Will AI and in-place management solve my records problem?

By Nigel Carruthers-Taylor

The AI revolution is upon us, and at the forefront stands Microsoft Copilot. Many are buzzing with excitement as Copilot steps onto the scene. It promises to transform document analysis, report generation and streamline processes. It’s also getting records managers excited: will it automate their records management?

Without a doubt, Copilot will supercharge productivity. Imagine a knowledge worker at a legal firm on a mission to undertake content-based evidence analysis. They get swamped with mountains of documents, videos and images. Copilot steps in and analyses these like a cyber detective.

It categorises the evidence, so it’s neatly filed: “Conspiracy Cache” next to “Cat Video Crypt.” It identifies key evidence: images spotted! Incriminating phrases flagged! Then it summarises and reports: legal briefs assembled, complete with footnotes and references. All done fast and efficiently, saving my legal eagle heaps of time, and no stone has been left unturned!

But wait! Recent research by Cornell University has spilt the beans on AIs: they can daydream! [1] They can hallucinate like poets on a caffeine binge – between 69% and 88% of the time! Even for simple tasks, it can make stuff up 50% of the time. This is such a risk that US regulators have determined that AI has the potential to perpetuate unlawful bias, unlawful discrimination, and produce other harmful outcomes.

The culprit? Inadequate training data – not enough of the right content, and not in the right context. Imagine an AI in a dark room, squinting at half-written scrolls. “Where’s the context?” it cries. “What’s the backstory?” And so, it invents tales of Shakespeare and tax forms.

This contextual conundrum is important! Not only can bad context make AIs hallucinate, but it can skew your understanding of a business transaction’s context, or make you miss something altogether.

Context is especially important now that everyone wants to use in-place records management. “If we can use AI to auto-classify and store records, in-place management seems like a no-brainer”, I hear you say. Well, is it?

Academic research warns against blind use of in-place management[2], and standards bodies recognise that in-place systems do not provide enough context to give full and accurate records[3]. Meanwhile, technologists tell us that manage-in-place results in higher storage, audit, security and privacy costs[4].

This is why experts are warning us to be careful about in-place management’s use[5]. Imagine you're in the midst of contract negotiations. You communicate via email and draft the contract using templates from a network drive. You diligently save the draft to Teams and share it with colleagues for collaborative input. Each colleague works independently to complete the shared contract. Of course, your AI and in-place management takes care to identify and place records controls on the contract.

After the contract is signed, you discover that the price was incorrect! Now, where did this erroneous price originate? Was it from the templates (and where are they stored?), your colleagues (and where are their records?), or did the price change during negotiations (with records hidden in emails)? In this scenario, in-place management has dutifully auto-classified and applied records controls - but the critical business context remains scattered across various locations. It resides in people's Outlook emails, your Teams site, personal repositories, and other business area storage.

That's why we've created Ingress by iCognition. Ingress ensures the safety of your important records. It combines various models to deliver complete records in the right context. With Ingress, you can automatically or intentionally manage records in the context of the business transaction, either in-place or in a secure corporate store using OpenText Content Manager. That way your vital records are full, accurate and accessible, saving you time and money, as well as reducing the risk of reputational damage.

And the best part? You can do it all without leaving your favourite business app. Ingress is a Content Services Platform that delivers enterprise-wide automated processes, customer-centric services and agile, and seamless information flows within the user’s favourite business application. Use it to find, manage and protect content quickly and easily across multiple repositories: Teams, SharePoint, Exchange, OneDrive, OpenText and others. All while operating inside those applications!

Ingress is not just a compliance platform, it’s a productivity platform. Let’s get back to our legal eagle reviewing his AI-generated report. Something seems off, and he needs to inject some precedent text pronto. But where's that text hiding? Is it buried in his network drive, Outlook, or somewhere else entirely?

Luckily, Ingress comes to the rescue. With a single click in Microsoft Word, he swiftly conducts a search, locates the relevant information within the appropriate context, and seamlessly inserts the text into his report. All done quickly and easily, and without leaving Microsoft Word. Talk about a lifesaver, right?

Furthermore, Ingress incorporates our proven innovative RM Workspace and RM Workflow to enhance business efficiency and content productivity while ensuring transparency in record-keeping. Our legal eagle protects his document in the right context using behind-the-scenes corporate records structures.

But he doesn’t care about these structures! He wants to make sure he can quickly access his document when needed, share it with some colleagues, convert it to a PDF and initiate an approval workflow. He does this by tagging the document with his personalised terms, placing it in an RM Workspace portlet within Teams, and triggering a formal RM Workflow approval process with his superiors.

So, will AI and in-place management be the ultimate solution to your records chaos? While AI definitely has its perks for analysing documents and generating reports, it's not infallible, especially when it comes to understanding context. Similarly, in-place management is handy for short-term or low-value records, but it can leave you sinking in the quicksand if you're trying to piece together information from multiple sources.

The key takeaway? While AI can significantly enhance productivity and provide valuable advice, it's essential not to treat it as the sole authority in records management. Instead, consider it a powerful tool within a broader strategy. When implementing solutions like Ingress, ensure they support various records management approaches, just like a prominent client who recently opted to combine in-place management and a corporate store to address their records management challenges.

With Ingress and its flexible records management, especially when complementing it with AI, you’ll not only steer clear of the quicksand but also gain context, accuracy, and long-term access to your records. Using Ingress you will deliver the right information to the right person at the right time, with minimal effort. With Ingress, say goodbye to record-keeping headaches and hello to accurate content management to fuel your digital transformation!

Nigel Carruthers-Taylor is Executive Director & Principal at Information Management and Governance Specialists, iCognition. To view a detailed video demonstration of Ingress by iCognition Register HERE. For more information contact iCognition on info@icognition.com.au.

[1] Dahl, et. al.  Large Legal Fictions:  Profiling Hallucinations in Large Language Models.  arXiv:2401.01301v1 [cs.CL], 2 January 2024. This is nicely summarised in the article Wrong at the Speed of AI: Hallucinations and Regulators, Why Both Provide Insight into Organizational Risk and AI (archive360.com)

[2] Lappin, et. al. Rival records management models in an era of partial automation, January 2021

[3]Australasian Digital Recordkeeping Initiative (ADRI), Functional Requirements for Managing Records in Microsoft 365, October 2021

[4] Bill Tolson, Why In Place Information Management Is Not A Complete Solution, April 2021

[5] Bardara Reed, Reflection on ‘Recordkeeping theory, models & strategies and today’s workplace’, March 2021