AI-based prediction is a boon to human decision-making that will enhance productivity, income and the quality of life.There is a division of labor between humans and technology. Artificial Intelligence frees people to focus on what they as individuals do best. AI-based prediction improves human decision-making and enhances wealth.
The main
concern of classical economics revolved around how specialization and exchange
make human life better. One of the main advantages of artificial intelligence
(AI) lies in enabling further specialization and division of labor. AI
processes data to predict and humans use the predictions to make decisions.
Let us
unpack this phenomenon now.
Humans
and technology
Early-generation,
or classical, economists were particularly concerned with how the division of
labor along the lines of freely chosen specialization allows people to benefit
from a society-wide chain of value creation. The father of modern economics,
Adam Smith (1723-1790) established that specialization drives surpluses and
innovation. David Ricardo (1772-1823) identified how even the least productive
members of society could benefit from specialization. And even the angry Karl
Marx (1818-1883) saw the advantage of the division of labor. His erroneous
qualm was with the capitalist retaining the benefits of the division of labor
instead of allowing them to flow to the specialized worker.
Specialization,
however, is not restricted to humans. The more technology permeates economic
exchange, the more there is a labor division between humans and technology.
Instead of weaving by hand, most textile producers use machinery for that task.
The machine makes the fabric that has been designed and engineered by people.
Instead of doing arduous calculations ourselves, we let the calculator perform
the arithmetic of models conceived by humans. The Bloomberg Terminal freed the
human analyst from writing quotes on a chalkboard to be able to focus on
deciding how to invest. In short, utilizing technology is about finding its
specialized employment along the value creation chain.
The
technology is becoming cheaper to use, making it increasingly significant
economically.
Technology
enhances the division of labor between humans and machines. It allows each to
specialize even further in what they respectively do best. And this comes to
the greatest benefit to humans. This division of labor enables them to create
novelties, produce more goods and increase quality. By making people able to
focus on what they individually do best, specialization and the division of
labor enhance their productivity and, along with it, their income and the
quality of life.
Making
prediction cheaper
How does
AI fit into this picture? As the prices of weaving machines, calculators, and
computers decreased, they were incorporated into the division of labor. The
same is happening with AI. The technology is becoming cheaper to use, making it
increasingly significant economically. And while AI is making progress in
different applications, it is especially salient in one area – the making of
predictions.
Presently,
predicting is expensive because it involves gathering an enormous amount of
data, analyzing it, identifying patterns, and calculating possibilities. But,
as AI is increasingly employed in performing exactly these tasks, prediction is
becoming less arduous and less expensive. In a virtuous circle, AI is becoming
cheaper itself. If predicting becomes cheaper through AI, the scope of its use
will widen, and its use will intensify.
Prediction
is the process of filling in the missing information. It uses available data to
find patterns to generate new information and calculate probabilities for these
patterns to repeat themselves or change. Prediction is used for traditional
tasks, like inventory management and demand forecasting. More significantly,
because it is becoming cheaper, it is being used for undertakings that were not
until recently prediction problems, such as driving, translating, or medical
care.
The drop
in the cost of AI prediction will influence the value of other things. A
cheaper AI increases the value of such complements as data, judgment, and
action and diminishes the value of its substitutes (human prediction).
AI
prediction
Machines
and humans have distinct strengths and weaknesses in the context of prediction.
Humans, including professional experts, make poor predictions under certain
conditions: they often overweight salient information and do not account for
statistical properties. As prediction machines improve and become cheaper,
businesses will likely adjust the labor division between humans and machines in
response.
Prediction
machines are better than humans at factoring in complex interactions among
different indicators, especially in settings with rich data. As the number of
dimensions for such interactions grows, the ability of humans to form accurate
predictions diminishes, especially in comparison to machines. However, humans
are often better than machines when their understanding of the data generation
process confers a prediction advantage, especially in settings with limited
data. Humans are better at handling uncertainty and significantly better at
making decisions based on prediction.
As
prediction machines make predictions increasingly better, faster, and cheaper,
the value of human judgment will increase.
So, AI’s
specialization in prediction is valuable because it can often produce better,
faster, and cheaper predictions than humans. Prediction is a crucial ingredient
in decision-making under uncertainty, and decision-making is ubiquitous
throughout the economy and social lives. However, a prediction is not a
decision, only a decision component, and another crucial component is judgment.
And here lies the advantage of the human – and therefore, their area of
specialization, human judgment.
It helps
to break down a decision into its components to understand better the impact of
prediction machines on the value of humans and other assets. The value of
substitutes to prediction machines – namely, human prediction – will decline.
However, the value of prediction-making complements, such as human judgment
skills, will increase.
Judgment
involves determining the relative payoff associated with each possible decision
outcome, including those associated with a correct decision and those
associated with mistakes. Judgment requires specifying the objective that one
is pursuing and is a necessary step in decision-making. As prediction machines
make predictions increasingly better, faster, and cheaper, the value of human
judgment will increase because society will need more of it and will value it
better – or price higher. Humans may be more willing to exert effort and apply
judgment to the cases where previously they abstained from deciding (e.g., by
accepting the default situation or setting).
The
quality of AI prediction would most likely increase faster than its price goes
down.
Why is
it that humans are better at judging than AI? While the machine can account for
more data, and therefore, more patterns, to make a prediction, the human has a
gut feeling. Human agents can deal with gaps in a dataset and fill those gaps
based on intuition, and no AI can do that. Also, humans are good at dealing
with and responding to “unknown unknowns” or “black swans.” Arguably, this
skill set comes from two different sources. First, humans see uncertainty as a
resource or a chance leading to opportunities. Second, humans need to take
responsibility for their judgments, which hone their decision-making
capability.
Summary
AI is
best thought of in terms of how it is used and what it does instead of what it
is. Presently, prediction is one of its most vital applications. The cheaper AI
predictions become, and the more and better AI predictions are made, the more
the technology can be used as an asset by economic agents. It could be employed
in tasks specialized in prediction making, which free humans to specialize in
judgment and decision-making activities. This incorporation of AI in a chain of
value creation, utilizing its specialization and division of labor with humans,
is broadly beneficial.
By
allowing the human to focus on what it does best, prediction-specialized AI
increases human efficacy and efficiency and, as a result, income and quality of
life.
Scenarios
Four
broad scenarios emerge from this reasoning.
The
first, base scenario
AI
continues to increase its prediction capabilities, making them better and
cheaper. This outcome could be expected if current AI-related investment and
research trends continue. The quality of AI prediction would most likely
increase faster than its price goes down. As a result, a lower rate of adoption
of AI prediction among economic agents could be expected than the rate of the
technology’s ability increase.
The
second scenario
This
scenario builds on the first. Under this script, the economy will incorporate
AI into various activities, increasing the degree of specialization. Machines
will specialize in making predictions and humans will specialize in making
judgments and deciding. Human decision-making will become better and more
valuable with the increased division of labor within this alliance. Also, the
demand for it will grow. Likely, the increases in decision-making quality will
be faster than its valuation growth. Under what conditions could this scenario
materialize? It hinges on the degree of freedom that economies and individual
entrepreneurs will have in developing and implementing the technology.
The
third scenario
This one
also builds on the first. But it concedes that, at least in the short to medium
run, specialization and division of labor might not occur. There could be
different reasons for this. If economic agents perceive AI as a threat instead
of an asset, they may try to compete against it. Instead of allowing for
specialization along the lines of the different advantages, humans would direct
their abilities to less productive tasks, such as data gathering and
predicting. Another possible reason for this scenario would be regulation
preventing the use of AI or curbing its functions. The third trigger for this
scenario would also be regulation barring or diminishing the economic payback
from using AI. For example, by not allowing investors to finance it or retain
the revenues from its use.
The
fourth scenario
This
scenario does not relate to any of the previous three. Here the focus of AI
changes completely, and it shifts away from predicting to some different
application, e.g., sensors or robotics. In this scenario, AI can still be used
to the advantage or disadvantage of economic agents; AI could even lead to
specialization and division of labor. But in this variant, specialization would
not occur along the lines of machine prediction and human decision but result
from other applications. Similarly, in this scenario, there is an upside and
downside potential. The fourth scenario could materialize if the returns from
prediction fall considerably or if other uses of AI become cheaper faster,
making these uses easier to fit with the chain of added value.
****Henrique
Schneider is the chief economist of the Swiss Federation of Small and
Medium-sized Enterprises as well as professor of economics at the Nordakademie
university of applied sciences in Germany. He serves on several non-executive
boards in Switzerland, Asia, and at the United Nations level.