Event description
Summary Report from the session with Dr Jovan Kurbalija
Welcome to the ‘Wondering Birds’ club!
Membership in this club is not required! However, it is ‘required’ to use critical thinking when navigating AI-driven changes, which are frequently counter intuitive.
In this spirit, you can consult AI assistant for the Summit of the Future:
1. EspriTech de Geneve: Where technology meets humanity
EspriTech de Geneva is about Geneva’s role in shaping an interplay between technology and humanity. One starting point for exploring this historical and current role of Geneva could be the International Telecommunication Union (ITU), the venue of our meeting. Among many interesting moments in ITU’s history one was the first online meeting held between UN HQ in New York and ITU back in October 1963.
If you walk in Geneva, you can find many places where critical historical figures of EspriTech Geneva were born or live.
Here is a summary of the main contributions of EspriTech de Geneve
We come to today’s relevance of EspriTech de Geneva. More than half of global digital governance is centered in Geneva.
2. Demistifying AI: AI is simple, but not simpler
AI is often presented (and perceived) as a complex technology which cannot be understood. It is communicated via a complex terminology and surrounded by mysticism that could be uncovered only by AI ‘priesthood’ (e.g. experts).
This is not a good situation for two main reasons. First, AI is not as complex as it looks. Second, AI ‘priesthood’ is not – often – professionally equipped to deal with critical non-technical issues of AI governance.
In such situation, I prepared a book that describes AI through flags, starting from the core of AI, patterns and predictions, and moving to neural networks and other technical issues that trigger more confusion than clarity in current AI debates.
Patterns
Patterns show up in both AI and flags. Flags can be put into families based on their colours, patterns, and symbols. Nordic crosses are on flags from Scandinavia. Most Arab flags are black, green, white, and red.
AI can identify patterns in large amounts of data, just as we can in a variety of flag designs. Pattern recognition is the functional and conceptual foundation of AI. When you scale it up to the massive amounts of data and computing power of modern computers, you get AI platforms like ChatGPT and OpenAI.
Probability
After we identified patterns among flags, we asked the question: What would be the design of a potentially newly independent state? Here we come to probability, the second pillar of AI. Probability is not a certainty.
This distinction between certanity and probability can be illustrated on the selection of the new flag of Greenland back in 1984. Flags of Nordic region where Greenland is located have the nordic cross in common: Iceland, Faroe Islands, Denmark, Norway, Sweden, and Finland. Based on this rather specific pattern, one would expect Greenland to have a Nordic cross in its flag as well.
In the 1984 selection, one of the proposals was to design the Greenland flag around the Nordic cross. However, this expected certainty was not realised, as 14 members of the selection committee voted in favour of the flag with circle which became new flag of Greenland. The proposal with the Nordic cross received eleven votes.
The moral of this story is that probability does not equal certainty. Even though the pattern of previous flags made no exception for the use of the Nordic cross, the new flag deviated from this pattern due to a political decision.
Here you can go to Diplo’s app for looking for patterns in the flags…
Patterns and predictions underlie the way ChatGPT generates text responses to our questions, as illustrated below…
As we saw previously in an example of flags, AI concepts are simple in their core. AI is an umbrella term with different facets and meanings, as illustrated below with the ‘elephant and blind person metaphor’. Apart from the diversity of AI, this field is heavily loaded with technical terminologies, which add to the overall mystification of AI. In grasping AI, it is essential to insist on simple answers to common sense questions.
3. AI will change us as we discuss AI’s future, including the risks to humanity’s extinction
This slogan paraphrases John Lennon’s song “Beautiful Boy” lyrics:
‘Life is what happens while you are busy making other plans.’
It accurately describes the framing of AI debates in 2023. We were busy debating AI’s extinction risks for humanity, as depicted on Time’s cover page in June last year, while less spectacular but more profound changes started impacting education, jobs, and the economy.
The dominance of long-term AI risks in 2023 has to be re-balanced with more policy focus on short-term and mid-term risks by AI, as outlined below.
Three types of AI risks
Short-term risks include loss of jobs, protection of data and intellectual property, loss of human agency, mass generation of fake texts, videos, and sounds, misuse of AI in education processes, and new cybersecurity threats. We are familiar with most of these risks, and while existing regulatory tools can often be used to address them, more concerted efforts are needed in this regard.
For example, AI transforms education profoundly. The first glimpse of changes came with the impact of ChatGPT on writing essays. The first knee-jerk reaction was to ban the use of AI at some universities. However, it is changing. Educators are increasingly acknowledging AI as new reality which opens new possibilities. You can read here how we use AI in Diplo’s pedagogy. Diplomacy, as a text-intensive activity, will undergo structural and profound changes as a result of LLMs’ language capabilities. More information can be found in the AI and Reporting section, as well as throughout this text.
Mid-term risks are those we can see coming but aren’t quite sure how bad or profound they could be. Imagine a future where a few big companies control all the AI knowledge, just as they currently control people’s data, which they have amassed over the years. They have the data and the powerful computers. That could lead to them calling the shots in business, our lives, and politics. It’s like something out of a George Orwell book, and if we don’t figure out how to handle it, we could end up there in 5 to 10 years. Some policy and regulatory tools can help deal with AI monopolies, such as antitrust and competition regulation, as well as protection of data and intellectual property. Provided that we acknowledge these risks and decide we want and need to address them.
Long-term risks are the scary sci-fi stuff – the unknown unknowns. These are the existential threats, the extinction risks that could see AI evolve from servant to master, jeopardising even humanity’s very survival. These threats haunt the collective psyche and dominate the global narrative with an intensity paralleling that of nuclear armageddon, pandemics, or climate cataclysms. Dealing with long-term risks is a major governance challenge due to the uncertainty of AI developments and their interplay with short-term and mid-term AI risks.
4. The beginning of the end of traditional publishing, PDF and Web
We are on the eve of the next major shift in communication with the gradually diminishing importance of traditional publishing, PDF, and web. This evolution will take some time. However, it should be on the radar of strategic communication planners and younger colleagues.
The main conceptual ‘battles’ will be between, on the one hand, the fluidity and flexibility of the new AI and multimedia technologies and, on the other hand, our need to comprehend data and information through occasional ‘closures’ of shifting arguments and importance of storytelling for comprehending reality. There will be tension between the ‘horizontal’ nature of hypertextual information with so many links and the ‘vertical’ nature of storytelling sequence. You can explore this tension in Borges’s writings including his book The Library of Babel.
PDF as closure and mimic of printed publications
PDF files resemble traditional printed publications. PDF files are sequential, going page by page. It is an excellent form for following texts with a ‘backbone’ narrative, such as most fiction.
PDF is useful in official UN communication where ‘closure’ and formality are required (treaties, diplomatic letters). However, in most cases, PDF’s sequential format does not suit the textual nature of UN documents, which rarely follow a strict narrative structure. You can get in the middle of a UN report or resolution and understand the information without following the previous narrative.
Concretely, you can compare access to information and knowledge in the policy brief via official PDF to the way we presented them, with the option of making ‘deep links’ at any point in the narrative. One caveat: UN website has a good multimedia intro into policy briefs which help access to the content.
Traditional printed publications
Two years ago, I was asked to contribute text to a new publication on digital diplomacy. The production time was two years. It did not make sense. Here is my proposal for dealing with the rapidly changing nature of our knowledge space and the need for occasional ‘closure’ in the form of traditional publications.
Web: from passive ‘postcard’ to knowledge space
The web regained relevance following COVID. However, there will be significant changes ahead of us. Most websites are designed around the organisational ‘postcard’ paradigm, with designers playing the primary role. Websites should evolve into knowledge spaces where visitors can interact with information and knowledge generated by organisations.
This shift will be facilitated by the arrival of AI, which will allow us to access relevant knowledge in massive amounts of data based on our specific needs. Journalists, students, diplomats, and researchers all interact with our websites in unique ways.
Last year, we did analysis of websites of international organisations which can show ‘paradigm’ and thinking behind organising information on websites. You can consult this study here.
Next week, Diplo will introduce this new type of web, which shifts the logic of web development from ‘postcard’ to knowledge ecology.
AI assistants
AI assistants are tangible examples of the AI-driven shift in access to data, information, and knowledge. This shift will affect how we develop websites and share information and knowledge. Here are a few examples from DiploAI:
- access to full knowledge and information on one website: an example of Diplo website (new design will be launched next week).
- access to knowledge generated at one event: workshop on cyber diplomacy in Africa held (Addis Ababa, 15-16 April 2024)
- access to knowledge generated by one policy process such as years of deliberations in the UN Open Ended Working Group on Cybersecurity
- access to knowledge in video materials: Ask Sir Humphrey from Yes Minister
- access to knowledge produced by person in books and blogs: ‘Ask Jovan‘ and ‘Ask Aldo‘ (Diplo’s chief contrarian officer)
- access to one book or text: EU AI Act | PhD thesis on cybersecurity relations between USA and China
5. Story-telling is dead. Long live story-telling.
In previous examples, we saw how information and knowledge can be accessed directly through the use of AI assistants. We access ‘deep knowledge’ contained in documents, videos, and books.
However, storytelling matters! It is ingrained in human cognition, dating back to our distant ancestors. TikTok and other social media platforms are introducing young people to storytelling. Stories are shorter, but they still convey our emotions, expectations, and perspectives. In public spaces, storytelling can act as a pressure valve, a problem-solving tool, and a strategic maneuvering tool. It combines emotional and imaginative engagement with the strength of an argument to make the lesson more impactful and memorable.
Read more about storytelling.
Finding clever ways to blend narration and dissection is a crucial task in managing text and information that lies ahead of us!
6. Beyond hype: UN as safe harbor in turbulent digital times
Solid anchors are essential in an era of technological hype, geopolitical confusion, and global societal tensions. The UN is uniquely positioned to perform this vital and critical function for the future of technology and humanity.
However, it is not often a shared view. During my time at the UN High-Level Panel, I was concerned about how tech companies treated UN diplomats and high officials. We were viewed as ignorants who needed to be enlightened by cutting-edge technology. When we asked specific technical questions, as I did on several occasions, they were dismissed.
My point was and continues to be that the UN and governments have a noble role in representing humanity’s priorities and core values.
Unfortunately, we are not always up to this noble task for a wide range of reasons. Among them is low esteem for diplomacy and politics in general by people worldwide. This gap between a noble role and low esteem must be changed rapidly for the sake of the future of humanity. The change should come with technology as both primary hope and fear of humanity.
It is far more than simply mimicking “the latest tech trends,” as some UN communication attempts to do. The United Nations should use cutting-edge technology. However, this should not be the UN’s main message in the technology field. The main message should be that the UN is assisting in this critical interplay between rapid technological advancements and fundamental human values.
7. Connecting: Left hand learning what right hand is doing
One of the most frequently mentioned themes in the United Nations and governments around the world is the need to break down policy silos. Unfortunately, reality does not align with rhetoric. In 2023, we crowded 43 UN organisation and agency websites. We discovered that 6% of the more than 120 million links connected to one another. In our epistemological study, 20% cross-referencing is required to classify a single system (organisation, business, university) as knowledge ecology.
We also investigated the Geneva scene, including not only UN organisations but also NGOs, academia, and businesses. The percentage of crosslinking was 0.49%. Paradoxically, organisational or even physical proximity has no effect on cognitive performance.
One of the benchmarks of the UN system or any other space that aims to create knowledge ecology is the use of hypertext linkages to evaluate real knowledge among different parts of an organisation or family of organisations, as the UN does.
Mapping of AI initiatives
UN should map AI initiatives in innovative way. Instead of indicaitng what organisaiton X does, it should evaluate how AI system of country X functions in terms of outcomes, data, and weights.
Rediscover knowledge management
Today’s focus is primarily on data. However, this focus is insufficient for quantifying the impact of AI. Knowledge must be restored to organisational rhetoric and dynamics. Fortunately, the United Nations worked extensively on knowledge management a decade ago. These studies and policies must be ‘undusted’ and repurposed for the AI era. Knowledge is at the heart of the AI era, far more so than simple data. The UN Joint Inspection Unit’s an excellent study of knowledge management at the UN could serve as a good starting point. It was led by Ambassador Petru Dumitriu, Diplo’s lecturer. Please let me know if you’d like to be connected with him.
Digital footprint
The digital footprint shows the visibility of Geneva-based organisations from 50 cities around the world for 500 issues searched on Google. Here’s a video explaining how we perform digital footprint analysis.
In addition to Geneva’s digital footprint, you can see digital footpring of the major media houses in covering the Gaza war.
8. AI-supported reporting: Major shift in modus operandi of UN and diplomatic services
AI-supported reporting will have the most profound impact on UN and diplomatic practices. According to our estimates, diplomats and international officials spend more than half of their time reporting. It has significantly increased in the last ten years due to the emergence of a ‘compliance culture’. Diplomacy has become more bureaucratic, focusing on reporting rather than its primary function of resolving conflicts peacefully through negotiations, engagement, and persuasion.
In this context, AI-supported reporting will positively impact diplomacy and the United Nations. However, this transition will be complex as it will require reskilling of staff, refocusing of resources, and overall organisation reform from ‘bureaucratic diplomacy’, focused on heavy reporting in ‘compliance paradigm’, towards core diplomacy aimed at the core functions of multilateral diplomacy to ensure security, facilitate development, and support human rights. This AI-triggered transformation will be the msot critical for the future of the UN 2.0.
Additional advantages of AI-supported reporting will be:
- Efficiency: AI can automate repetitive tasks like recording, transcribing, summarizing, and analyzing large amounts of data quickly and accurately, saving time and resources.
- Accuracy: AI tools can help ensure accuracy in reporting by reducing human errors in tasks like data analysis and summarisation.
- Data Analysis: AI can process and analyze vast amounts of data to identify trends, patterns, and insights that may not be immediately apparent to humans.
- Support for Decision-Making: AI-generated reports can provide diplomats with valuable insights and information to support their decision-making processes.
- Time-Saving: By automating certain reporting tasks, AI can free up diplomats’ time to focus on more strategic and analytical aspects of their work.
- Accessibility: AI tools can be particularly beneficial for smaller missions and developing countries that may lack the human and institutional capacity to follow and analyze complex diplomatic processes.
- Consistency: AI can provide consistent and standardized reporting formats, ensuring uniformity across different reports.
As a practical illustration of AI-supported reporting ou can consult Diplo’s reporting hub with a few examples of reporting:
- UN GA 2023
- UN Security Council Meeting
- Complex events with 100+ sessions: IGF 2023, UNCTAD eWeek 2023.
- WEF 2024 meeting
Here you can find knowledge graph representing relations among topics and speakers during one day of UNCTAD eWeek.
9. AI Impartiality: Major threat (and opportunity) for the UN
The forthcoming AI transformation of the UN will have profound impact on the organisation’s ‘operating system’, which is based on two key principles: impartiality and inclusivity. If the UN Secretariat’s ‘thinking’ is shaped by a specific corporate or national AI platform, it can undermine the UN’s impartiality and pose an existential threat to the organisation.
The upcoming AI transition differs from the UN’s previous major foray into digital innovation, which included data migration to the cloud in the 2010s and online meetings during the COVID pandemic in the early 2020s. In all previous cases, digital technology was used passively for UN negotiations, analysis, and reporting. However, artificial intelligence can actively sharpen the thinking of UN machinery and, ultimately, Member State decisions.
The good news is that the UN can turn this major risk into an advantage by developing UN AI based on the following 15 principles: openness, public good, inclusivity, modularity, professionalism, explainability and traceability, diversity, data and intellectual property protection, security, sustainability, environmental friendliness, capacity development, future orientation, accessibility, and multilateralism.
UN AI can be developed through open-source contributions of AI models from companies and countries. Similar to the UN 1.0 building, which received gifts and donations from member states, UN 2.0 should be developed through the contribution of AI models.
I began drafting an op-ed for UN AI. Please let me know if you have any comments or suggestions for where they could be published in order to spark positive discussions about the risks that the use of AI and the UN will bring.
10. UN 2.0 and Summit of the Future: How to get it right
Policy Brief 11 provides comprehensive, insightful, and creative context or discussion for UN 2.0. As there is little to add, my comments focus on how this change can be implemented, particularly ahead of the Summit of the Future.
Revisiting main building blocks of UN 2.0
It is crucial that the UN 2.0 Policy Brief gives a “quintet of change” cultural context. I would add developing “border spanners'” sensitivity and skills—which should help them see AI and other policy issues from a variety of perspectives—to the skillset section. The UN and governments should begin advocating for the ‘bounmdary spanners’ skills through hiring new employees and providing training to current ones.
Data
As previously stated, data overshadowed the concept of knowledge. It distorts reality as knowledge is more appropriate than data for describing what AI does. Data is a more passive, technical concept. Knowledge entails human interaction and values. It would be great if the UN system could begin revitalising knowledge as a better descriptor of artefacts created by AI and humans.
In search of data-driven solutions, we should not be obsessed with only more data. The focus should be shifted from quantity to quality of data.
Innovation
Innovation should be centred on the organisational response of combining human and artificial intelligence. It is a more non-technical challenge. One of the risks is to package ‘innovation’ as an event, application, or even center/hub or innovation. Such innovation promotion could be helpful but only as it is part of broader organisational cultural change, as Policy Brief 11. suggests. Unfortunately, most current thinking centres condense innovation in events/products/speeches.
I can report a few reflections from Diplo’s approach to changing organisation around the concept of ‘cognitive proximity’ among humans and between us and machines. Diplo has always been very agile with many internal dynamics and matrix organisational structures. However, the shift towards a ‘cognitive proximity’ approach has been challenging over the last four years since we intensified this shift. Diplo has had an almost perfect setting for experimenting with the new type of organisation, including agile organisation in Diplo’s DNA, creative and innovative staff, leadership, technological expertise, and limited ‘external constraints which IOs and governments have (e.g. Diplo’s Board has been consistently supportive for innovation).
Yet, this change of integrating AI into Diplo’s modus operandi has not been simple and easy because it challenges how we have been taught to deal with knowledge and how it has been done for centuries in typical knowledge settings of academia, businesses, and governments. Knowledge and information have been subject to monopolies of a wide range from essential individuals within the structure or companies and organisations about external societal actors.
Our experience from this painful transition requires a lot of patience, understanding individual specificities regarding learning and knowledge, and fast adaptation in organisation. This transition mainly involves a lot of ‘carrots’ (incentives), but, sometimes, ‘sticks’ are needed as institutional monopolies and inertia can primarily foster passive sabotage of new approaches.
You can read here more about Diplo’s Cognitive Proximity approach.
Digital solutions
Digital solutions should advance core UN principles and approaches in practical and tangible ways. For example, whenever the UN asks the global community to provide an opinion – e.g. ‘Have your say’ actions, the UN should respond by indicating what happened with specific submissions and proposals. AI can help identify the ‘destiny’ of proposals by comparing each submission and the final report or document of policy processes. It is just one example of how digital solutions can increase legitimacy and ‘buy-in’ from citizens, communities, and countries worldwide.
Foresight and behavioural science
The core of this aspect of Web 2.0 is the future relevance of ‘choice’. As the illustration shows, we make choices by using logos (brain), ethos (heart), and pathos (stomach), as Aristotle argued.
Today, this ancient decision-making trinity is challenged by machines as computers can make more optimal and informed choices than we can do. Algorithms tend to know us better than we know ourselves. They can do that based on the information they collect about us (what we click on, our likes, interests, data gathered from a Fitbit watch…).
Human choices (freedom) and optimisation (modernity) are two pillars of enlightenment and the above visualised dichotomy between the two could be regarded as the so-called autoimmune disease of enlightenment.
If machines can make optimal choices, they can decide instead of us about our selection of partners, purchase of things, and political preferences. It is a fundamental dilemma ahead of us. We may consider having the ‘right to be imperfect’, which could leave space for our genuine choices even if they are not as optimal as those made by machines.
Making Summit of the Future a new type of meeting
The Summit of the Future could be a pivotal moment for UN 2.0. It should ‘walk the talk’ in presenting concrete tools and actions about the future of the UN and the world.
In preparation for the Summit, we will run online training on ‘Futures literacy’ focusing on how future narratives are developed and used by academia, businesses, media, and governments.
Here are a few suggestions on how the Summit could ‘walk the talk’ by using innovative and future-oriented tools and approaches:
- AI Reporting from the Summit of the Future and other UN GA deliberations;
- Launch the UN AI initiative with the first set of applications donated by member states in accordance open source and public good principles;
- Launch ‘Ask UN’ platform.
