We have entered a strange and powerful era of artificial intelligence (AI). Today’s systems can write poetry, create music, and even help develop new drugs. Yet a Google engineer can claim a conversational system is sentient, only to be met with the scepticism of experts who know the truth: these systems don’t understand what they are doing.
This is the central paradox of modern AI. The recent explosion of progress is based on engineering—the “how” of making things work. We use ever-bigger neural network models, train them on all the text on the internet, and apply unprecedented computing power. The results are impressive artificial “savants” that make brilliant predictions.
Science, however, is concerned with “why” things work—understanding the underlying mechanisms. This “why-knowledge,” as scientist Judea Pearl would call it, is largely absent in mainstream AI today. So, can this powerful but un-understanding AI truly be useful to us?
The Human Bridge to Business Value
The answer is a definitive yes. This “how-AI” is a great tool to extend our capabilities to explore and be creative. However, it has not given us an AI that autonomously solves problems and explains its way to the solution. It has not given us artificial general intelligence (AGI).
The pressing problems that businesses face lie in the “understanding” part of the intelligence spectrum. So how can companies profit from AI in the meantime?
Given that humans are still the only guardians of why-knowledge, there is only one way: relying on clever data scientists coupled with AI power.
Data scientists are often misinterpreted as technicians who run algorithms on datasets. On the contrary, the data scientists who deliver true value are those who reason deeply about a problem; they are scientists before they are data scientists. They understand business needs, creatively link them to problems that AI can address, and then solve them—often with a great deal of discussion and thought. They then convince management that the solution can work and help software engineers industrialise it.
Building the Team That Can Harness AI
Much more than relying on algorithms, relying on people is the core challenge companies must address. I have been fortunate to work at a first-class AI research institute for the last 25 years—the Dalle Molle Institute for Artificial Intelligence, founded in Lugano in 1988—and have recently gained entrepreneurial perspective by co-founding an AI company.
Based on this experience, here are my principles for building a team capable of true AI-based innovation:
- Selection is central: The selection of excellent and experienced data scientists is key. Technical qualities are as important as personal ones and motivation.
- AI requires a team: One data scientist alone won’t do much—AI is a very big field.
- Engage subject-matter experts: A strong data-science team won’t help much if its business counterpart is not truly engaged in the process.
- Create the right environment: Data scientists value informal, academic-like environments where they can openly discuss, learn, and vary the problems they work on.
- Stay connected to research: Staying up-to-date with recent advances in the scientific literature is fundamental. Having tight connections with top-notch AI researchers in academia is a great plus.
The People-First AI Strategy
It may seem unrealistic for traditional companies to set up an internal data science team in such a demanding way. I envisage the need for private and public “AI-lab on demand” services so that companies can flexibly tap into this expertise.
But the core message remains. The main obstacle to boosting competitiveness with AI is never technological; it is people. We need clever data scientists as well as open-minded management and experts working together to promote innovation. It is, and always will be, a team play.
