To casual observers, technological innovation is a fast-paced environment where every other day seems to bring something entirely new to the market. Which is true, to an extent. For decades we’ve registered growing amounts of capital poured into young ventures whose only objective is to revolutionize the way we do things, processing power has grown exponentially as have the ubiquity of computers and Internet connectivity. All of this inevitably leads to thousands of startups that fight for the spotlight through improved ways to solve ordinary problems.
On the other hand, for those working in this field the pace of things is much slower. A tech professional can easily spot a product that’s not truly pushing the envelope, or when a marketing message is the only original concept behind a brand. The large majority of everyday launches don’t add anything revolutionary to the mix, but they do offer something tangible: an increasingly clearer view of the underlying subject most technology providers chase. Companies may look for unique methodologies to address a problem, but typically they’re all targeting a common, bigger quest. Later on, especially at the end of a decade, it gets easier for all of us to take two steps back, look at the bigger picture, and see which trend really dominated the hearts and minds of practitioners in a field.
It's All About Automation
While the discourse around Automation today seems to be losing steam in favor of other subjects (the press is in constant need of new oxygen…), the work that happens behind the curtain in the real world is still very much revolving around the advancement of this practice, which is far from having reached its maturity but at the same time is already producing an impressive output. My opinion is that RPA (Robotic Process Automation) will come out of this decade as the indisputable winner, and every other small or big piece of technology that’s now joining the scene will inevitably be attached to Automation platforms, these functioning as the necessary stage for all the actors to play their role. In fact, I see the current condition as an extension of the imperative around platforms that most solution providers are embracing (a deep dive on the subject here in this interesting article from a16z).
To better explain where I’m coming from, let’s take a quick look at the history of innovation and the tech waves that drove change in the past, with a focus on Applied Computational Linguistics since this is the area I operated in for the last 30 years or so.
A Revolution 30 Years in the Making
The 90s were all about personal productivity. Microsoft Word, spell checkers, thesauri, digital editions of dictionaries, and so on. Personal computing was growing fast. “A computer in every home” was transitioning to a computer in every pocket. Handheld devices were already having quite the success when mobile phones and on-the-go communication boomed, kickstarting the first iterations of what we’ll later call smartphones. The company I was working for at the time offered incredibly cutting-edge products that went a step beyond a typical spell checker, analyzing sentence structure and fixing typos as well as grammatical mistakes.
The aughts focused on team productivity. Not only did computers’ presence in every company grew greatly (thanks to software dedicated to every function in the organization), but also intra-company networks became commonplace, as well as Internet connectivity which allowed to easily make computers in offices around the world talk to each other. This sparked the first applications and operating systems that natively managed workgroups, the active collaboration on shared documents, instant messaging. The boost in productivity this phenomenon generated was so immense that in many ways it dragged into the following decade, while mobile and internet networks matured into clouds, and every single process moved, slowly but surely, online. The leap forward I observed in Computational Linguistics during the 00s was primarily in Knowledge Management. From the first meaningful instances of Knowledge Graphs to corporate databases, from Data Analytics to customer Sentiment Analysis, the impressive computing power suddenly available combined with an almost infinite storage space to bring the ability to collect and manipulate large amounts of data, and, more interestingly, have people and applications everywhere contribute to this knowledge pool. The adoption of digital taxonomies and ontologies became commonplace and generated a wealthy market of its own, basically overnight.
This particular flavor of innovation kept on going strong in the following decade, bouncing off new elements like NoSql databases, triple stores, etc. And I might add Taxonomy Management still has a lot to say in NLP, though now more in a vertical, industry-specific way. Which, in turn, was the underlying thread of the 2010s, when NLP abandoned a general purpose, almost vanilla, application once primarily powering the Media&Publishing industry, to make a push in other sectors. From Finance to Insurance, from Intelligence to social media Opinion Mining, the past decade reflected a renewed cross-industry awareness that unstructured data is a bottomless well that can be mined to generate value.
And this leads us to today. The shift we're observing in recent years is related to workflow productivity. We’re quickly abandoning the small-minded concept of making single tasks easier to the advantage of a more wholesome approach that betters entire processes start to end, top to bottom. This revolution relies on a combination of AI and robotic automation, and it's having the highest impact because it transforms the way we work. Workflow Automation is about freeing up time that Subject-Matter Experts can spend on activities that actually require their expertise, as opposed to reading documents, searching the Internet, moving files around. It unleashes the potential of cognitive technologies to a point where entire operations can be automated. It’s when RPA embraces technologies like Natural Language Understanding (NLU) to become what’s now known as Intelligent Process Automation.
To offer a clear example, I'm thinking of solutions around, for instance, Claims Management in the Insurance industry, where a sufficiently-advanced NLU software can read an entire claim package (including medical reports, police reports, and so on) in order to provide the Claims Handler with the information they need, instead of having them read hundreds of pages before they can start doing their job. The same can be said for financial reports, medical trials, all the documentation healthcare providers work with in hospitals and clinics. Intelligent Automation frees up the experts' time, speeds up processes while keeping them consistent, and offers an overall better work-life balance for all the individuals involved. Put differently, all of this can easily impact multiple layers of an organization; this is a story about technology, therefore a story about productivity, and productivity is something every department ultimately contributes to.
This level of innovation comes at a cost, as one might expect: the challenge is around the way this changes the business of technology vendors, the people they talk to in an organization that wants to adopt these solutions, the Change Management on the customer's side, and, last but not least, the necessity for these companies to embrace transformation at an enterprise level.