I Solved the Solow Paradox: Technology and Economic Productivity in AI GPT, money, science, and manufacturing
Asher Idan, Ph.D
Summary
Solow’s paradox is the most important paradox in the technology economy. The paradox is that although technology makes life more and more efficient, economic productivity (productivity is more production with fewer resources) sometimes does not increase with technological improvement. The solution I developed to the paradox lies in understanding the interrelationship between technology (Rich) on the one hand, and the users (Reach) in the firm and society in general on the other. The computer contributed to the productivity of the firm and the country, only when the Internet spread the computer to the masses. When and which technology will spread Bitcoin and GPT technologies to the masses? This is the most important question for the next decade.
1, Introduction: The paradox and the research question
A series of studies over the past year show how GPT can increase productivity (more output, with less input) for people doing different jobs. On the other hand, total productivity is still not increasing, neither in the last two years of the GPT revolution nor in the last twenty years since the iPhone and social media revolutions. Welcome to the Solow Paradox.
Economists at Stanford and MIT found that call center workers are 14% more productive when they use AI call assistance; In particular, there was a 35% improvement in the performance of inexperienced and low-skilled workers. Another study showed that software engineers could code twice as fast with the technology.
So why does it still not affect the general productivity? Does it take time for technological revolutions to be assimilated into the general population, and only then do they increase productivity? how long does it take? Does this paradox also exist in other technologies such as Bitcoin and social networks?
For example, in the field of manufacturing, for many years, it was one of the most important sources of productivity growth in the American economy. It still forms a large part of the country’s research and development. And recent increases in automation and the use of industrial robots might suggest that manufacturing is becoming more productive — but that hasn’t been the case. For somewhat mysterious reasons, US manufacturing productivity has been a disaster since about 2005, which has played a large role in slowing overall productivity.
2, productivity in manufacturing and research and development
In manufacturing: In a paper published in March, a team of MIT economists and mechanical engineers (including Acemoglu and Ahmed) identified many opportunities for generative AI in design and manufacturing, before concluding that “current [artificial intelligence] solutions cannot lead to productivity, due to several key deficiencies.” The main drawbacks of ChatGPT and other AI models are their inability to provide reliable information, their lack of “knowledge in the relevant field” and their “lack of awareness of industry standard requirements”. The models are also not designed to address the spatial issues on production floors and the various types of data generated by production equipment, including old machines.
In research and development: Several years ago, a team of leading economists wrote a paper called “Is it Harder to Find New Ideas?” And they discovered that it takes more and more researchers and money to find the types of new ideas that are the key to preserving the progress of technology. The problem, in technical terms, is that the output of research — the output of ideas given the number of scientists — is rapidly declining. In other words — yes, it’s harder to find ideas. We’ve generally kept pace by adding more researchers and investing more in R&D, but overall research productivity in the US itself is in deep decline.
To keep up with Moore’s Law, which predicts that the number of transistors on a chip will double roughly every two years, the semiconductor industry needs 18 times more researchers than it had in the early 1970s. Similarly, it takes many more scientists to invent roughly the same number of new drugs than it did a few decades ago.
The paradox is exacerbated because, on the other side, “it looks like a goldmine of new things -” an order-of-magnitude expansion in stable materials known to mankind,” DeepMind researchers wrote in Nature. DeepMind’s database, called GNoME (an acronym for “Graph Networks for Materials Research” ), “equivalent to 800 years of knowledge”, according to a press release from the company.
Before joining Google in 2022, Manyika spent several decades as a consultant, researcher, and eventually chairman of the McKinsey Global Institute, the economic research arm of the consulting giant. At McKinsey, she became a leading authority on the connection between technology and economic growth, drawing on Robert Solow, the MIT economist who won in the 1987 Nobel Prize for explaining how technological progress the main source of productivity growth is.
https://www.technologyreview.com/2024/08/20/1096733/how-to-fine-tune-ai-for-prosperity/
Among the lessons of Solow, who died late last year at the age of 99, is that even powerful technologies can take time to affect economic growth. In 1987, Solow said: “You can see the computer age everywhere except in the productivity statistics.” At that time, information technology underwent a revolution, seen mainly with the introduction of the personal computer. However, productivity, as measured by economists, has been sluggish. This became known as the Solo Paradox. It was only in the late nineties, decades later !!! The birth of the computer age, the growth in productivity finally began to take hold. Why only in the late 90s? Is it related to the distribution capacity of the Internet? My answer is an unequivocal yes!
3, The solution to the paradox: how will the ability to distribute and decentralize the Internet’s networks accelerate both Productivity and social prosperity, for example in GPT and Bitcoin? Ray Dalio and Plan G
Two articles that appeared this year from two completely different directions, can solve the productivity paradox.
3.1 Ray Dalio’s article deals with the two negative side effects of technologies:
First, the Solow paradox according to which technologies that do not reach a critical mass of workers will not affect economic productivity.
Second, technologies that do not reach the majority of society create a struggle against the new technology by those who do not benefit from the technology.
3,2 The second article by PLAN G deals with the Bitcoin network in particular, and networks in general. The article shows:
First, Moore Law (and I will extend this to any technology) greatly increases productivity because it doubles every two years (one generation) the output of the computer’s processor, without increasing the investment (or the price): 2^n where n indicates the number of generations.
Second, the Metcalfe Law greatly increases the distribution of the new technology both among the workers, which increases economic productivity, and among the entire population of users, which reduces social conflicts. Distribution grows in the network by the number of users squared N².
Moreover, the combination of the exponential multiplication of the productivity of processors, with the exponential doubling of the distribution of technology, creates a spiral like a 2^n&N² snowball.
