At the Rise of AI Summit 2020, Christian Guttmann, Chief AI Officer at TietoEVRY, draws on almost 20 years working with businesses across four continents to outline what boardrooms and grassroots employees each need for AI adoption to succeed. Structured around eight reflections, or 'ducks,' he argues boards need entrepreneurial, technically literate members, CEOs must personally learn the technology rather than delegate it, and organizations must build trust with employees by framing AI as augmenting tasks rather than replacing jobs. He challenges the assumption that AI always requires massive legacy data, citing the second version of AlphaGo, which learned purely from the rules of the game, and describes a 'second wave' of AI maturity in which purely technical data-science teams gain business, change-management and organizational mandate. Guttmann discusses the scarcity of top AI talent, estimated at roughly 10,000 people worldwide, the risks of choosing the wrong technology partner, and why so many proof-of-concept projects end up in what he calls the 'POC graveyard' instead of scaling into full production.
Key takeaways
- Christian Guttmann recommends boards include entrepreneurial or technically literate members, since dividend-focused, non-digital boards struggle to accept short-term trade-offs for AI investment.
- He cites Nokia chairman Risto Siilasmaa, who personally studied machine learning with Andrew Ng at Stanford before advising the Finnish government on AI education.
- AlphaGo's second version learned Go strategy purely from the rules of the game rather than historical human data, showing AI does not always require massive legacy datasets.
- Guttmann estimates the global pool of AI leaders able to 'move the needle' at roughly 10,000 people, with Europe a net exporter of that talent to the US and other parts of the world.
- He recommends measuring 'proof of value' with clear business KPIs rather than stopping at technical proof of concept, to avoid projects ending up in the 'POC graveyard.'
Detailed notes
Boardroom readiness: are your ducks aligned
Guttmann frames his talk around eight reflections on organizational readiness for AI, starting with the boardroom. He argues boards, who are ultimately responsible for setting company direction, need executive or non-executive members who are entrepreneurial or deeply understand AI technology, since boards with a mostly non-digital, dividend-focused background struggle to accept short-term revenue trade-offs for longer-term technology gains.
Connecting grassroots enthusiasm to the boardroom
He describes seeing many strong proof-of-concept AI projects in European and Nordic companies over the past five to six years stall and get buried in what he calls the 'POC cemetery.' Leaders, he says, must stay close to enthusiastic employees running these projects, study their impact, and build a path from prototype to full end-to-end deployment. He also stresses building employee trust in AI systems, citing examples from recruitment, where resume-scanning tools must be fair and transparent with a human still assessing outcomes, and sales, where AI recommendations on which customers to call must still let commission-paid sales staff trust they can hit their targets.
Advice for CEOs: learn the technology directly
Guttmann urges CEOs not to let outsiders determine their company's future, recommending they personally learn machine learning fundamentals. He cites Risto Siilasmaa, chairman of Nokia and CEO of F-Secure, who studied machine learning with Andrew Ng at Stanford, later advised the Finnish government, and contributed to making AI education mandatory in the Finnish parliament and to the globally distributed 'Elements of AI' course. His practical advice is to hire the best AI people available and spend regular time, including informal coffee and lunch conversations, with them to build a tangible feel for what the technology can do.
Data myths and organizational structure
He challenges the assumption that AI always needs large amounts of legacy data, pointing to the second version of AlphaGo, which learned its game strategy from the rules of Go rather than historical human data, and to cases where new data sources, such as newly installed cameras for factory anomaly detection, can be more useful than connecting all existing legacy systems. On organization, he describes a 'second wave' of AI maturity in which companies move beyond pure data-science teams to give AI teams business change capabilities and the mandate to implement end-to-end processes, while avoiding both full decentralization, which risks duplicated and competing AI initiatives, and full centralization, which loses insight from individual business units.
Talent scarcity and choosing partners
Guttmann cites estimates from the World Economic Forum, the New York Times and Wired putting the global pool of AI leaders who can 'move the needle' at about 10,000 people, translating to roughly five in Sweden, three in Finland and three in Norway, with Europe a net exporter of AI talent to the US and other parts of the world. He advises companies choosing an external AI partner to evaluate whether that partner's people combine deep technical understanding with business and regulatory insight in the same person, rather than treating technical expertise and business translation as separate roles.
Q&A highlights
Asked why boardrooms often lack entrepreneurial roles, Guttmann says companies with more entrepreneurial board and leadership teams are more likely to spot new AI-driven opportunities, while boards focused on executing a mature business tend to favor caution. Asked how leaders manage employee fear of new AI hires, he says introducing uncertainty naturally makes people protective of established, comfortable ways of working, and describes examples of companies deliberately exposing leadership teams to stressful experiences, such as helicopter skiing, to build comfort with uncertainty. On the 'POC graveyard,' he recommends measuring 'proof of value,' with clear KPIs on cost, revenue or operational impact, rather than stopping once a project is technologically feasible, and notes that scaling a working pilot across an entire organization often requires tens of millions of euros that only the boardroom can commit. On talent retention, he says top researchers at large US tech firms can become 'spoiled' because business outcomes are not always the top priority there, whereas European AI teams tend to be more focused on business impact. He also credits the Nordic 'fika' coffee culture and flat organizational hierarchy, including easy access to TietoEVRY's own CEO, as helpful models for fostering curiosity about AI across an organization.
Quotes
No one sells AI, right, you're selling the impact and the business and the value to the organization.
Christian Guttmann
Talent is still scarce, and to put it in numbers, WEF, New York Times and Wired have estimated the global talent pool of real leaders that can move the needle at about 10,000 experts in the world.
Christian Guttmann
Work your way backwards from the value that you want to create, from the value proposition, and see what type of data is necessary.
Christian Guttmann
Questions answered in this talk
Why do so many AI proof-of-concept projects fail to scale?
According to Christian Guttmann at the Rise of AI Summit 2020, projects often stop once they are shown to be technologically feasible instead of being measured as 'proof of value' against business KPIs, and scaling a working pilot across an entire organization typically needs an investment, tens of millions of euros, that only the boardroom can approve.
Does AI always require large amounts of historical data?
According to Christian Guttmann at the Rise of AI Summit 2020, no: the second version of AlphaGo learned its strategy purely from the rules of Go rather than historical game data, and companies can often start with new, targeted data sources, such as newly installed cameras, instead of connecting all their legacy systems.
How scarce is top AI leadership talent globally?
According to Christian Guttmann at the Rise of AI Summit 2020, estimates from the World Economic Forum, the New York Times and Wired put the global pool of AI leaders who can meaningfully move an organization forward at about 10,000 people, with Europe a net exporter of that talent to the US and China.