The Acceleration Economy Analyst Network recently got a unique opportunity to connect with one of the pioneers in the AI field — Ingo Mierswa, founder and Chief Technology Officer of RapidMiner — to get his point of view on AI uptake and new directions in his company’s strategy to maximize usage of the technology.
RapidMiner has publicly outlined a corporate and product direction that aims to make AI broadly accessible. For Mierswa, this goes back to his earliest days and thinking (RapidMiner was born from a data science project at the University of Dortmund in 2001) around AI, namely taking a practical approach to applying the technology. “There was always something that was important to me personally: Not just doing science for the purpose of doing science, but actually solving real-world problems,” he recalls.
There are several ways that RapidMiner’s latest directions put that thinking into practice, including:
- Making sure all types of stakeholders can use, or at least take advantage of, AI technology through approaches that eliminate coding and make the tech available to users at all skill levels.
- Providing training and upskilling opportunities for those who want to gain AI and data skills that boost their careers and their company.
- Overcoming AI skepticism or reluctance by building trust in the tech.
- Capitalizing on the most broadly used IT architecture — the cloud — to eliminate barriers to entry.
Additional details on each of these initiatives, and inputs from Mierswa, follow:
How RapidMiner Maximizes AI Accessibility
One of the underlying goals of computer science is to solve a problem and do so in a way that you can quickly solve similar, future problems — ideally without creating new code. Mierswa, and the RapidMiner approach, similarly emphasize no coding where possible in order to accelerate projects and delivery.
“We need to find a way to avoid coding so that instead of taking six months, we could take six hours and can enjoy the rest of the day sitting on some beach drinking long drinks,” Mierswa quips.
On a more serious note, he says that being overly reliant on code creates unique challenges for presenting and explaining results or recommendations from an AI tool, versus coming up with more easily understood and visually appealing approaches.
The latest RapidMiner release targets this objective specifically. If a company has a business expert who is data-savvy but doesn’t code, that expert can tell RapidMiner what they would like to predict. Then, the system will create a model, generate insights, and predict outcomes based on those models.
The company calls this “fully automated AI” and is striving for “a really good balance between the power and flexibility from coding” with greater “understandability, and let’s say ease of use, that you have from the visual approach.”
This mindset developed along with the maturation of the company. “The thing we learned is that you can have the best product in the world, but that doesn’t alone create the necessary changes when it comes to AI and ML,” Mierswa says. “Because keep in mind, if you want to get the biggest value, you will need to change the business process, you need to change the way you make decisions.”
Providing Training and Upskilling Opportunities
The skillsets of employees — from data scientists to business power users to novices — relate closely to the first point above: namely, how to maximize value and leverage from AI. The more collectively skilled a company’s employees are, the more value there is to be realized.
In parallel to offering product features (like fully automated AI) that make it more accessible, RapidMiner acknowledges the need to offer training and other opportunities for companies to upskill their employees who will maximize adoption of and leverage from AI models.
In particular, expanding the use of AI to non-technical users could have a major payoff. “They don’t need to become coders, but they need to understand the concepts behind ML,” Mierswa says. “At the same time, we also need to help other people with, for example, those communication skills, like what are we doing with those results, how we share those results.”
To that end, the company is offering training for all levels of users, helping them build skills and confidence as they adopt best practices in AI. It’s also building a curriculum that includes repeatable use cases.
Overcoming Skepticism and Building Trust
One major factor that prevents the realization of full value from AI today is that many organizations lack data science familiarity and comfort level — think of it as not trusting AI or the people supporting the technology.
As a result, not enough of the AI models that get built make it into production, so companies don’t get (or perceive) adequate value from their investment in the technology.
One of RapidMiner’s latest objectives relates directly to this non-technical consideration: how to build and gain trust for AI initiatives and recommendations coming from systems. As we all know, transparency and accurate results (or forecasts) are two sure ways to build credibility and confidence in the technology.
To deliver on these objectives, the company is increasingly focused on offering reusable building blocks and tools, like whiteboards, to make AI concepts understandable to non-technical business people.
From Mierswa’s perspective, the company aims to show visuals and workflows as opposed to “35,000” lines of code that will never be understood. “Most people are good at that level, they know what happens, they understand how the data was prepared, they can build the necessary amount of trust.”
Capitalizing on the Cloud
At the time of RapidMiner’s June product update, the company explained the benefits of offering a fully cloud-based version: ensuring maximum accessibility and alignment with the ubiquitous IT architecture.
The new platform includes the full functionality of the company’s existing software — in use among more than one million analytics users worldwide — in a multi-tenant cloud platform that also delivers an enhanced user experience to simplify problem-solving and analysis.
The new release fully leverages the cloud’s elastic nature and increases scalability dramatically to meet the most demanding enterprise requirements.
The cloud version maintains the three modalities for using RapidMiner (fully automated, notebooks, and workflow designer). The big win is that there’s no provisioning of computing or other infrastructure to support the work. “The cloud platform obviously makes that even 10,000 times easier,” Mierswa says, “because you don’t need to install anything, you need to just get started.”
For more exclusive coverage of innovative cloud companies, check out Cloud Wars Horizon here:
