Artificial Intelligence and Knowledge Technologies in a Post-Corona Society
This is a preview of a chapter of The Knowledge Graph Cookbook that was published on 20 April (get your copy here).
As of this writing, we’ve entered the fourth week of quarantine and are probably only at the beginning of what has become the world's largest crisis since World War II. In a few months, the fog will lift and we will be able to see more clearly the consequences caused by the coronavirus. One thing is certain, the outbreak of the pandemic will change all of our lives forever: our patterns of social behavior, the way we work together—now and in the future—how we research and search for solutions as a global community, how we reorganize our supply chains, and how we will think about big data, surveillance and privacy.
At the present time, the internet has become a central infrastructure to ensure the continued existence of many of the systems around the world, and how crucial it is to have data, information, news, and facts that can be trusted, accessed, processed, and networked at lightning speed. Many people, even entire industries, did not see it that way until very recently, but now it has probably become clear to everyone.
“As humans have spread across the world, so have infectious diseases. Even in this modern era, outbreaks are nearly constant, though not every outbreak reaches pandemic level as the Novel Coronavirus (COVID-19) has .” Virus outbreaks are inevitable, but next time we should be better prepared, and for that, we should build systems and societies based on trust.
The post-corona era will divide the world into two:
- Countries where the acceleration of digital transformation is based on recognizing the importance of evidence-based decision-making, the need for data quality, and the crucial importance of linking people and organizations across borders to benefit from explainable AI
- Countries that use Big Data and AI to build societies that are centrally governed by a few, using pandemics as a pretext to increasingly instrumentalize people as data points.
In which environment do smart networking technologies unfold—where the benefits of people and citizens are at the center, where the diversity of ideas, knowledge, and research is stimulated in such a way that sustainable and countable results are achieved?
Where are resilient societies  emerging in the post-corona era, developing strategies that will be effective in the next—possibly even more catastrophic—pandemic?
Let's take a look at some of the possible building blocks of a post-corona society and at upcoming trends that we should pay attention to in order to shape our new future in a humane way.
Self-servicing Based on Explainable AI
The economy and public administration are now in turmoil and under enormous pressure to cut costs, and at the same time, a door has opened that is pushing the use of AI to provide cost-saving self services.
Digital self-service services will be ubiquitous, they will support many more interactions between citizens and public administration than today, they will complement existing e-learning services (for teachers and students), they will serve younger and older people, in health care, to acquire financial literacy or even to plan the next trip to be economically and ecologically balanced, in short: conversational AI will help to make the "right" decisions.
As described above, however, this is happening in different countries under diametrically different circumstances. While in some regions of the world explainable AI (XAI)  and Big Data are being developed for peoples’ benefit , in other regions this is happening under very different auspices: by using knowledge graphs, complete digital twins of citizens are being generated and ultimately used against the individual in order to prevent individual behaviour, to destroy diversity, to make the future allegedly “predictable”.
Gartner recommends that Government CIOs must “leverage the urgency created by the virus outbreak to accelerate the development of data-centric transformation initiatives”, and further on they state that “the increased need for transparency and improved decision making is putting greater emphasis on data centricity, while exacerbating ethical issues .”
Fight Fake News and Hate Speech
To a large extent, the degree of the pandemic is due to the fact that even before the outbreak of the crisis, but primarily during it, false news and opinions were constantly spread via fake news spinners like Facebook and other social networks, but also via so-called 'established' media. As mentioned above, the foundation of a resilient society and its organizations is built on trust. Every wrong message and every hate posting undermines this foundation a little bit more. And it was during the pandemic that the vulnerability of digital systems in this respect became apparent, with Facebook having to send home thousands of content moderators while at the same time relying on AI algorithms to ensure that false messages like medical hoaxes could not spread virally across the platform. Facebook’s CEO Mark Zuckerberg acknowledged the decision could result in “false positives,” including the removal of content that should not be taken down .
Considering that even big data technology giants have to employ thousands of people who have to manually classify their content, one can easily deduce how impossible it will be—at least in the near future—to rely on any AI without the human-in-the-loop (HITL) . The approaches to combat fake news and hate speech will be a mixture of AI, HITL, and stricter policies and regulations. Let's stop trusting tech giants who have told us over and over again how resilient their AI algorithms are. The virus revealed their limitations within days.
HR at the Heart of Learning Organizations
Qualified employees and human resources will become increasingly important in a post-corona society and its organizations that want to base their values and business models not only on data, but above all on knowledge, in response to increasingly dynamic environments. Many organizations will have learned at least one thing from the Corona pandemic: self-motivated, self-determined, networkable and knowledgeable employees form the foundation of every company, one which can remain resilient and capable of action even in times of crisis. While some have closed their borders and put up their blinders, others have sought out collaborators and have intensified global networking, especially within the pharmaceutical industry. “While political leaders have locked their borders, scientists have been shattering theirs, creating a global collaboration unlike any in history .”
Paradoxically, where networking is becoming more important, the human being is again at the centre, and on a level above this, the "learning organisation"  now comes into play.
“It is not the strongest of the species who survive, nor the most intelligent; rather it is those most responsive to change.”—Charles Darwin
HR management in a learning organization can benefit from semantic AI and knowledge graphs in many ways: semi-automated and more accurate recruitment, more precise identification of skills gaps, semi-automatic orchestration of knowledge communities within an organization, working law intelligence based on deep text analytics, e-learning systems based on semantics , job seekers identify opportunities that match their skill sets, etc.
Rebirth of Linked Open (Government) Data
"Linked Open Data" experienced its first heyday around 2010, when organizations around the world and government bodies in particular—at least in the long term and in terms of society—recognized and invested in the added value of open data. It has since become clearer that added value is created when data is based on interoperable standards and is therefore machine-readable across borders. For example, even in 2015 the European Commission still looked optimistically into the future and announced in their study on the impact of re-use of public data resources that ''The total market value of Open Data is estimated between €193B and €209B for 2016 with an estimated projection of €265B to €286B for 2020, including inflation corrections .”
Expectations were probably very high and since then, the Open Data movement in general has stagnated and what the 'Global Open Data Index' stated in its last report  in 2017 continues to be the main obstacle to overcome before we can make use of open data on a large scale:
- Data findability is a major challenge and a prerequisite for open data to fulfill its potential. Currently, most data is very hard to find.
- A lot of ‘data’ is online, but the ways in which it is presented are limiting their openness. Governments publish data in many forms, not only as tabular datasets but also visualisations, maps, graphs, and texts. While this is a good effort to make data relatable, it sometimes makes the data very hard or even impossible for reuse.
The scientific community is already doing better, which has paid off during the pandemic. By applying the FAIR principles to their data, such as the open research data set COVID-19 , which contains the text of more than 24,000 research papers, or the COVID-19 image data collection , which is supporting the joint development of a system for identifying COVID-19 in lung scans, a cohort of data scientists from around the world has been brought together to achieve a common goal.
Governments and public administrations would be well advised to finally learn from science and, after years of chaotic Open Data efforts, to finally bring their data strategies to a level that takes into account the FAIR principles, and thus Semantic Web standards.
The Beginning of a New AI Era
Before the outbreak of the pandemic, AI had been heralded as a great promise of salvation, and its litmus test: the virus. So could AI pass this test? Yes and no. COVID-19 has turned reality and the future upside down, and with it all the models that were trained before the outbreak .
The COVID-19 crisis has exposed some of the key shortfalls of the current state of AI. Machine learning always requires a large amount of historical data, and this data is not available at the beginning of a pandemic, or more generally, during times of change. By the time they are available, it is often too late. So Deep Learning is AI for the good weather, but what we need is an AI that can learn quicker and can produce answers to questions, not only predictions based on obsolete data.
This can only work when AI can make use of human knowledge and creativity, and is able to make abstractions. Thus, AI systems need support from machine readable knowledge models; additionally, collaboration is key! “Efforts to leverage AI tools in the time of COVID-19 will be most effective when they involve the input and collaboration of humans in several different roles .”
This all requires a major reworking of our AI architectures as depicted above, which should be based on the Semantic AI design principle.
For everyone’s safety, the use of personal health data will experience an unprecedented proliferation and it is imperative that it is based on the HITL and FAIR principles, otherwise we will either live in societies that are underperforming in combating pandemic outbreaks or other crises, or that are overperforming in surveillance . Only by applying the FAIR and HITL principles to AI we can bring this into balance. This must be placed in an appropriate legal framework and should become the cornerstones of a new AI era.
 Visualizing the History of Pandemics (Nicholas LePan, 2020), https://www.visualcapitalist.com/history-of-pandemics-deadliest/
 After corona: The Resilient Society? (Zukunftsinstitut, 2020), https://www.youtube.com/watch?v=g0lncocYIiY
 Explaining Explanations: An Overview of Interpretability of Machine Learning (Leilani H. Gilpin et al., 2019), https://arxiv.org/pdf/1806.00069.pdf
 Gartner, Inc: ‘How COVID-19 Will Impact Government Digital Transformation and Innovation’ (Andrea Di Maio, Ben Kaner, Michael Brown, 2020), https://www.gartner.com/en/documents/3982374
 Facebook sent home thousands of human moderators due to the coronavirus. Now the algorithms are in charge (The Washington Post, 2020), https://www.washingtonpost.com/technology/2020/03/23/facebook-moderators-coronavirus/
 Gartner, Inc: Design Principles of Human-in-the-Loop Systems for Control, Performance and Transparency of AI (Anthony Mullen, Magnus Revang, Pieter den Hamer, 2019), https://www.gartner.com/en/documents/3970687
 Covid-19 Changed How the World Does Science, Together (The New York Times, 2020), https://www.nytimes.com/2020/04/01/world/europe/coronavirus-science-research-cooperation.html
 Building a Learning Organization (Olivier Serrat, 2017), https://link.springer.com/chapter/10.1007/978-981-10-0983-9_11
 A Survey of Semantic Technology and Ontology for e-Learning (Yi Wang, Ying Wang, 2019), http://www.semantic-web-journal.net/content/survey-semantic-technology-and-ontology-e-learning
 The FAIR Guiding Principles for scientific data management and stewardship (Mark D. Wilkinson et al in: Scientific Data, 2016), https://doi.org/10.1038/sdata.2016.18
 Creating Value through Open Data (European Commission, 2015), https://www.europeandataportal.eu/en/highlights/creating-value-through-open-data
 The State of Open Government Data in 2017 (Danny Lämmerhirt et al, 2017), https://index.okfn.org/insights/
 COVID-19 Open Research Dataset (Allen Institute for AI), https://pages.semanticscholar.org/coronavirus-research
 COVID-19 image data collection, https://github.com/ieee8023/covid-chestxray-dataset
 What Happens to AI When the World Stops(COVID-19)? (Ian Rowan, 2020), https://towardsdatascience.com/cf905a331b2f
 AI can help with the COVID-19 crisis - but the right human input is key (Matissa Hollister, 2020), https://www.weforum.org/agenda/2020/03/covid-19-crisis-artificial-intelligence-creativity/
 COVID-19 and Digital Rights (The Electronic Frontier Foundation), https://www.eff.org/issues/covid-19