The Center for Data Innovation spoke with Florian Goossens, chief operations officer at Radix, a company based in Belgium which helps companies, research institutes, and public organizations accelerate the implementation of AI-powered systems. Goossens discussed how AI can support many different types of organizations and the characteristics of successful AI projects.
Eline Chivot: Which trends or challenges did the three founders identify that led them to create Radix?
Florian Goossens: AI is a huge opportunity for organizations, but we noticed many find it hard to understand what AI can do for them or how to get started. Once an organization selects an AI use case, they are not sure which steps to take to make it a success. Generally, they do not have the expertise in house, so we at Radix fill that gap. AI is not just about coding and building models. The real challenge is to be able to build a use case for the business, a case that has been validated by both technical and non-technical stakeholders. At Radix, we spend time to deeply understand the business and the end-users. From the beginning of each AI project we do, we test it with the end-users. If organizations don’t have the end-users in mind when building an AI project, it will likely not be a success. If the user experience is not there, the project is not going to work. If the project is not aligned with the business strategy, it’s not going to work either.
We see two trends in AI: AI as an auto-pilot, and AI as a co-pilot. Auto-pilot means you completely automate a process, for example, automated data entry or automated production planning. Co-pilot means that the AI will not replace a human, but rather assist the human. For example, recommending which products to buy, or suggesting a response to a message you received. We think that the biggest part of the value that AI will create will be in co-pilot solutions, where interaction between humans and AI is crucial.
Chivot: Radix is active in a wide range of industries, from human resources to logistics, from media to pharmaceutical companies. How is AI useful in supporting various types of industries?
Goossens: As previously mentioned, it’s about automating current processes and tools, as well as allowing new opportunities, new business models. Specifically for the human resources market, for example, there is a huge opportunity to shape the future of the workforce. It’s about understanding what people want and can do, and helping them see opportunities in terms of learning, jobs, coaching, courses, etc. AI can help with assisting people, such as helping recruiters be more efficient and effective in scanning CVs, or allowing new applications, such as offering personalized career guidance.
Chivot: Radix has worked with VDAB, the Belgian public employment service, by building a job recommendation system to help job seekers. How does this project use AI?
Goossens: Our AI-powered talent data platform Talent API provides a job-matching solution that predicts how likely a job seeker will be interested in a vacancy, or how likely employers or recruiters will be interested in potential employees.
Classical, rule-based systems achieve this goal by manually defining matching criteria. For example: “To work as an engineer, you need an engineering degree.” This is a fragile system, which often doesn’t return the most relevant suggestions. Talent API has been trained on millions of vacancies in different languages. The AI system finds talent where people don’t, by detecting implicit and related or similar skills of a candidate. Using the state-of-the-art in natural language processing techniques, we are able to extract semantic meaning in data to provide high quality matching. For instance, for a candidate mentioning experience in data science, Talent API will detect implicit skills that are needed to do data science, such as a specific programming language.
Apart from matching job vacancies with candidates, we have worked with VDAB on other projects, such as an intelligent questionnaire for profession suggestions, or a framework for handling potential bias in recruiting and job matching.
Chivot: Radix is also active in transport services and has worked with a major European airport. Can you walk me through this project, and how it has improved the airport’s services as well as passengers’ experience?
Goossens: We helped a major airport innovate. We forecast waiting time at the security gates (screening) in order to decrease the number of people that miss their flights, and we also forecast how long passengers need to wait for their luggage to arrive at pick up.
The objective was to improve both the passenger experience and operational efficiency using data. By using modern machine learning methods, we are able to capture and predict temporal trends in the flow of passengers, and to derive waiting times from it.
The luggage retrieval time prediction is already in active use at the airport as part of our bTag App, successfully predicting 95 percent of the time by when a luggage will be available on the belt. The observed impact is an improved passenger experience in the form of less worrying about the luggage, and more free time spent visiting the shops in the area.
Chivot: Where do you see AI being the most promising in its various applications? And what could support or hinder its progress?
Goossens: We see the biggest promise is in assisting humans and enabling new applications. What is hindering progress is the lack of a deep understanding of use cases, or prioritizing the wrong use cases. If you start with the wrong use case, companies lose hope in AI and think it’s not useful for them. Understanding where AI can really help your business and where you should start will really catapult organizations. At Radix we believe in having AI champions, people who understand both tech and business. That is a real force for change. Where will progress take more time? Where being human is important, such as caring, having emotional intelligence, creativity, or making important decisions. AI is not capable of that yet.