the exception

Morgan Stanley pointed out that theme bubbles tend to have a 3-year correction period after peaking, which is a faster cycle than macro themes, which tend to recover after 4 years.

The current rally is an outlier compared to the 70 hype cycles we have tracked over the past 100 years. The question now is, to what extent is this hype cycle stickier than past hype cycles?

AI-style technology diffusion”: 10 lessons learned” from the past 100 years

For this we will be paying close attention to Google Trends:

• There was an initial surge in activity for code generation AI tools such as GitHub Copilot, but this has fallen back to less than 45% of peak search interest.

•Image generation AI (such as Midjourney) was launched before text models, and its interest dropped from its peak to 50%.

• Text generation AI (e.g. ChatGPT/Bard) was last deployed in November 2022 and is proving to be stickier than the methods above, but we will continue to track consumer engagement.

#6 Find GPT’s “killer app”

When the iPhone was released, it was clear which manufacturers would be disrupted, but the final "winners" at that time may be new startups, most of which use hardware as a new point of contact with consumers.

AI-style technology diffusion”: 10 lessons learned” from the past 100 years

From the innovation of more than 50 "platform applications", Morgan Stanley concluded that "killer applications" often take about 1.6 years to emerge from new disruptive technology platforms:

Of the $13 billion deployed by these startups to date, most of it has gone to large language model makers like OpenAI, Anthropic, and Adept, to name a few. These are the unicorns in this space.

The remaining approximately 20% of the funding is for downstream applications such as HuggingFace and other leading vendors and platforms supporting new centralized and open source LLMs.

It is within this latter category of funding that we expect to see killer apps emerge within the next two years.

#7 The “winner” almost “takes it all”

Morgan Stanley pointed out that over the past 100 years, market capitalization leadership has been more susceptible to change. Stocks that are in a leading position in technological changes often no longer are leaders 10 years after major technological innovations.

The lessons from electricity and the Internet show that companies that take advantage of automation and electrification early can gain huge development and value in the market.

The diffusion of transformative technologies often leads to changes in stock market leadership, Since 1990, 2.3% of companies have generated $75 trillion in shareholder returns.

AI-style technology diffusion”: 10 lessons learned” from the past 100 years

For potential buyers who feel that the upside to AI is already priced in, History shows that even missing out on the first year of a long-term trend doesn’t reward it with “winner” stocks for many years What impact does the return have.

#8 High prices for pure investment stocks

"Pure investment stocks" refer to stocks of companies that are only engaged in a specific industry or field. The main business activities and sources of income of these companies are concentrated in a specific industry, so the performance of their stocks is highly correlated with the performance of their specific industry or sector.

Morgan Stanley pointed out that In the long term, stocks related to topics such as artificial intelligence are expected to be valued at 10-50% higher than "non-pure" stocks, with an average premium of 25%.

AI-style technology diffusion”: 10 lessons learned” from the past 100 years

Although (generative) AI does not appear as a stand-alone theme in our sustainable solutions database, we still recommend that investors focus on those with: (1) The highest and most sustainable barriers to entry, (2) the best data and quality, (3) the lowest internal risk, and (4) the purest business model regarding AI exposure that generates revenue upside or sustainable cost reduction enterprise.

#9 Increase productivity

Morgan Stanley said that after widespread industrial adoption of electricity and the spread of the Internet, technology diffusion tends to lead to a 2.5-fold increase in productivity (measured by GDP per capita).

A growing body of corporate communications and academic literature proposes productivity gains across industries, ranging from a 55% developer productivity gain using GitHub Copilot to a 14% gain using generative AI Copilot Improved contact center agent resolution rates, to a 26% increase in legal use cases, and a 79% increase in patient-physician empathy and response engagement using ChatGPT compared to a control sample.

AI-style technology diffusion”: 10 lessons learned” from the past 100 years

#10 Deflation problem

Morgan Stanley proposed that technology is deflationary ("Technology is deflationary"), which means that technology can promote deflation.

First, Technology reduces the demand for labor, which puts downward pressure on wages and employment levels, thereby reducing demand for goods and services as workers have less disposable income .

Second, technology allows the production of goods and services to scale up efficiently. If the production of goods can be efficiently expanded to meet the market's current and future demand for the goods, then the prices of those goods will not rise even if demand increases. As technology improves, more and more industries will reach this inflection point, and inflation in the entire market will become weaker and weaker.

One question that needs to be determined, however, is when deflation will occur and how severe it will be.

ChatGPT and other generative AI text LLMs are ideally suited to bring efficiency gains to industries that have been structurally inflated over the past 30 years, especially education, healthcare, legal, finance, construction, and licensing. How much of these efficiency and deflationary benefits will accrue to companies remains to be seen, depending on the robustness of their barriers to entry.

AI-style technology diffusion”: 10 lessons learned” from the past 100 years

We are more interested in how the emerging discussion about virtual agents will exacerbate enterprise customer churn and subscription business challenges.

The above is the detailed content of AI-style 'technology diffusion”: 10 'lessons learned” from the past 100 years. For more information, please follow other related articles on the PHP Chinese website!

Statement
This article is reproduced at:搜狐. If there is any infringement, please contact admin@php.cn delete
What is Graph of Thought in Prompt EngineeringWhat is Graph of Thought in Prompt EngineeringApr 13, 2025 am 11:53 AM

Introduction In prompt engineering, “Graph of Thought” refers to a novel approach that uses graph theory to structure and guide AI’s reasoning process. Unlike traditional methods, which often involve linear s

Optimize Your Organisation's Email Marketing with GenAI AgentsOptimize Your Organisation's Email Marketing with GenAI AgentsApr 13, 2025 am 11:44 AM

Introduction Congratulations! You run a successful business. Through your web pages, social media campaigns, webinars, conferences, free resources, and other sources, you collect 5000 email IDs daily. The next obvious step is

Real-Time App Performance Monitoring with Apache PinotReal-Time App Performance Monitoring with Apache PinotApr 13, 2025 am 11:40 AM

Introduction In today’s fast-paced software development environment, ensuring optimal application performance is crucial. Monitoring real-time metrics such as response times, error rates, and resource utilization can help main

ChatGPT Hits 1 Billion Users? 'Doubled In Just Weeks' Says OpenAI CEOChatGPT Hits 1 Billion Users? 'Doubled In Just Weeks' Says OpenAI CEOApr 13, 2025 am 11:23 AM

“How many users do you have?” he prodded. “I think the last time we said was 500 million weekly actives, and it is growing very rapidly,” replied Altman. “You told me that it like doubled in just a few weeks,” Anderson continued. “I said that priv

Pixtral-12B: Mistral AI's First Multimodal Model - Analytics VidhyaPixtral-12B: Mistral AI's First Multimodal Model - Analytics VidhyaApr 13, 2025 am 11:20 AM

Introduction Mistral has released its very first multimodal model, namely the Pixtral-12B-2409. This model is built upon Mistral’s 12 Billion parameter, Nemo 12B. What sets this model apart? It can now take both images and tex

Agentic Frameworks for Generative AI Applications - Analytics VidhyaAgentic Frameworks for Generative AI Applications - Analytics VidhyaApr 13, 2025 am 11:13 AM

Imagine having an AI-powered assistant that not only responds to your queries but also autonomously gathers information, executes tasks, and even handles multiple types of data—text, images, and code. Sounds futuristic? In this a

Applications of Generative AI in the Financial SectorApplications of Generative AI in the Financial SectorApr 13, 2025 am 11:12 AM

Introduction The finance industry is the cornerstone of any country’s development, as it drives economic growth by facilitating efficient transactions and credit availability. The ease with which transactions occur and credit

Guide to Online Learning and Passive-Aggressive AlgorithmsGuide to Online Learning and Passive-Aggressive AlgorithmsApr 13, 2025 am 11:09 AM

Introduction Data is being generated at an unprecedented rate from sources such as social media, financial transactions, and e-commerce platforms. Handling this continuous stream of information is a challenge, but it offers an

See all articles

Hot AI Tools

Undresser.AI Undress

Undresser.AI Undress

AI-powered app for creating realistic nude photos

AI Clothes Remover

AI Clothes Remover

Online AI tool for removing clothes from photos.

Undress AI Tool

Undress AI Tool

Undress images for free

Clothoff.io

Clothoff.io

AI clothes remover

AI Hentai Generator

AI Hentai Generator

Generate AI Hentai for free.

Hot Article

R.E.P.O. Energy Crystals Explained and What They Do (Yellow Crystal)
3 weeks agoBy尊渡假赌尊渡假赌尊渡假赌
R.E.P.O. Best Graphic Settings
3 weeks agoBy尊渡假赌尊渡假赌尊渡假赌
R.E.P.O. How to Fix Audio if You Can't Hear Anyone
3 weeks agoBy尊渡假赌尊渡假赌尊渡假赌
WWE 2K25: How To Unlock Everything In MyRise
4 weeks agoBy尊渡假赌尊渡假赌尊渡假赌

Hot Tools

SublimeText3 Chinese version

SublimeText3 Chinese version

Chinese version, very easy to use

mPDF

mPDF

mPDF is a PHP library that can generate PDF files from UTF-8 encoded HTML. The original author, Ian Back, wrote mPDF to output PDF files "on the fly" from his website and handle different languages. It is slower than original scripts like HTML2FPDF and produces larger files when using Unicode fonts, but supports CSS styles etc. and has a lot of enhancements. Supports almost all languages, including RTL (Arabic and Hebrew) and CJK (Chinese, Japanese and Korean). Supports nested block-level elements (such as P, DIV),

DVWA

DVWA

Damn Vulnerable Web App (DVWA) is a PHP/MySQL web application that is very vulnerable. Its main goals are to be an aid for security professionals to test their skills and tools in a legal environment, to help web developers better understand the process of securing web applications, and to help teachers/students teach/learn in a classroom environment Web application security. The goal of DVWA is to practice some of the most common web vulnerabilities through a simple and straightforward interface, with varying degrees of difficulty. Please note that this software

Dreamweaver Mac version

Dreamweaver Mac version

Visual web development tools

SecLists

SecLists

SecLists is the ultimate security tester's companion. It is a collection of various types of lists that are frequently used during security assessments, all in one place. SecLists helps make security testing more efficient and productive by conveniently providing all the lists a security tester might need. List types include usernames, passwords, URLs, fuzzing payloads, sensitive data patterns, web shells, and more. The tester can simply pull this repository onto a new test machine and he will have access to every type of list he needs.