At Rise of AI 2019 in Berlin, Prof. Dr. Christian Guttmann, Executive Director at the Nordic AI Institute, argues that AI is entering a phase where it will reshape how people work, receive information and interact with machines, drawing on 25 years in the field spanning IBM, big-company AI projects and academia. He predicts children growing up today may never learn to drive, will live in hyper-personalized information environments, and will see precision medicine transform healthcare within five to forty years. He revisits AI's history since Alan Turing's 1950s work and the Dartmouth researchers who founded the field, cautions against overregulation, and cites World Economic Forum estimates of a net 58 million new jobs over three years alongside a global shortage of true AI experts. Drawing on his own research into robotic soccer, multi-agent decision-making and human-AI trust, he argues companies must focus on trust and collaboration with AI, distinguishing three ways AI changes work: augmenting human judgment, automating tasks, and creating new jobs.
Key takeaways
- Guttmann predicts children under five today will likely never need to learn to drive, given self-driving cars within 10 to 15 years.
- The World Economic Forum estimated a net gain of about 58 million jobs (75 million lost, 133 million created) over three years, according to Guttmann.
- Guttmann says fewer than 10,000 people worldwide truly understand AI's technology, business and societal implications well enough to move the needle.
- Guttmann's own psychology thesis found extroverts trusted explanation-giving chatbots more, while introverts were not positively affected by the same explanations.
- Guttmann distinguishes three effects of AI on work: augmenting human judgment, automating repetitive tasks, and creating entirely new jobs.
Detailed notes
Generational shifts and hyper-personalization
Guttmann opens with a background note: 25 years in AI, starting with a psychology degree before a PhD in AI, roles at IBM, British Telecom and Hewlett-Packard, and now a vice president role plus entrepreneurial and VC activity. He observes that children today have no experience with cassette tapes or fixed telephone lines, and argues the next generation, children now under five, will likely never learn to drive because self-driving cars will be common within 10 to 15 years. He predicts kids will live in a hyper-personalized world where not just the channel but the content itself is tailored to the individual, and, drawing on his affiliation with the Karolinska Institute, predicts AI-assisted detection of abnormalities in MRIs and X-rays will become common within about five years, standard within twenty, and taken for granted, the way today's medicine looks primitive in hindsight, within forty.
Defining AI and why it matters now
Guttmann traces AI's history to Alan Turing's work introducing the Turing test around 1950, and to John McCarthy, Marvin Minsky and colleagues who founded the field roughly seventy years ago, which has since grown into about 25 subfields. He argues against demanding an overly strict definition of AI but expects growing calls for professional registration in AI, similar to medicine, given AI's effects on people's finances, health and rights, while cautioning against the kind of heavy regulation he sees emerging in Brussels and Berlin. On why AI matters now, he cites Gartner's assessment of AI as the most disruptive technology class for the next one to two decades, an Obama-era US government report on AI as a general-purpose technology, and Finland's early national AI strategy. He cites World Economic Forum estimates of 75 million jobs lost and 133 million created within three years, a net of about 58 million, arguing "AI does not kill jobs, it kills tasks." He also notes, citing Wired and New York Times estimates, that fewer than 10,000 people worldwide truly understand AI's technology, business and societal implications well enough to move the needle, with only about five in Sweden and three in Finland by his own mapping, and warns that Europe is a net exporter of AI talent to the US and China.
Building trust between humans and AI
Guttmann connects his interest in AI to psychology and human connection, describing his own psychology thesis building chatbots that explained their decisions in a real-estate scenario to increase user trust, finding that extroverts responded positively to more explanation while introverts did not. He says trust-building is now a central concern across industries, involving techniques such as intent recognition (understanding what a user means, not just what they type), sentiment analysis, biometric identification through typing patterns, and multimodal interaction that reads stress or emotion, noting that maintaining a dialogue over many turns remains difficult for most chatbots today.
From robotic soccer to autonomous team decisions
Guttmann describes his master's thesis on autonomous robotic soccer, where robots built a shared world model and coordinated as a team, calling it one of AI's grand challenges alongside chess, referencing Kasparov's 1998 match and AlphaGo Zero. He notes such multi-agent coordination has direct applications in search and rescue or hospital settings where robots must collaborate with staff. His PhD work investigated how autonomous teams exchange and aggregate information to make optimal joint decisions, research he says is now applied in healthcare, where care teams must coordinate decisions together.
How AI transforms value chains
Guttmann distinguishes three ways AI changes the tasks that create value. Augmentation improves human judgment, for example helping radiologists or pathologists interpret patterns in MRIs and X-rays better than either humans or AI alone. Automation replaces repetitive tasks, illustrated by a solar-powered robot with a deep learning algorithm that identifies and sprays weeds in a field, work he says few people want as a career. New tasks are also emerging, such as designing chatbot personalities that reflect a company's values for banks and insurers. He closes by urging companies to focus on trust in AI systems among employees and smooth collaboration between humans and increasingly autonomous, "agentic" AI systems that must align goals with each other and with people, a research area he says has existed for 35 years.
Q&A highlights
Asked whether people might over-trust AI systems the way they already over-trust computer workflows even against their own judgment, Guttmann agrees this is already happening in some fields and expects it to increase as AI outperforms humans on specific tasks, but says the bigger current problem is under-trust and workflow friction, such as sales teams not naturally following AI-recommended leads. He adds that in some cases the choice to trust or not may disappear altogether, citing an Amazon patent for a "ship-then-shop" model that would predict and send items to customers before they order, arguing companies that adopt such predictive approaches could outcompete those still waiting for customers to browse and choose.
Quotes
AI does not kill jobs, it kills tasks.
Christian Guttmann
I'm not a blind advocate of AI.
Christian Guttmann
Questions answered in this talk
How many jobs will AI create and destroy, according to the World Economic Forum?
According to Christian Guttmann at Rise of AI 2019, the WEF estimated roughly 75 million jobs lost and 133 million new ones created within three years, a net gain of about 58 million jobs.
How many true AI experts exist worldwide?
According to Guttmann, citing Wired and New York Times estimates, fewer than 10,000 people worldwide truly understand AI's business, technical and societal implications well enough to move the needle, with only about five in Sweden and three in Finland by his own mapping.
How does AI change the way work gets done, according to Guttmann?
According to Guttmann, AI transforms work in three ways: it augments human judgment (for example helping radiologists interpret scans), automates repetitive tasks (such as a solar-powered robot spraying weeds in a field), and creates new jobs, such as designing chatbot personalities.