In this 5-part series, two innovation heavyweights go head-to-head to discuss the current and future state of affairs. AI entrepreneur and author Jim Stolze and MultiMinds’ Head of Innovation Nicolas Lierman have an in-depth conversation on innovation and technology. Part 1: Do we have true AI yet?
Nicolas, as head of innovation, what is your relationship to AI and machine learning?
Nicolas Lierman: “AI is one research track of DIAL, our Data Innovation and Acceleration Lab. AI and machine learning (ML) within customer experience (CX) is the mid-term track. The long-term track deals with everything concerning Web 3.0, blockchain and the metaverse. I mention it because there’s an interesting observation to be made when comparing AI with Web 3.0.
The latter is clearly focused on the consumer, or the citizen if you will. If we compare that to AI, the focus is mostly on the marketeer. In our industry at least, I feel like the use cases for AI are focused on what it can do for companies, instead of consumers. It’s all about improving efficiency or saving costs for companies, not about increasing the quality of life for consumers.”
Jim Stolze: “That’s an astute, but somewhat painful observation. In my opinion, a real marketeer should be focused on the consumer exclusively. There are some examples of AI and ML applications that are aimed at consumers, but they often fall a bit short. Think of chatbots for instance, which are – let’s be honest – nothing more than pimped up FAQ’s.”
How should AI evolve to benefit the consumer then?
Jim Stolze: “AI is still quite immature today. And Nicolas is right, it is mostly focused on helping marketeers, for the simple reason that they are paying the people developing AI. With my own company, Aigency, we don’t work for consumers either, but for governments and companies. Because they come to us with concrete problems that we can help solve with AI. Hopefully, when AI matures, we can take what we learned in the B2B market and start applying it to solve problems in B2C.”
Nicolas Lierman: “A lot has to do with awareness as well. Professionals are more aware of what AI can do, and what type of problems it can solve. The public at large is still mostly unaware of the possibilities, so perhaps it makes sense that companies are driving innovation. Consumers are using AI daily, without realizing it. Just think of recommendation engines or a lot of smartphone applications. They are using AI passively. We need to improve the general knowledge of AI to bridge that gap.”
“There hasn’t been any time to make AI comprehensible for people other than researchers or engineers. As a result, many people find it ‘too complex’ to deal with, while others have unrealistic expectations about what it can do." - Mieke De Ketelaere
Do you agree with what Mieke De Ketelaere said in her Eureka! interview? Is this because of this knowledge gap?
Jim Stolze: “Absolutely. A lot of discussions about AI get sidetracked quickly. That’s because AI is still a specific field of expertise. It’s a continuous quest to make machines behave intelligently. That’s the long-term goal, but we’re not there yet. We are still applying smart algorithms to ‘dumb data’. The real question is: what is intelligence? We can make machines appear somewhat smart, but they are very far from what we consider ‘intelligent’. Today we are simply making predictions based on historical data. I don’t call this AI, but algorithmization.”
So how can we define ‘true AI’ then?
Nicolas Lierman: “We prefer speaking in terms of ‘AI capabilities’ to describe where we are today. We have smart algorithms, but no true AI. My definition of AI hinges on context. If the AI can function based on contextual information, we can consider it intelligent. But this is extremely hard. Facebook, for instance, can only detect 6% of all hate speech through AI. This is a purely contextual assignment. If a tech behemoth like Facebook doesn’t succeed in applying AI, you realize we are still in the very early stages of the technology.”
Jim Stolze: “The definition of AI is constantly changing, too. In the 80s, we considered playing chess as a difficult task. Today, AI has no problem whatsoever beating the world’s best chess players. The best definition of AI for me is the one Professor Tom Heskes put forth: ‘AI is whatever information technology can not do yet’. The things that seem impossible today, might be easy in 5 years time.”
Nicolas Lierman: “The hardest task for AI is to make it funny. Humor is viewed as a manifestation of intelligence. The applications we see today are basically just extrapolations of existing data. An AI can create a new composition in the style of Mozart. Deepfakes can make anyone say and do anything. But no AI can be truly original or funny. It needs historical data to pretend to be intelligent. I guess we can conclude that we haven’t seen true AI yet, in whichever definition you apply.”