AI
Research5
According
to an 11/1/24 Financial Times article,
“Big Tech’s capital spending is on track to surpass $200bn this year (about 1%
of 2020 GDP) and is set to rise even further in 2025. This is a fantastic
amount of money to pour mainly into data centers, computing a mathematical
vector similarity technology, that mimics the real world. In a subsequent
article, we shall examine in detail why precise predictions,
except in very restricted systems, are not possible. But the generative AI bill
is now coming due, as Wall Street now looks for profits.
“Microsoft
said its revenue from AI was on the brink of hitting an annualized $10bn.”
Microsoft, however, is an exception in its specificity; but it estimates that
total 2025 revenue will grow by only 6.3%. “…and few other software
companies have revealed anything about the effects of AI on their
revenue…leaving the stock market to fret. And investors are freaking out about
the costs…”
MIT
economist and Nobel Prize winner, Daron Acemoglu, does highly practical
research. In “The
Simple Macroeconomics of AI” (2024), he estimates only a very modest
improvement in the economy from AI, “…an (task-based) increase of about
0.53-0.66%” over a total of 10 years, without substantial additional
inputs of labor and capital. However, “…if AI is used for generating new tasks
for workers, it can have more beneficial productivity, wage and inequality
consequences, and it can even increase wages…. Many production workers today,
including electricians, repair workers, plumbers, nurses, educators, clerical
workers, and increasingly many blue-collar workers in factories, are engaged in
problem-solving tasks. These tasks require real-time, context-dependent
(our note, this is really an important point that we shall cover in a future essay)
and reliable information…Reliable information that can be provided
quickly by generative AI tools can lead to significant improvements in
productivity.”
But that
improvement will require a significant reorganization of the workplace. “My
assessment is that that there are indeed much bigger gains to be had from
generative AI, which is a promising technology, but these gains will remain
elusive unless there is a fundamental reorientation of the industry, including
perhaps a major change in the architecture of the most common generative AI
models, such as the LLMs, in order focus on reliable information that can
(also) increase the marginal productivity of different kinds of workers, rather
than prioritizing the development of general human-like conversational tools.
The general-purpose nature of the current approach to generative AI could be
ill suited for providing such reliable information.”
Current
generative AI may not give people in production systems the information that
they actually need. More about this later.