How Low-Code Development Will Bring Data Science to the Masses
Data science has traditionally been the playground for the highly trained and highly skilled. But low-code and no-code programming has that potential to put it in the hands of the average knowledge working, unleashing vast new potential for the enterprise.
Data science is becoming a must-have for the new digital economy. As organizations become more adept at capturing, storing and analyzing data from numerous sources, they are finding previously untapped founts of knowledge that lead to new sales opportunities, new markets and even entirely new business models.
The challenge, however, is that only the most well-heeled enterprise has the means to acquire the highly specialized skillsets needed to turn data into actionable information. Demand is high for what is usually a PH.D-level employee, and viable candidates are scarce.
Low Code to the Rescue
Enter low-code development. As Alluxio CEO Steven Mih points out, low-code, and even no-code, tools are hitting the channel that should allow even non-technical knowledge works to query databases, mash up disparate data sets and derive insights from random information — all without having to learn the intricacies of coding, DB management or any other data science-related skills.
In this world, business teams become the experts in leveraging data for their own needs rather than rely upon a mystical cabal of scientists who have limited knowledge and intuition when it comes to business matters.
As early as 2020, Mih says “these no-code or low-code technologies will bring machine learning to the forefront and make services smarter so the business isn’t reliant on individuals with specific expertise. Instead of building and deploying models, for instance, we will see ‘bring your own model and we’ll run the training on it for you’ autonomous technology.”
We can already see these kinds of tools in action. Google's Cloud AutoML and Teachable Machine 2.0 services offer quick and easy access to advanced artificial intelligence (AI) and machine learning capabilities, while firms like C3, Mendix and Appian are offering low-code options for their clients.
Rajeev Dutt, CEO of DimensionalMechanics, recently told SiliconANGLE that he’s seen high school students with less than a week of training on his NeoPulse Modelling Language build fully functional AI models with as little as 14 lines of code.
This kind of capability, in fact, is expected to finally deliver on the promise of big data following years of disappointing results. With AI and other technologies delivering actionable results in real time directly to people with little or no technical know-how, businesses can finally become responsive to the demands of an increasingly digital economy.
But even as low/no-code is making it easier to leverage AI, the reverse is happening as well. Recently, a low-code platform developer called OutSystems added AI and ML features to its system, allowing non-developers and novices to incorporate advanced automation, voice recognition, self-service and a range of other capabilities to their products. At the same time, it simplifies the process of integrating other tools, such as Siri, Skype and Facebook Messenger.
António Alegria, OutSystems’ head of AI, said these tools now give enterprises of all sizes the ability to quickly incorporate AI and other forms of intelligence into their mission-critical applications without having the pay upfront for high-level data science skills.
At the same time, OutSystems is hoping to benefit from its partner development relationships by bringing new tools and capabilities into the platform that can address the needs of key industry verticals like finance, retail and healthcare.
Coding for the People
All of this activity falls under the broader goal of democratizing data science throughout the knowledge workforce, according to business services firm Deloitte. With the U.S. alone is looking at a shortfall of some 250,000 data scientists by 2024, low/no-code, along with other developments like automated machine learning, AI pre-training, self-service analytics and accelerated learning, aims to break the bottleneck of data science skills that is currently holding many companies back.
To be sure, there will always be a need for highly specialized data scientists, but with more usable tools in the hands of more people, those highly paid employees can be tasked with projects more suited to their talents.
The key to successfully implementing low/no-code tools into enterprise workflows is to make them as easy as possible to use. Simplified GUIs with drag and drop modules, along with an overall user-friendly design and architecture, will go a long way toward accelerating adoption.
To do this, of course, platform designers will have to stop thinking like, well, platform designers and get more into the heads of salespeople, doctors, bankers and others who will actually use these systems. Deloitte claims that application and service development times can be shortened 10-fold with the right low/no-code design.
Organizations looking to leverage low/no-code should be aware, however, that it can disrupt business models as easily as it enhances them. Once people have the ability to create their own apps and services, demand for commercial offerings will likely trail off.
While it may be tempting to restrict the ability to alter a given ecosystem — a car, for example, or a phone — organizations would do better to encourage this trend, under a proper governance regime, of course. In this way, users gain the autonomy they desire while the enterprise gains the benefits of an expanding and innovative development community.
As with most advancements in the digital age, the spoils won’t go to the person who captures and leverages a particular technology, but who encourages others to play in their sandbox.