Technology and automation alone are not enough. It takes intelligence, too. A plant governed by AI transforms production (but also the life of the technical manager!)
Artificial Intelligence: a brief history
Is intelligence an exclusively human, or even biological, faculty or can it be replicated? Discussions and analyses have long revolved around intelligence.
Conscience has always been one of the central focuses of the debate. Can a technical system develop self-awareness? Can a machine act on anything other than mere computation?
If these are fundamental aspects for the very concept of Artificial Intelligence (to delve deeper into the topic, you can read René Descartes, Kurt Gödel, Alan Turing, John McCarthy, Warren McCulloch, Walter Pitts…), the real history of AI originates in the American summer of 1956, at Dartmouth College (NH, USA), when John McCarthy coined the term “Artificial Intelligence” during a meeting between scholars and researchers. During the days of Dartmouth, models such as the Logic Theorist, the General Problem Solver, and the Dendral were presented, which were able to solve various types of problems using logic.
Two years later (1958), McCarthy introduced a programming language dedicated to AI: LISP.
In 1966 came ELIZA, developed by Joseph Weizenbaum, the first chatbot in history that spoke through written messages (PS: an online version of ELIZA still exists today). The conversation was managed through an algorithm but those who chatted with the “supercomputer”had the general impression of conversing with another human.
In 1976, MYCIN diagnosed blood infections. But the turning point came in 1980: R1 (also known as XCON, which stood for eXpert CONfigurer) began to manage computer orders on behalf of Digital Equipment Corporation, saving the Maynard-based company 240 million dollars in about 6 years of operation (the state of Massachusetts is renowned for being a fertile ground for innovation).
In 1997, IBM’s Deep Blue beat chess player Garry Kasparov.
In 2005, some self-driving cars completed in the DARPA Grand Challenge, crossing 100 km of desert.
In 2015 the ConceptNet 4 system passed an intelligence test (one of those designed to determine human IQ) and was demonstrated to have the true abilities of a 4-year-old child (no, that would not be a low score; read on if you want to find out why).
Artificial Intelligence between ethics and legal boundaries
Supranational bodies such as the European Union have worked hard to develop regulatory frameworks that can serve as a reference for the development of Artificial Intelligence. In particular, the recent European Commission document on AI states its mission to: “Accelerate, act and align to seize opportunities of AI technologies and to facilitate the European approach to AI, that is human-centric, trustworthy, secure, sustainable and inclusive AI, in full respect of our core European values”. In addition to this, the central role of human control must be emphasized, mainly because AI systems are designed by people, and it may happen that they carry with them the errors, prejudices or faultylogic of their designers.
Precisely for this reason, the partnership between people and AI is always going to be beneficial, because it ensures maximum efficiency, maximum precision and maximum utility. And we must not forget that this is a continuous process.
Strong Artificial Intelligence and weak Artificial Intelligence
The first machines developed with Artificial Intelligence were based on a concept of weak AI, that is to say, a system that replicates human behavior but which is not truly intelligent. The theory of strong AI is instead based on the assumption of being able to create systems with authentic intelligence, just like human beings.
In the case of weak AI, the symptomatic behaviors of intelligence are replicated through deductive logic. With the progress of technology and based on an approach that sees computers as neural network models, a different model has been developed: “The inductive typology is instead typical of that area known as machine learning, based on statistical techniques, such as the identification of correlations between the variables analyzed and the prediction of new behaviors through regression algorithms. This approach has recently taken a significant step forward with modeling through the use of artificial neural networks in which different levels of representation are identified, similar to what happens in human neural networks. Since this form of machine learning occurs on the basis of the depth of the network, we speak of deep learning.” [Giovanni Amendola].
And sonow, Artificial Intelligence observes, evaluates, decides, foresees, learns and recognizes habits. Which makes it a reliable partner, working 24 hours a day, being fully active in the operation of systems as well as constantly oriented towards objectives.
How Artificial Intelligence in tissue production can change the life of the technical manager
Artificial Intelligence in tissue production can change the life of the technical manager at a plant, because all phases, from procurement to the final consumer using the products, generate lots of data. It follows that, if we can collect, analyze and process these data, something can happen…
Artificial Intelligence in tissue production: procurement
Raw materials arrive at the plant based on specific orders determined by forecasts. Orderrigidity or flexibility can make the difference between making the most of market variability or getting into trouble. Let’s take, for example, 2020: last year, with the pandemic, toilet paper consumption skyrocketed beyond all plausible predictions (Covid-19 is an event that no one could have foreseen). In this scenario, Artificial Intelligence represented a competitive advantage because the companies that organized themselves were able to redefine the entire production process to optimize supply but also to make distribution more efficient. The JIT (Just-In-Time) procurement logic has shown its limits in overcoming a delicate situation such as that of 2020, but the application of AI logic has made it possible to coordinate the reorganization of the entire production process (storage and final distribution included), modifying at the margins downstream to be able to support the increases upstream. There are two keys in a situation like that of 2020: on the one hand, forecast modeling, which some companies have pushed to the limit by creating hundreds of simulations a day; on the other hand, agility, a natural consequence of the adoption of AI systems (flexible by nature), that in scenarios of sudden change has allowed intensifying some production phases and slowing down others.
The advantages? Uninterrupted production, profit maintained, consumers satisfied.
Artificial Intelligence in tissue production: converting
Converting is a critical phase, which clearly affectsthe quality of the final product. Every single step is a potential source of data which, if collected, related to other surveys, and processed, can bring appreciable benefits in terms of process (and profit). Let’s imagine that several consumers report problems and distributors collect these complaints. What does all this have to do with the converting phase? Simple: first of all, if the data are centralized, Artificial Intelligence has the ability to isolate the production lot and make an automatic retro-investigation to determine which were – for that particular lot– the conditions or processing parameters that differed from traditional values. By analyzing the type of problem and the data from the converting phase, AI will be able to isolate the source of the anomaly, simulate the outcome of possible interventions and apply the correction. Finally, AI will be able to initiate an automatic recall of the other products belonging to the same lot (preventing further discontentedconsumers).Is all this possible without Artificial Intelligence? Yes, of course, but time (which is money of course) is the fundamental variable. Doing everything manually while working on a spreadsheet is not an effective method. A problem that is difficult to identify may require weeks of analyses, simulations, checks, crosschecks and interventions. All this would not be sustainable for a company because it would be economically damaging, in addition to the risk that disappointed consumers abandon the brand (in which case recovery becomes more difficult than ever). And then, imagine the difference between presenting yourself to the management board with a problem compared with presenting yourself with a solution 
Artificial Intelligence in tissue production: packaging, storage and distribution
What has been said for the procurement and converting phases can – in essence – also be applied to the other process phases. IoT machines that detect and share data with an AI system in the Cloud allow punctual and in-depth analyses, transforming rough and chaotic data into information, automatic corrections, forecasts and lists of possible interventions. And so every activity – from systems maintenance to stock logics, from machine arrangement to line speed, from energy optimization to product quality controls (in each production step) – becomes a strategic element to improve production. The warehouse is also optimized. The advantages? Process chains that are always active, scheduled downtime, maintenance that prevents breakdowns, in-out flows from the plant that make the most of environmental and economic conditions, elimination of human error.This last point deserves further study. There are errors due to working conditions, excessive stimuli in the plant, multitasking, incorrect forecasts, the omission of variables deemed insignificant, cognitive biases. There are also inevitable errors because they are linked to elements which fly beneath the radar; they are anomalies, imperfections, defects that nobody would be able to detect. But a machine can, and then intervene before such non-compliancegenerates a cascade of economic damage.Among the possibilities afforded by Artificial Intelligence, is the direct connection with customers’ and suppliers’ AI systems to transform the process of product development, and as such introducing servitization, where the outcome is a service rather than a one-off sale.
Artificial intelligence in tissue production: the value of processes
Big data processing opens up the possibility of using automation based on data analysis to make forecast estimates for the anticipated management of specific events. Thanks to AI, it is possible to introduce predictive maintenance, preventing unplanned line stops or component failures due to wear.
Well, imagine yourself heading up a plant where your tireless partner is Artificial Intelligence. And now imagine how calm and collected you would be… Not bad, right?
Focus: Technological Singularity
Among the forecasts for the evolution of machines, is that of the technological singularity (which starts from Moore’s Law according to which the processing capacity of equipment doubles every two years) and which sees the date of 2045 as the starting point of the new era. Singularity is the condition in which computers will be so powerful as to revolutionize the world in every possible aspect, a condition made possible by the fact that at some point computers will be able to build other computers that will be more “intelligent” than human beings, giving rise to an infinite loop.
Many are the discussions around the possibilities (and consequences) of this scenario. In any case, Moore’s Law is already setting the pace today because the development of computing capacity collides with the limits of physics in the building of components.
What matters, however, is the almost absolute degree of perfection that artificial intelligence has achieved to support industrial automation processes.
Among the recommended reading on singularity, Vernor Vinge‘s article coining the term (1993), Isaac Asimov’s Reason in which robots’ self-awareness appears (1941), and Fredric Brown’s short story Answer (1954), which “prophesizes” the building of a supercomputer to which humanity asks for confirmation of the existence of a God (no spoilers, read it if you want to know the answer).
Let us have your opinion on what you have read by leaving a comment but first, please accept the gift we have set aside for you. NB: do not print it if you don’t need to