Delegate the part of your job you dislike! – Artificial intelligence in the office
“Smart machines” were first mentioned among the strategic technology trends by Gartner in 20141. Ever since, some form of artificial intelligence has been in the top 10 every year. By now it has become a real buzzword. Every company in software development uses the technology in one way or another. So, we had a look, what specific applications are there in the field of enterprise content and document management (ECM).
A bit of history
The term artificial intelligence comes from Alan Mathison Turing. The English mathematician formulated the concept of a machine, that could operate by modifying and improving its own programme2, already back in 1935. He is considered as one of the pioneers of the field, that is why the Turing test, by which computers were measured on how “smart” they were, was named after him.
In the early days, the goal of building intelligent machines was to reach or surpass the intellectual abilities of humans. Therefore, it was considered a major milestone when a computer defeated the world chess champion in 1997, the Jeopardy champion in 2011 or the Go champion in 2017.
But let’s quote here Noam Chomsky, professor of cognitive sciences at MIT: "A computer beating a grandmaster at chess is about as interesting as a bulldozer winning an Olympic weightlifting competition."
So, no wonder, that in parallel with the above efforts, artificial intelligence research has attempted to reach two further goals:
- Applied Artificial Intelligence (AI), also known as advanced information processing, that aims to create “smart” systems for commercial use. Such as medical diagnoses systems or analytical / forecasting systems for stock exchange. The success of applied AI became tangible for the first time by the dynamic spreading of Expert systems (knowledge-based systems).
- Cognitive simulation, aiming to use computers for testing theories about how the human mind works. I.e., how people recognize faces or how they recall memories. The results of cognitive simulation researches are now being effectively applied in neuroscience or cognitive psychology.
Artificial intelligence is present in a wide variety of forms
From 2018 to ’19, the number of companies using some form of artificial intelligence-based technology increased almost fourfold3.
But while a few years ago machine learning (ML) was the only alternative to building your own solutions, nowadays AI is reaching organizations in many different ways. Gartner Hype Cycles is a graphic representation of the maturity and adoption of AI technologies and applications. As almost all innovations make bold promises, it’s often hard to discern the hype from what’s commercially viable and decide on when to start using a technology. Therefore Gartner’s report also provides guidance on how the various forms of AI can contribute to solving real business problems and exploiting new opportunities.
The figure shows that automated machine learning (AutoML), chatbots, and intelligent applications are at the phase, when they already have various success stories behind. Thus, expectations for them are high, even though the use of the technology is not yet typical for most companies. Conversational AI (ed: messaging applications, speech-based assistants, chat bots designed to automate communication and provide a personalized user experience) is a field that continues to remain a top priority for companies, that is likely to be driven by the worldwide success of Amazon Alexa, Google Assistant - and others.
Gartner’s graph also points out that, like in case of all new technologies, interest in many applications of artificial intelligence is declining following the initially rising expectations. Especially if they do not deliver the expected results as they spread. In such cases, further investment will only continue if technology providers can improve their products to a level that satisfies the early adopters.
AI can support the entire document lifecycle
And what does Innodox use this technology for? Here are some examples of how artificial intelligence can make enterprise information management way more efficient:
- Processing of incoming information / documents
A company receives hundreds (or depending on the size, thousands) of messages, letters and e-mails a day. These may be orders, requests for quotations, invoices, complaints or other documents. In order to process the information, they contain, you need to digitize the documents (if they are paper based), file them, sort them and find the person responsible: that is, read them and decide on who will get the task.
This last phase is especially labour intensive. While advanced image recognition technologies make it relatively easy to automate this phase of incoming document processing for structured content types (i.e. invoices, delivery notes etc.), the same method does not work for unstructured documents. To process information that does not follow some sort of regularity or has no repetitive elements that a machine could easily recognize, i.e. emails, you need to involve human resources or use artificial intelligence.
- Manage complex workflows
But let's go one step further! Suppose a new customer “knocks in” to a service provider after a successful campaign. The tasks related to opening a new customer account are relatively standard for all service providers: enter customer’s data into the system, digitize related personal documents, assign a product/service to the customer, open the account, prepare related documents (see contract), check data, activate the account, possibly offer additional service, if accepted have it activated and update the customer's file accordingly.
This standard process can be almost completely automated. Clicks associated with account opening, activation and updating can also be performed by a software robot (RPA). Automated verification of information can be accomplished with artificial intelligence, while intelligent recommendation systems based on data mining methods can be used to determine add-on offers.
- Support the quality management in case of mass document production
Despite mass document production is a highly automated process, it requires human intervention at certain points. An example is the checking of the layouts before they get printed/produced. The system automatically generates samples from each “batch of documents” for verification. In practice, however, this means that for a service provider producing millions of documents, someone must go through thousands of pages. In such cases, a machine pre-filtering can be used to review the documents’ layout. As some of the errors can be easily identified by a machine: i.e. if a number sticks out of a cell, the logo is not in the upper right corner, the QR code is missing from the check, or it is simply out of place.
Thus, we can reduce the human resource requirement of the Quality Management process to a great extent by supporting this stage of document production with machine vision and deep learning algorithm based artificial intelligence.
Artificial intelligence is transforming jobs
The above examples demonstrate well that AI can release human labor which, at the end of the day, is a “win-win” situation. The employer is happy because he obtains a loyal resource that only needs to be trained once and then is available 24/7, working without mistakes for decades. This results in a more efficient operation, so the return on investment is also positive. On the other hand, the employee is also happy, because he can devote his time to tasks that require true human qualities. At least, this is what we learned from the study of ABBYY4, according to which nearly half of office workers would like to "delegate" the part of their tasks they dislike, to a robot.
Should you like to discuss this topic further, do not hesitate to contact our colleague responsible!
Gabriel Fellenberg: +49 176 6100 67 44