Recently, the image of artificial intelligence has become blurred, and undoubtedly Europe has taken up this topic. Indeed, the European Commission has just presented its proposals for a legal framework for AI in the European Union. The pursuit of a European risk-control model does little to hide the general public’s ignorance of AI in general. Indeed, there is not one artificial intelligence, but several.
The goal is more important than the tool
Inspired by the AI technological aura that has been widely promoted in the press in recent years, some companies have stated that they are AI-friendly, AI pioneers … often because they just deployed a chatbot. However, this was only a step towards integrating a true AI strategy. True digital transformation must be based on transformation of processes in the interests of business efficiency. The sustainability of the company is ensured by the pursuit of increased productivity, increased turnover, automation and improved knowledge (market, customers), including through IA.
Less cosmetic than installing a chatbot on a website, this approach involves close collaboration between CIOs, CMOs, and business teams to create economies of scale. Indeed, it takes time to analyze; because AI is rarely a plug & play technology, even if the symbolic branch shows results faster, thereby eliminating the tedious learning phase that machine learning offers today. This is not a problem in itself, because the time for interdisciplinary training is also important: it will allow you to deeply understand the company’s strategy and the use of its data.
Data: a strategic asset
The untapped potential of unstructured data is greatly underestimated. Unstructured data is data presented without a predefined format, sometimes referred to as dark data (a term not related to the dark web). However, the use of this data is of strategic importance to the company, and awareness of it has emerged recently. A 2020 study by A-Team * shows that only 31% of companies can use and benefit from unstructured data. This is one of the promises of AI to make them useful and effective. However, Evgeny Morozov ** suggests that we do not “drown in the sea of algorithms”: you need to choose the application areas. In 2020, 47% of decision-makers rated their organization as underdeveloped or immature in the digitization of documentary and business processes ***. AI-assisted automation is a concern of many companies and requires thinking about data management.
Data management strategy
Data is strategic and should enable a company to move from a decision-making process based on experience or instinct to a process based on data. Being data driven means using data for leadership, driving strategies … But above all, it means using relevant data. There are many applications for evaluating this data, from consolidating accounts for financial reporting to investigating bank non-compliance; or even automation of information retrieval to increase tenfold the number of tender bids processed …
It is possible that adopting a data-driven approach implies a re-architecture of the company and its operating model. CDO will enable this awareness, even if its role in the long term will also contribute to the implementation of a data culture throughout the company and its business units. It is also a guarantor of global consistency: the required data-driven approach requires a unified and integrated vision of data, a vision that an organization cannot live with in silos.
Ultimately, how we use our data tells us not how technology works, but how we decide to make it work. And with the current debate about the ethics of AI, explainability is becoming an issue. The result of transparency is the ability to describe an AI object, its logic and make decisions in a way that is understandable to all for an ethical and useful AI.
* Unstructured Data Management and Benefit. A-Team 2020 https://a-teaminsight.com/
** MOROZOV, E., To solve everything, click here: Aberration of a technological solution, Limoges, FYP editions, 2014.
*** Documentary and business processes: trends for the end of 2022. Marquese 2021.http: //blog.markess.com/2021/01/automatiser-securiser-deux-enjeux-prioritaires-a-relever-dici-2022- en-terms-of-document-management-and-business-process -Trends-2022 /
<< Также прочтите: Этика ИИ: этику нельзя делегировать! >>>