To begin contemplating the future influence of artificial intelligence, we should go back hundreds of years. Humanity has always used tools and simple machines. At some point in history, machines emerged that were significantly different from primitive tools and mechanisms.
In the 18th century, the first textile machines were created, capable of embroidering according to programmed patterns and weaving materials, essentially performing tasks that artisans once did by hand. In the early 19th century, the first industrial steam engines appeared, followed by the first locomotives on railways.
Try to imagine what people of that era thought of these new inventions. Even the smallest steam engine was more powerful than any living draft animal. Each loom with automatic weft control by Joseph Jacquard could embroider with unparalleled precision, 24 hours a day, seven days a week. A single mechanical loom by James Northrop could produce hundreds of meters of fabric daily, and one worker could easily operate 200 such machines.
But what about jobs for the other 199 workers? People of that era felt anxious about factory jobs, and others worried about their horse-drawn transport companies being displaced by the railway’s incredible freight capacity. Groups even formed to destroy these machines, like the famous Ned Ludd, a fierce fighter against steam engines. In Britain, a powerful lobby of horse-drawn postal workers opposed building railway connections. These groups, however, could not stop technological development, as it brought enormous societal benefits. Those who adapted thrived. Just consider that an annual salary at the automatic looms in Israel Poznański’s factory in Łódź could afford a small apartment, shop, or small farm. People working with large textile machines earned what would have been considered fortunes at the time, gained significant technical expertise, and established a new class of technical workers. The steam railway allowed agricultural products from remote areas to reach markets, significantly reducing rural poverty. Industrial goods became affordable, and clothing widespread. Technological advancement substantially increased societal wealth, fostering growth opportunities. Fortunes grew among the wealthy, and a middle class of small entrepreneurs actively competed for resources. So was it worth fighting the machines?
Artificial Intelligence is the Steam Engine of the 21st Century
After the great steam engine revolution came the revolution of electricity, cars, telephones, television, and airplanes. People stopped fearing new waves of technology. We are now witnessing another wave, the wave of artificial intelligence. How does this new technological revolution differ from previous ones? Previously, it involved replacing human hands with mechanical hands, muscle strength replaced by piston or electric motor power. Gradually, the first fully automatic machines, sensors, and feedback systems appeared, the beginning of primitive artificial intelligence on a single-celled level. For instance, when the temperature in boilers exceeded 120 degrees, the system cut off the heat supply; when material production encountered a snag, the system halted the process. This was the first stage, where humans began overseeing the operation of technical systems.
Artificial Intelligence Today
We are now surrounded by systems that we can define as artificial intelligence. A significant AI system that changed the world was Google’s search algorithm, introduced in 1998. This modest, now indispensable part of our daily lives pushed out most search engines like Yahoo or AltaVista. Google proved to be the most efficient, intelligent information search system, a first display of basic artificial intelligence. I remember times before Google, when finding information was slow and cumbersome. The Google algorithm marked a major improvement in quality of life and was one of the first conveniences arising from AI application.
First Level of Artificial Intelligence – Rule-Based Systems (RBS)
A system based on pre-programmed rules is artificial intelligence fully coded by humans. The best example is a chess program. Programmers input the rules of piece movement and the history of all configurations from countless chess tournaments. The computer, when playing chess, analyzes the current layout based on previously stored historic chess setups. Knowing the rules, it makes decisions, considering the future piece positions after each move. This algorithm relies on historical data and programmed game rules. Such algorithms are single-task systems based on operational research, seeking the best outcome under specific constraints.
Returning to the Google algorithm, the initial version used slightly more advanced contextual systems. Today, Google has advanced significantly, based on a second-generation artificial intelligence algorithm.
Second-Generation Artificial Intelligence – Context Awareness and Retention System (CARS)
With first-generation AI, there was full control over its actions and a certain level of understanding of algorithmic decisions by humans. Furthermore, it was possible to replicate selected AI behaviors using Excel or even pencil-and-paper calculations. With second-generation AI, we sometimes feel as if we’re interacting with something that has feelings or a controversial character. This is, of course, an illusion. While replicating AI’s actions on paper remains theoretically possible, it’s increasingly challenging. A flagship example of second-generation AI is ChatGPT. For a while, I engaged in conversation with this algorithm. When I asked for certain information, it would sometimes, as if out of laziness, refer me to other sources. For instance, when I asked about the number of immigrants coming to Poland between 2000 and 2020, it referred me to the Ministry of Foreign Affairs. When I inquired about specific years, it finally provided the figures, as if hoping this would stop further questions. The inconsistency here is noteworthy, as first-generation AI systems were perfectly consistent. Now, we have an algorithm that responds differently depending on its „mood” and even can… imagine. ChatGPT is capable of generating plausible but fictional responses.
Third-Generation Artificial Intelligence – Domain Specific Mastery System (DSMS)
This generation of AI has existed for a few years. A spectacular showcase of this technology’s potential was the Go tournament, a highly complex ancient Chinese board game with nearly infinite possible combinations. Computers recently defeated the top Go champions. AI algorithms here don’t just gather data and arrange it in context but also build new skills. This is not just interpretation or imitation but specialization—a narrow focus in which no human can compete. At this level, machines learn from each other, playing against one another, sharing insights, and racing in cognitive reasoning.
Fourth-Generation Artificial Intelligence – Thinking and Reasoning AL System (TRALS)
This technology has not yet been developed, but we understand its potential. It hasn’t been built because computers with the computational power to run such complex algorithms in a short time do not yet exist. However, the computing power needed is in laboratory testing phases, particularly in quantum computing.
Fifth Generation of Artificial Intelligence – Artificial General Intelligence (AGI)
This entirely theoretical algorithm represents the next generation of AI. The typical feature of existing and soon-to-be AI is specialization; the algorithm can play chess well but cannot drive a car, manage thousands of warehouse operations but can’t predict currency rates next week. In the fifth generation of AI, we’d have something resembling humans, with their versatility and sensitivity.
The goal of the journal is to provide scholarship that enables
decision-makers to thoughtfully and intentionally engage in the decisions that shape this dynamic.