Whose AI is Smarter?

July 2017

Whose AI is Smarter?

Recently, I had the opportunity to participate in the “Clarifying Artificial Intelligence” panel at the Goldman Sachs TechNet Conference Asia Pacific 2017 in Hong Kong. We were asked to discuss AI technology, its applications and implications. During the panel and the follow-up meetings afterwards, there was a common theme to the questions asked by the audience. It was evident that AI and its related technologies were regarded as a holistic, all-in-one, mega-brain “thingy” which was expected to be “fed with questions or problems” and subsequently spit out automatic “answers and optimal solutions” with the push of a button. With this flawed logic, the questions asked were focused on comparing different AI “alternatives” or, bluntly “which company has a ‘smarter’ AI?”

Honestly, I wasn’t particularly surprised by these questions. Firstly, on the public and media front, we are hammered with news items and articles (some are advertisements in disguise) which focus on the “easy to understand” or attention-grabbing AI applications, such as AI machines beating human chess masters or vehicles driving flawlessly on the highway. Secondly, at its core, AI technology utilizes various advanced computer science and sophisticated mathematical algorithms, and isn’t easily understandable by the average person. So, there are numerous attempts to compose simplified explanations – which consequently, in most cases, do this technology harm. The bottom line – combining oversimplification with provocative examples has led to the belief that very soon we will all be able to purchase robots from US Robots & Mechanical Men (I, Robot – Asimov).

The somewhat shocking truth is that we are still only at the beginning of all it. It is correct that many companies are focused on establishing common platforms for hosting or accelerating AI-related algorithms, but, as we speak, many industries are still struggling with the basic way they do business and seeking immediate solutions for their existing problems. These industries lack the internal expertise for utilizing common AI technology platforms and tuning them to their own needs, and in many cases, as these platforms try to solve a generic problem, cannot match the scale and subtleties of the explicit issues they face.

At Voyager Labs, we take a pragmatic approach. We address some of society’s and businesses’ biggest challenges with products that employ leading edge, innovative AI technology at their core. Although the AI component is vital to the products’ operation and success, the focus is on the practicalities of the product offering. The attention on a specific vertical also resolves the need for a flat AI to AI comparison. It is not about which AI is smarter; it’s about the quality and impact of the end result. Whether it be a solution for optimizing a bank’s loan portfolio by assessing an individual’s financial risk — the success measurement is in the savings of lost funds (non-returned loans) or in the increase of approved viable applicants. If it’s the enhanced capability of ecommerce shops to understand consumer better — the success measurement will simply be the metric of raised conversion rates and increased sales.

In summary – hype aside, AI can definitely be used to solve various industry relevant problems and it is the solution benefits which should provide the appropriate measurement means for the technology’s “smartness”.

Share this post