The first theories on artificial intelligence appeared in 1955. That same year, Fortune magazine launched the annual ranking of the top 500 American companies. Today half of them have disappeared because they failed to take the onramp to the digital revolution. With the emergence of AI in all sectors, will organizations be able to adapt to this new technological breakthrough? And more precisely, in the ERP universe, what can we expect from AI?

For some, artificial intelligence appears to be the most important technological revolution in history. It is true that it increases the natural knowledge of humankind, pushes back the limits of automation and shakes up innovation in the entire society.

Narendra Mulani, Head of Applied Intelligence at Accenture says, "AI is not an exponential technology, but it is the combination of AI technologies that delivers the exponential results." Its rise is based on the emergence of Big Data and the almost infinite processing power of Cloud computing. AI is therefore more introduced as a set of distinct and evolving technologies, deployed in four fields: perception, understanding, action and learning. Nowadays, AI is already widely at work in our daily lives: Siri on iPhone, Google Home, chatbots, music or film recommendations on platforms...

Two main currents

To gather a global view of the situation, let's go back to the origins where two currents found the evolution of artificial intelligence, symbolic AI and connectionist AI. Intellectual rivals for a long time, the second current lords these days, explains Olivier Ezratty, consultant, author and speaker, a specialist in innovation and innovation ecosystems. Very trendy in the 80s, symbolic AI manages both formal reasoning and logic. It is found in rule engines and expert systems. It is demonstrable because its logic can be dissected. However, it has been clearly in decline for the past fifteen years. Connectionist AI, continues Olivier Ezratty, is embodied by machine learning, neural networks and deep learning which train on the basis of data and are probabilistic approaches. The connectionist current dominates AI advances in data analysis, language and image processing. We also owe to it the premises of reasoning. As for Artificial General Intelligence (AGI), it is the next strategic issue for research in AI, seeking to complete the synthesis between the symbolic and the connectionist.

Weak AI and strong AI

Going further, it should be noted that there is a distinction between weak AI and strong AI. The former is built around human-created algorithms. It is based on machine and deep learning. It focuses on performing specific tasks through the analysis of existing data. Strong AI, still in the experimental stage, would be able to supplant the human brain by being endowed with consciousness and mind, even capable of feeling emotions. One is still far from the memorable robots with integrated emotions such as HAL* in "2001: A Space Odyssey" by Kubrick (1968) or David in "IA" by Spielberg (2001). “We have built systems capable of recognizing cats with a 95% success rate by providing them with 100,000 images of the animal. But a child only needs two cat images to identify one all his life, with a 100% success rate", points out Luc Julia, Siri's co-creator. AI is, however, truly here to stay. It provides services in all sectors, from environment to defense, including health and education. Drifts off the ‘right’ path are possible, of course. Also, the way between liberties-killing demonization and blissful technological optimism will be found alongside the uses to be made of this technology.

And what about ERP in all this?

Already, in the corporate world, everyone agrees that AI and ERP possess everything to get along. 75% of executives state they will implement AI within three years’ time. In business, AI's vocation is to expand the capacity of machines to handle tasks for which human intelligence is required. Its application is illustrated in machine learning, also known as self-learning. The machine bases itself on similarities to recognize an object or to differentiate it from others. Thanks to AI, machines will extend their capabilities to automate increasingly complex tasks and require less and less validation by individuals. In this sense, ERP has all the elements necessary for self-learning such as data history, traceability of modeling iterations and the modeling of task processes.

Augmented by AI, ERP is able to provide employees with much more ease and fluidity in the various processes. On repetitive tasks, thanks to machine learning, the software will be able to make the right decision every time. More time will therefore be left for employees to devote themselves to more strategic work. Eliminating repetitive, time-consuming and low-value-added tasks, ERP combined with AI will also detect inefficient processes. It will be able to suggest solutions to improve them and reduce costs. AI will also make possible predictive diagnostics, to minimize waste of resources, facilitate all projections and assist organizations to better adapt.

Among the expected changes for AI-aided ERP, centralization of data in an optimized manner is a recurring theme. As a guarantee of flexibility, one can mention these sensors which, placed in the right spot, will be able to collect all types of data in real time to feed them directly into the ERP databases.

Relative to the occurrence of incidents, AI will allow the establishment of alerts capable of analyzing the potential impact of the event on the immediate or the future activity of the company. In the area of IT security, intelligent ERPs with facial or voice recognition will be able to prevent anyone from accessing unauthorized data.

When it comes to ERP modules, one can also expect noticeable effects due to artificial intelligence. For HR, automation of repetitive tasks will generate significant time savings in the payroll and recruiting processes. More time is expected to be saved with the handling of sourcing and the sorting of work applications. Fed with all kinds of possible information, the machine will be able to scan and cross-reference millions of pieces of data to select the best candidates. With IA, databases will also be able to self-clean themselves of any unnecessary data.

An inescapable digital future

Artificial intelligence in the ERP realm offers more insight, performance and automation opportunities for businesses. With a view to optimizing business processes, all departments will be impacted: purchasing, sales, marketing, logistics and of course, finance. For their part, finance ERPs must integrate data of a very diverse nature and granularity, but above and beyond make them usable by businesses that are increasingly demanding in terms of operational efficiency and analytical capacity. When publishers enrich their ERP with functionalities such as RPA (robotic process automation) and AI/machine learning, there is no longer any doubt that the finance of the future will be digital. However, there are still reservations, especially in the face of the loss of the human connection inherent in the prospect of total automation of ERP systems.

Whatever happens, the clear progression of AI in the company is for real. The Boston Consulting Group and MIT Sloan Management Review's study of 3,000 managers, executives and researchers from organizations and businesses with revenues of over USD 100 million is enlightening. Of these, 57% say they use AI. This proportion is in net progression, as it was 44% in 2018. Plainly, the growing share of artificial intelligence will lead the company to truly real evolution. In terms of ERP, the ground is fertile since, beyond the automation of processes and predictive diagnostics for better management of resources, AI opens up towards a significantly improved user experience.

(*) HAL 9000. Some sharp-eyed observers have discovered that this name is composed of the letters that in the alphabet precede these: I, B, M. For author A.C. Clarke and the moviemaker, this is unplanned. HAL is the acronym of “Heuristically programmed ALgorithmic computer”. In the French version, HAL 9000 becomes CARL 500 of “Cerveau Analytique de Recherche et de Liaison [Analytic Research and Linking Brain].”

Sources: datascientest.com, accenture.com, figarovox/tribune, journaldugeek.com, oezratty.net/opinions libres, insuranceblog.accenture.com, FitnetManager.com, pegase-caraibe.com, geniuserp.com, archipelia.com, Wikipedia.