The epochs of the dramatic technological changes, similar to the current AI invasion, happened before. The introduction of gunpowder firearms and open-sea sailing ships, steam machines and electrical power, nuclear energy and computers – all preceding waves of dramatic technological changes – had similar patterns. It is not surprising that AI diffusion repeats them. The common pattern of arrival and implementation of the radical technological innovations looks like a steep hill and quagmire.
The innovation’s capability is the hill, while its employment is the quagmire. The first effective gunpowder mixtures appeared in the mid-14th century, and only in the late 15th and early 16th centuries did they deliver some siege and battlefield advantages. The rig of sails and ship design for blue water navigation were introduced in the late 14th century, and only a century later the epoch of Geographical Explorations came.
The workable steam machine was built in the mid-18th century, but steam power became the principal driver of the industrial equipment only a half-century later. Electricity was experimented with since the middle of the 19th century, but it was employed for practical use only at the turn of the 20th century.
The nuclear power repeated the pattern a half-century later, and the computer, invented in the mid-20th century, became a widely used technology only in the 1980s. The gaps between the introduction of the new technology and its common employment narrowed but remain nevertheless. Employment of the radical new technology is the quagmire invariably. The problem is not its capability but the technique and tactics of employment. There is the inevitable period of making the useful devices between the technology’s practical readiness and its widespread implementation.
The devices are necessary for the implementation of the new technology; however, unlike the introduction of the technology itself, they are the products of not an inventing genius but the interaction of the manufacturers and final users. Until the final users signal to the manufacturers what they expect of the new technologies and what for and how they are going to employ them, the usable devices will not appear. Today, it is a key obstacle for LLMs’ employment. Currently, implementation of AI is the process of proposal and not demand.
The demand of the final users for the AI-equipped tools does not shape up in the same way, as infantry tactics were not ready for firearm deployment in the 15th century, and naval tactics were not ready for steamers’ deployment in the early 19th century. The final users look at AI as an appendix to their routine, although it is ready to turn the routine over. Today the focus of AI development has to move from refining the LLMs to the “tactics and technique” of usage of the AI-driven equipment.
The new generation of AI practitioners have to grow. Not the AI training engineers must employ AI for practical tasks in the same way as not powder-mixing monks commanded infantry with firearms and composed its ranks. Only when this new generation of applicants arrives could AI be judged for its efficiency and effectiveness. In “Strategy by AI”, we develop the methodology for AI employment for strategic practice of the organisations – from strategic analysis through the strategy’s making to its implementation. Its analytical and controlling results confirm the effectiveness of well-defined, systematic AI practice. www.strategybyai.org