Artificial Intelligence Conference at UNESCO – A conference report
by Claudia Schmitz
Paris, July 11 & 12, 2019: Full Program Artificial intelligence and the next generation of competences: How Digital – and Artificial Intelligence – will impact jobs and competences profiles?
Organised by The European Chair on Intellectual Capital, the University Paris-Sud and UNESCO’s Intergovernmental Information for All programme (IFAP)Scientific Direction of the conference: Ahmed Bounfour, Professor, European Chair on Intangibles, Paris-Sud-University, email@example.com
There were ten panels and dozens of speakers during two days. This personal conference report focuses on those that resonated with me, or provoked with new ideas. The purpose is to keep some aspects and give references to more data online. As I am a practitioner, I am more interested in what is done in the industry, which is reflected in my long report about AXA.
The conference had the goal to address the impact of digital technologies on the future of employment and the accompanying skills. Within this setting the discussion went from the modification of the content and working conditions in terms of job losses in some sectors and their creation in others. Most of the speakers confirmed the studies around this but made clear, there are new jobs created for which we do not even have a name for today. The other main topic of the conference was the evolution of technical and social skills through new initiatives and digital advances in technologies as well as the challenges that accompany them.
Session 2: The keynote by Leonard Nakamura, Emeritus Economist, Federal Reserve Bank of Philadelphia, USA started with an example of the use of AI in wholesale business, when Amazon invested highly in structures (structural capital) 20 years ago. Many could not understand that the strategy was growth over profit, however, with this strategy Amazon could attack the profit of the other players like Walmart in the long run. Walmart was weakened, lost profit and needed to act to keep customers.
Another area which we do not know how to handle well is education: Nakamura pointed out that we already are a knowledge society, but the big question is the value of education for example in the field of medicine. Will the old model of presence learning in universities keep relevant? Or will online courses like the Khan academy and MOOCs (Massive open online courses) like Coursera take over?
His big point is our measurement system with GDP. This model is based on transactions – of goods produced, distributed and soled. We measure the amount of the invoices in the GDP. But there is a blind eye: There are services that cost no money. So there is no GDP. Within the new business models there is no transaction fee that we can measure in GDP. Like Facebook, like it or not. In the social media we use a service for free and in return we offer our private data. The published turnover of Facebook is the advertisement and the data sold. Apples exports its IP knowhow to Foxcoon China but there is no invoice for this export. But after the production Apples re-imports smart phones to the US. So China generates an export of I Phones and this adds to their GDP. What do we measure here? Does this linear system reflect the complexity of trade in this world? Article about the GDP in China For GDP we measure transactions and movement of money…. if there are no revenues – we cannot measure them. And it has to be in dollars – as we do not look at cryptocurrencies (and not at barter). We have been lacking measurement for intangibles for decades, but as they will becoming more important, we need models that take them into account.Read more https://www.philadelphiafed.org/research-and-data/economists/nakamura More about GDP and measurement https://www.philadelphiafed.org/-/media/research-and-data/publications/working-papers/2019/wp19-11.pdf?la=en
Session 3: Valerio Dilda, McKinsey France, presented on “AI and the future competences” and the Digital Transformation. His summary of the impact of AI: there is a lot of research on transformation, on companies, on people, on jobs etc. His three messages are:
- The shift is bigger and faster than we thought
- The change is good
- The role of companies is important
Valerio Dilda said: „We know how to do it. There have been massive large scale upskilling of people in the past century (after WW2), business schools trained managers, the labour market moved – we did it already, we have that positive experience. But in those days it was largely institutions who provided the training, in the future it will be companies who train their staff. Our learning from the recent past is, that all jobs today will increasingly require digital skills. The speed of change is fast, new jobs will comprise 70% of digital skills in the future. Even jobs like nursing and car repair will use data. Most of these trainings will be done on the job. Till 2030 45% of the population will go through retraining. By 2030, only 15% of the jobs we have today will exist, but there will also be 40% creation of new jobs for which we don’t have a name yet.“
The skills that are needed are higher cognitive skills for complex issues on one hand and emotional skills for better leadership and better communication. The third part consists of technological digital skills, as people need to understand what the computer does and why. Based on the, the role of HR will change. To create the workforce of the future we need to establish:
- Organisational flexibility
- Independent work – freelance
- “new colour” jobs
- Learning culture
More in the McKinsey article https://McKinsey/featured-insights/future-of-work/jobs-lost-jobs-gained-what-the-future-of-work-will-mean-for-jobs-skills-and-wages
Most of the speakers brought up learning perspectives and learning content discussed over the last decade and although nobody mentioned it, Peter Drucker and Peter Senge came to my mind. Some thought we have reached this challenge already, but besides all the soft-skills of better management, there was a consistent and practical topic: How can people in a society learn what is important in the future? What can we offer for self-learning?
Session 4: One answer came from Finland, where a collaboration between the University of Helsinki and Reaktor, created the online course Elements of AI. Ilmi Salminen presented the online course www.elementsofaI.com
“The Elements of AI is a series of free online courses created by Reaktor and the University of Helsinki. We want to encourage as broad a group of people as possible to learn what AI is, what can (and can’t) be done with AI, and how to start creating AI methods. The courses combine theory with practical exercises and can be completed at your own pace.”It takes 30 hours and includes 25 exercises.
“The future of work needs agile training systems and lifelong learning” UNESCO
Definition of AI “Artificial Intelligence” refers to a set of computer science techniques that enable systems to perform tasks normally requiring human intelligence such as visual perception, speech recognition, decision-making and language translation. Machine learning and deep learning are branches of AI which, based on algorithms and powerful data analysis, enable computers to learn and adapt independently.”
The title “I´d blush if I could” refers to the critical point in the gender bias of machines that behave like young women. This is the answer of Siri to a sexist remark of her user …
Session 5: Mark West, spoke about the UNESCO study “I´d Blush if I could”: Closing the Digital Skills Gender Divide with Education and training”. He presented the findings of the Gender Gap which is extreme in AI. Most people use AI in their mobile phones and with smart Siris and Alexas – all of them serving with a female voice. The study provides a powerful illustration of gender biases coded into technology products, pervasive in the technology sector and apparent in digital skills education.
- reflects, reinforces and spreads gender bias;
- models acceptance of sexual harassment and verbal abuse;
- sends messages about how women and girls should respond to requests and express themselves;
- makes women the ‘face’ of glitches and errors that result from the limitations of hardware and software designed predominately by men; and
- forces a synthetic ‘female’ voice and personality to defer questions and commands to higher (and often male) authorities.
What to do about it? Where to start? Is here an ethical code for programmers and companies? Do we need legislation to step into that process? We clearly need to address this trend, as the sexist pattern will influence all users now and in the future – it is a process of changing and undermining culture by using biased design.
Saniye Gülser Corat, UNESCO Director for Gender Equality, says “The world needs to pay much closer attention to how, when and whether AI technologies are gendered and, crucially, who is gendering them.”
The publication shares the first United Nations agency recommendations regarding the gendering of AI technologies, imploring companies and governments to (quote):
- end the practice of making digital assistants female by default;
- explore the feasibility of developing a neutral machine gender for voice assistants that is neither male nor female;
- programme digital assistants to discourage gender-based insults and abusive language;
- encourage interoperability so that users can change digital assistants, as desired; and
- require that operators of AI-powered voice assistants announce the technology as non-human at the outset of interactions with human users.
The study is online and open access, so hopefully, many read it and take some action. What is absolutely striking is the fact that there is a relation of society’s culture and the attractiveness for women to study technology: Main point of the findings quote: “Countries that score higher on gender equality indices, such as those in Europe, have the fewest women pursuing the advanced skills needed for careers in the technology sector. Conversely, countries with lower levels of gender equality, such as those in the Arab region, have the largest percentage of women pursuing advanced technology degrees. As an illustration, in Belgium only 6% of ICT graduates are women, while in the United Arab Emirates this figure is 58%. This paradox is explored in detail and underscores the need for measures to encourage women’s inclusion in digital skills education in all countries.”
Find the study here: https://en.unesco.org/news/first-unesco-recommendations-combat-gender-bias-applications-using-artificial-intelligence
Session 7: A practitioner from the industry, Cécile Wendling, AXA France, Head of Foresight, presented on: “The future of work and upskilling”. The starting question: Is human work threatened by automation? That is not the case, she suggested, but the threat are in demography. AI and technology are onlythe tip of the iceberg.The real problems we have are still under water and are much bigger: Skills, education, mindset, age.
The actual topic is: Is basic income so humanistic as an idea? We need to ask the question: “To replace people by robots – what does it mean in our society and in different societies (e.g. Japan)”. How can we use robots to reduce risks? Like in the discussion on self-driving cars, most of the discussion around it is about ethics. Another ethical topic about using AI: If we are on the phone, do we know if we talk to a human or to a machine? Should we know? More details can be found in this article from AXA:
Cecile presented six domains about new ways of working (architecture) how to design the future of work.
Chart: Ongoing and upcoming transformations
Change 1: The diversity of the workers profile is influenced by the widening of the workers’ age gap, the new skills required by the digital transformation and populations´ movements and evolutions.
Change 2: The mosaic of working patterns derives from new management methods, workforce externalisation and rise of interactions between human and machines, also called cobotics
Change 3:The aim of work and sense of purpose are being questioned and redefined by growing automation.
Change 4: the job polarization may increase as the future of middle-skilled employment is threatened by automation at lower cost.
Here are some details for the mentioned changes to come:
Back to the topic “The future of work”: Cecile put up an interesting topic of demography. “How to work with old workers?” It is not about the age, but the reference frame, even the language is different. If we have diverse teams, it is not the skills of a person, but the social interrelation that makes the diversity profiles.Migrant add to diversity as well, but they come skilled to another country, and often have to be reskilled. Another topic in our workplaces is purpose. If you think that your work has no impact, you might get bore out and burn out. These are the new diseases we have to face. WE talk about sense making … but what if all the sense is not clear to me? What if AI creates meaning out of data which I do not understand? What gives me the feeling of doing something that counts?
Cecile reported about different attempts to get a relationship to learning – to be self- motivated, self-organised and driven by purpose. The old HR offerings cannot cope with the need to learn for all and always. AXA tried out a new partnership with Coursera to get certified, they also implemented a company-wide learning week, where even the CEO talks about his learnings. We need to deconstruct how we work and get a systems view – learn by doing to understand the transition.
In a related conference about AI she talked about TRUST and that AXA has made the decision not to sell the data of customers. Even if this sounds normal for most of people, it is common practice in business, to create business models around customer data.
Question from the audience: What did you change in HR so far?
Future of work: The network of HR in AXA is sharing experiments all over the world in the
Learning and Development team: employee experience on learning internally, social interaction with learners. For example: In the learning week I can challenge my colleague on his/her learning goals. HR organises peer-to-peer trainings. We need to get away from ticking the box, “what was my training this year?” to a habit of a day-to-day learning basis. Change your own habits on learning. AXA has established several online learning tools and calls it: download the app, upload to learn. It is a day-to-day process and reflects the lifelong learning idea. It is not one seminar, but I learn every day, online or offline. It is a mindset that is necessary. So as a manager I need to take care that my team is learning. At Axa we have signs in the offices asking: What did you learn today?It is a mindset that needs to be fostered in the whole company. Designed and developed by the Digital Academy, in close collaboration with AXA France & AXA Winterthur, „Do you speak Digital?“ aims, through a „social-media friendly“ platform, to help AXA employees to gain a deeper knowledge of the digital world by learning and sharpening their curiosity about the major digital trends, in order to better understand the impacts of the digital revolution. Learn how to adopt digital tools and make optimal use of them, while also knowing how to handle them safely and protect their e-reputation. More here https://www.axa.com/en/newsroom/news/new-learning-experiences
Session 7:“Real-Time Management: When AI goes Fast & Flow” presented by Pernille Ryden from theTechnical University of Denmark Associated Professor and Omar A El Sawy, Marshall School of Business at the University of Southern California.
Their research is based on the findings of behaviour, expectation and reaction. “We live in a real time enterprise – we order an Uber, get a pizza … our expectation is fast. But if we only go faster and faster, we will not improve and get burned out. So we need to think about a Fast & Flow framework – and this is time management. Scenario fast: Business as usual, just faster and more complex. So how can we keep doing it? We need to find a way to extreme customization and think from the customer journey perspective. When the order is done, we need to synchronize the systems around the customer and act in real time by confirming the order, one expectation is already met – the product comes in time (but not immediately). Externally the customer has different perceptions of time which need to be handled even when the time perception internally is different. More details: California Management Review
Abstract “In digital business environments real time has become a critical strategic enabler. Enforced by new digital and social technologies we consider the temporal phenomenon real time to play an increasingly central role for strategic value creation and competition. More than ever, managers use real-time information, real-time events, and real-time processes to improve their businesses. Real-time data flows enable managers to dissect events in real time, shorten their decision-making time, and at the same time deepen their understanding of markets. Moreover, real-time technologies offer the possibility of supplying markets with whole new types of services and experiences, which affects the tasks, the timing, continuity, and rhythms of the organization. But the developments in the digital economy also create a diverse sensing of real time, which triggers central questions of how should management adjust strategic decision-making processes towards a real-time-based business, and second, how can the organization build the conditions needed for capturing these strategic opportunities with real time? The increasing use of real-time technologies in business encourages us to unravel decision-makers’ real time perceptions and practices in strategy and organization. From a theoretical and historical outline and empirical findings this chapter aims at bringing knowledge to the increasing influence of real time on strategic management. This knowledge is vital for understanding the tenets of real-time management and to better guide managers and organizations into future strategic directions.”
Session 7: Colin de la Higuera, UNESCO Chair for training of teachers through OERs (Open Educational Resources) University of Nantes
Colin de la Higuera’s presentation “Key digital and information literacy competencies for AI” started with the finding: The gap between the real world and thhe digital world gets bigger – dramatically. Why do we need to teach AI? When people have the idea that AI is just another step of speed of PCs, they do not understand that algorithms do things differently – not just faster. Discussion point: do we understand 100 terabytes of data? Or are we just relying on another algorithm to “explain”? If there is an algorithm A which can explain what the algorithm B is doing, then A is more powerful than B. It is about understanding AI, not doing it. Data is not consistent – data is a source of uncertainty, but data is really what is making the decisions. When the order in data surge is excluding women, they will never show up in the data. The same with age, race, social background etc. If we rely on these searches, we are in the middle of bias which put it into it. It will matter to be or not to be inside the right population. Bias is a big problem. The other issue is randomness – or luck. If the computer filters out 5 students – is that fair? Language and words are another topic. The Inuit have 50 words for snow and ice. Our language is poorer: In our education system we have one word expressing randomness, stochasticity, probability, uncertainty, non-determinism etc. How can we avoid mixing up things and make two different decisions with the same data, because our reference frame is different? Another topic of importance is of making data visible. And lastly: critical thinking – relying on common sense? What is a fake, how do I recognise a fake? Is it enough? Some fakes are too good – better than the original. De la Higuera suggested that people coming out of school should have learned a bit of coding and should have learned that life is not deterministic.
We all know about computer and chess. They programmed the IBM computer Deep Blue – with all the data available for centuries about how to play. Then they worked with Poker – where there is nearly nothing written about to train a machine. The machine DeepStack has to play against itself – millions of times – and learn a strategy to win. The only given is the rules of the game. DeepStack bridges the gap between AI techniques for games with perfect information—like checkers, chess and Go—with ones for imperfect information games, like poker, to reason while it plays using “intuition” honed through deep learning to reassess its strategy with each decision. With a study completed in December 2016 and published in Science in March 2017, DeepStack became the first AI capable of beating professional poker players at heads-up no-limit Texas hold’em poker.https://www.deepstack.ai
YouTube Michael Bowling – „Artificial Intelligence Goes All-In: Computers Playing Poker“
We need to understand that AI is men/man made and that there is always bias in data (gender – race) and if you feed the machine with biased data – you receive biased results. We need to implement algorithms to find the bias ….
Session 8: In a special session on Russia: Society of digital equality – challenges and perspectives Mikhail Nasibulin presented on “Digital economy of the Russian Federation”- how Russia works with knowledge and information. The Government has the most data to be stored, so cybersecurity software is very important to protect data of people. Alexandra Adaskina from the State University of Management is a leading specialist of the UNESCO Chair “Societal, legal and ethical frame working of knowledge societies”. She presented a project about a business game that helps students understand the role of UNESCO.
Conclusions at the end
The conference had several more panels which are not covered in the report. Reflecting over the input some weeks later, I see there was not much really new information for me – but I see an ongoing attempts to get away from teaching “knowledge” and come closer to enable „the ability to act in context“ which was a key mantra of the knowledge society in the 90ies already. We need to train our employees for jobs that do not exist today. So how could we train facts and figures? It is much harder to find the real problems than to find a solution. The preparation to work with AI in the future is not that easy and it is not to teach like “how to” do it right. The companies and Universities are in the dilemma of preparing mindsets to learn quickly, making the right analyses and take conclusions from it to make decisions. This is not new – even in the old Rome – the military had to do this. Tomorrow we have more data and we have more speed, all is connected and complex. I wonder when we will question our learning systems in general – but this is a topic for another conference and for the Universities. The recent newsletter of EFMD had this topic again. „Shot down or broaden our horizon“
Perhaps our AI programs will be smarter than we are today, as they have larger capacities and less emotions to disturb. Again I can refer to my article in this blog about the books of Yuval Noah Harari, the history professor from Israel, who asks exactly the question in his new book: The Lessons from 21rd Century: „When will the AI take over?“ And Elon Musk has started a new company in Silicon Valley to implement IT in our brain so we can better cooperate with AI. We need to keep our eyes open and our minds sharp.