Automation & Artificial Intelligence – The future of work for Document Control

Artificial intelligence (AI) has been everywhere in the news since the recent release of ChatGPT on the 30^th of November 2022. We have seen countless examples of people testing the tool on a wide variety of topics, and with answers from the AI that were quite astonishing sometimes.

• It’s been used to help IT professionals with their coding, generating within seconds lines of codes that would have taken hours to write for an experienced programmer.
• We have seen students using ChatGPT to help with their college papers, or even during exams (that’s not allowed by the way 😉).
• We have seen the AI helping medical doctors to review scan results and identify anomalies, or even helping users to self-diagnose illness based on symptoms.

Those are just a couple of examples from millions of examples that have arisen in the past few months. After all, just after 5 days of being completed, ChatGPT reached a million subscribers. And after 2 months, it had 100 million users with 13 million visiting on a daily basis.

So, we are talking about a rolling wave that almost nothing could stop, and we now can’t ignore that generative AI will be a significant part of our daily lives in the very near future.

Which takes us to the question: what are the possible consequences for us in the workplace? And how will it change the way we work, and the way Document Control works in the near future?

We have all heard the news headlines saying that artificial intelligence will have a significant impact on the job security of workers throughout the world, with a McKinsey study saying that it could displace between 15% and 30% of workers.

Moving away from sensationalist headlines, this article aims at diving deeper into what it means for the future of Document Control professionals and of the Document Control practices in organisations.

Let’s start with terminology for a better understanding of what we will be discussing here.

• Artificial Intelligence (AI) If AI was to be summarised, one would say it is a science that aims at creating machines that are able to have intellectual capabilities that resemble those of humans. Machines capable of analysing, evaluating, reasoning, planning, communicating and even learning. 

• Automation - Even though it is often confused with AI, it is not exactly the same, as automation involves programming in order to carry out specific tasks in a recognisable pattern. For example, automation is used to help us with repetitive tasks such as sending an email to someone every time something specific happens, or to copy/paste something from one place to another place automatically, without human intervention.

• Machine Learning is a type of AI that can learn: rather than being programmed, the AI will analyse the data it receives and will be capable of making recommendations. One very common example of that (and that we all use everyday already) is with the ‘predictive text’ function used on our smartphones, email software or office software : the program will guess a writer's next word based on their writing style. Of course, this is just a simple example, that will be taken to an entirely new level with so many other applications in the future.

• Generative AI is a type of AI that can generate various types of data, such as images, videos, audio, text, and 3D models. An example of that would be automatically generating a PowerPoint presentation with text and images, pulling information from existing documents on the same topic.

• Large Language Models (LLM) are a type of AI with a deep learning ability that can understand, summarise, predict, generate data based on knowledge pulled from a significant variety of data and datasets. It also allows humans to “discuss” with the AI, using their own words. ChatGPT is a good example of LLM, but it is not the only one, of course.

Generative AIs and LLMs are only at the beginning of their capabilities, so, although we may understand what it can do for us for the next 2 to 5 years, it is almost impossible to predict exactly what that technology will be capable of in the next 10, 20 years or 30 years. Indeed, the power and the intelligence of AI is increasing at an exponential rate, and that makes it, by essence, unpredictable.

This makes writing this article challenging in itself, because it will need to evolve in the future, along with the evolution of the technology. 

A large portion of our daily lives in office jobs is spent using Microsoft Office tools like Word for generating documents, Excel to process and analyse data, PowerPoint for producing presentations, etc.

Therefore, this will be used as an example of what will change in the coming months and years in offices around the world.

But of course, note that similar changes are going to happen throughout all the applications that you are currently using in the workplace.

In March of 2023, Microsoft released a series of communications and videos about their own application of LLMs and generative AI, which they chose to call ‘Copilot’.

The idea behind Copilot, as explained by Satya Nadella (Chairman and CEO of Microsoft), is that we have now entered a new era of AI: indeed, until recently, AI was part of our daily lives, with predictive search engines, social media feeds that guess what we will be interested in seeing next, targeted advertising, recommendations in Netflix, etc. Microsoft calls that past era of AI – which  has become second nature to us - the era of using AI on “Autopilot”.

They propose that the new generation of AI that is in front of us will be different, as we will be moving from using AI “from autopilot to copilot”. 

Microsoft will be gradually launching their new Copilot software in the next coming months, which will be embedded into the existing Office tools of Microsoft (such as Word, Excel, PowerPoint, etc).

They announce that it will “radically change the way we work”, as Copilot will be used as a personal assistant with tremendous analysis capabilities, for example:

• Generating a document such as a proposal, by pulling information from existing data from minutes of meetings with a Client, and from the company brochure, for example: the AI creates a draft (let’s call this a starting point) of that document with the content it deemed important and proposes a structure for the proposal. The author then needs to thoroughly check, verify, add, modify the document to make it their own.

• Creating a presentation from existing documents: for example turning a Word Proposal into a PowerPoint presentation

• Analysing data from an Excel spreadsheet

• Or creating a new sheet with a dashboard style to analyse data

• Managing emails with inbox triage of the most important emails, summarizing of long email threads, drafting an email reply including data from various sources, or helping with the style of your email draft (make it more concise or longer, more professional or more casual, etc

• Summarizing meeting transcripts, analysing points discussed in a meeting or action points agreed on

• Summarizing long chat discussions

What information can we trust?

Over the past decade, we have witnessed a growing trend towards collaboration over control when it comes to documentation in the workplace, and it looks like the trend is going to continue growing.

It comes with its challenges and risks regarding traceability of decisions, record of the basis for decisions, designs and endeavours, and with a growing unease about what information can be trusted, and how do we differentiate it from what is only chatter and drafts.

Being able to automatically produce a document within seconds that pulls information from a lot of sources can seem rather exhilarating. And although it is quite exciting to think about all the manhours that we will save in the workplace, one can only wonder about the basis on which the AI will decide that an piece of information is “important” or not, or whether something will be mentioned in the document or not.

The AI is going to initially be as smart as the data sets that we will provide to it, and unfortunately we all can see that most data sets in the workplace cannot be fully trusted: have you ever wondered which was the trustworthy document after you found two versions of a document in a folder? Have you ever witnessed a document having been incorrectly classified in an inadequate location, or a document with inaccurate metadata, the result of which being incorrect reports and search results? 

So, if the information we feed to the AI is not necessarily trustworthy, the results of its analysis will be even less. The issue being that we won’t necessarily have the time to cross check where the information actually comes from, as the AI will have pulled it from a variety of sources.

So, we can see that, in the future, for the AI to work properly and to provide the expected results, we are going to have to think even more about how to guarantee the integrity and the quality of the information, how to push for consistency and compliance with the rules required for company protection, and how to ensure the traceability of where the information comes from, and where it goes to. Document Control by essence can be a key player in helping to manage all these challenges and risks.

Better and more reactive support to users

When a user request comes into the Document Controller’s inbox, it is sometimes challenging to answer in a timely manner as they are usually busy, checking, managing and processing hundreds of documents.

But with the growing presence of AI assisting us in replying to emails, drafting replies, and providing answers to frequently asked questions, we can see that the response rate as well as the response time is going to improve drastically.

As long as the DC double checks the email drafted by the AI before sending it out.

Reduction of manual repetitive tasks

In the recent years, technology has already started to help a lot when it comes to data entry, automatic data import from one system to another, automatic feed of databases based on information coming from a variety of sources. 

The trend will accelerate with AI, probably supporting even more extracts of data from documents to automatically input information into EDMSs (Electronic Document Management Systems), identifying discrepancies between the expected and the actual, etc.

With automations and AI, we will be able to support DCs even more with the repetitive tasks that they have to perform, which should free more time for DCs  to perform more complex operations. 

This means that current Document Control professionals are going to have to up their game with new in-depth skills (including via attending professional development training courses in the field), especially if currently a lot of their days is spent doing data entry, or uploading/downloading documents in a system, or from one system to another.

These tasks are likely to be very much automated in the future, therefore planning for the development of more value-adding skills is key for the future of our profession.

Fewer compliance and quality errors from document authors

A lot of our day in Document Control is spent checking documents for their compliance and quality, and to liaise with authors and originators, asking them to modify documents according to the list of discrepancies we identified. 

Having an AI assisting with document production and document writing will reduce the turnaround for creation and modification of documents, and the fact that it is likely to be capable of automatically formatting a document based on an existing template is going to save a lot of manhours both for the author and the Document Controller.

The above is true if DC does not forget how important it is to check the document before publishing, as even the most intelligent AI cannot be fully trusted on compliance, quality, consistency, integrity, traceability of the documentation. 

Document Control will still need to be the gatekeepers that they currently are.

Data Analysis

Document Control is at the centre of a massive amount of data: document registers, planned and actual delivery dates, information about progress, reviews, approvals, distributions, interfaces, etc.

One of the tasks of a Document Control professional is to provide visibility to their teams (e.g. managers, engineers, authors, interfaces, clients, contractors, and so on) about the progress of the documentation, the possible bottlenecks, late deliveries, and other key facts, stats and trends. 

The ability of AI to analyse quickly a large amount of information and its ability to summarise and to identify trends and unexpected changes in patterns, will most probably be of great help to Document Control. 

It may allow us to issue better reports and dashboards, which pull information from many sources; it may allow us to predict future performance on documentation progress and deadlines, to identify potential issues and threats, and to summarise all of it in a digestible format, useful to our busy users.

Support to compliance

We have seen it in the past 25 years, with the growing presence of modern EDMSs in the workplace: when an EDMS is well designed, it can support a better compliance overall, as people have to follow the established rules in order to publish their documents, have it reviewed or modify it. 

AI is going to accentuate this support and is going to help prevent non-authorised actions, which may have repercussions for the company.

Improved office skills

It is said that the average employee uses only 10% of the capability of software packages such as PowerPoint, Word, Excel. 

With the support of an AI with which we can discuss and ask for support, we can increase our usage of these office tools, and use much more of their capabilities. This can result in more professional-looking documents, presentations, registers and dashboards.

As a conclusion, we can say that we are entering a new era in the workplace, which is both thrilling on some aspects such as:

• efficiency
• rapidity
• reduction of repetitive tasks

and concerning in others:

• integrity of information
• traceability of decisions
• deterioration of the existing skillset of the workforce
• lack of transparency on its decisions

and that, as always with a new era of technology, we will have to observe closely, being neither naïve nor excessively conservative or cautious, looking at the positive changes it brings, and the regulations and rules that will have to accompany it along the way. 

What do you think of the future of work, with AI and automations? Does it excite or frightens you?  Let us know in the comments!