A booming technology, Artificial Intelligence (AI) is much more than a passing trend! With more and more organizations taking an interest in it, it is because its use represents numerous advantages. Nuxeo, a specialist in enterprise information management, and Océane Consulting DM (Data Management), an expert in data processing, take stock.
It seems like a long time ago when Yann Le Cun, Director of AI Research at Facebook, stated that "Artificial Intelligence is not even at the level of a rat's brain". And yet, this statement is only from 2018! If, since then, the machine has not yet replaced Man, its effectiveness is no longer in doubt.
PROPERLY DEFINING ARTIFICIAL INTELLIGENCE
For Océane Consulting DM, AI is above all the performance of a human action by a robot. This raises many questions: how to make a machine think, have it perform tasks usually assigned to a human, and do so quickly and efficiently, etc. More and more organizations
are thus turning to AI to have machines perform repetitive, time-consuming tasks with no added value. This is called Robotic Process Automation (RPA).
« For example, in the event of a claim, an individual will be able to send an email to their insurance company. Thanks to AI, a machine will analyze this email, identify its recipient and link it to their customer file, verify the associated contract,
and contact an expert by email in an automated manner », explains Dominique Costardi, Director of AI & Big Data at Océane Consulting DM. « The machine is
capable of performing more actions in a short time compared to a human, which is a guarantee of productivity and efficiency for organizations. »
INCREASINGLY ADVANCED USES
In detail, it is thanks to machine learning and access to Big Data that AI is able to train itself on process or document models and improve on a daily basis. Adapted to ECM, its uses are particularly varied. Among other things, AI is able to perform semantic analysis to verify the compliance of certain sensitive documents
(contracts for example) with current regulations. Along the same lines, the machine is capable of analyzing the sounds in videos to extract information from them, and even to understand the emotions that emerge from images: this is « affective computing ».
But AI doesn't stop there! If businesses are so keen on it, it is also because it contributes greatly to hyperpersonalization, which internet users love. The Netflix streaming platform, in particular, understood this well: from registration, the user is prompted to provide their film and series preferences in order to obtain viewing recommendations. AI is also very widespread in chatbots. While old scripted conversational agents could only answer 30% of requests, the most recent ones
are very promising: CloE, the virtual banking assistant from Caisse d'Épargne, reduced support calls by half.
PERSONALIZED AI TO GIVE MORE MEANING TO DATA
Aware of the impact of AI on their sales and customer satisfaction, more than 90% of companies plan to implement it in personalization. However, not all types of AI are equal. While generic AI uses public services for simple use cases (resulting in non-customizable generic models), personalized AI understands business-specific features
and can deliver intelligent content by enriching it with the organization's data. The customized models created by this process bring real added value on a daily basis.
And while GAFAM (Google and Amazon leading the way) rather propose generic service platforms, another publisher distinguishes itself by its approach based on personalized AI. Published by Nuxeo, the Nuxeo Insight solution brings genuine added value to information analysis. For a fashion photography example, generic AI will only be able to return certain data of little real value (indoor shooting, presence of clothing, etc.); on the contrary, the tool developed by Nuxeo, based on personalized AI, is notably capable of identifying the model or clothing worn. The same applies to text files, whose type (contract, invoice, form, etc.), audience or tone can be determined by the machine, which will automatically trigger associated tasks.
A TOOL THAT HAS ALREADY PROVEN ITSELF
Existing for 2 years, Nuxeo Insight already displays convincing indicators. Thanks to machine learning, analysis of only 50 models is necessary for AI to learn how to retrieve certain key information. For their part, content bots can achieve performance levels between 85 and 97% of correct predictions depending on information complexity, contributing to the return of relevant information.
« With Nuxeo Insight, we make predictions that are multiple, that is, they integrate a set of properties », indicates Grégory Carlin, Nuxeo product team. « We are specific about user needs and ready to deploy in enterprise contexts. We offer continuous learning. We do one-click deployment. The solution is easily auditable, with the ability to quickly see what is performing and what is not. We also involve the user in the implementation of models [active learning, note] and we offer continuous monitoring on the different existing models ».
All advantages that make Nuxeo Insight the ideal platform for meeting the business needs of organizations in all sectors.
