Tallinn Digital Summit 2018 took place on 15-16 October 2018 with delegations from 15 digitally-minded nations across the globe and the European Commission.
The focus of the Summit was on how artificial intelligence and free-moving data can be wielded in the service of our digital states, economies and societies.
Below are 17 general conclusions from the discussions at Tallinn Digital Summit 2018 as drawn by the Summit organisers.
AI will only live up to its promise if the public has confidence in it. We need to be able to explain the benefits of artificial intelligence to all citizens and policy makers in a very broad sense even if we do not have an AI which can easily be explained in technical sense. Thus, it is critical to establish trust and confidence in its abilities to do good at the same time as address potential misuses.
The participating nations at TDS expressed a strong will in getting it right in the area of safety and security of artificial intelligence. We can only get the benefits of AI if we deal with the risks related to it in a well-informed and intelligent manner. The best way to deal with the risks is by using the technology – by playing and experimenting with it. We should start experimenting with small scale AI projects in order to be able to advance with confidence to more sensitive areas.
AI can help to deliver better public services, hence improving peoples’ lives, but we need to realize that today it is still quite incompetent. We have a tendency to put too much faith in it – a lot of the ethical misuse comes from blind trust in AI – when there actually should be more controls in place. We also need to be ready if the AI is or might become too competent – carrying out cyber-attacks we cannot defend against and/or manipulating our own emotions. It is therefore essential that we embed the human factor into AI solutions.
We need to develop common standards, privacy and security practices to ensure good data protection. Since artificial intelligence systems are reliant on the quality of their input data, we need to ensure that there is enough quality input data available and that it cannot be compromised. From technology point of view governments should prioritize interoperable and open AI technologies during procurement and impose rigorous standards for all new digital public sector tools. The technical infrastructure should promote transparency and facilitate external audits so that citizens have access to the data and reasoning of any AI systems that may affect them.
In order to build citizens’ trust in government institutions, they must believe that those institutions will act ethically, rigorously and with careful consideration of their interests. We need to ensure that even against the backdrop of mistrust and uncertainty, legitimacy is an achievable aim.
The ability to innovate and acquire the necessary human capital skills will be among the most important enablers of the AI-driven economic growth. We need to encourage further AI literacy among policy makers, business leaders and the wider public to guide informed decision making. Currently most governments rely heavily on external contractors in the field of AI. To be able to order, build and offer good AI-powered public services, we need to increase the knowhow within the governments.
We also need a regulative framework that supports innovation, and we need to progressively invest in AI research and innovation in a manner that ensures benefits can be shared by all.
AI’s impact on work will be profound. While most scenarios suggest that more jobs will be created than lost to automation, the transition is likely to be disruptive, and occupations and skill requirements will shift significantly. Some occupations as well as demand for some skills will decline, while others grow. Many occupations will change as people start working alongside machines to solve problems together that neither could accomplish alone. There is a need to come up with a solution on how to do skilling at scale.
The adoption of AI is still at a relatively early stage and uneven among countries, sectors and companies. In building a shared vision of an AI-augmented future, governments need to foster education and debate about AI more widely and collaborate with both technology experts and public sector professionals. While there is experimentation going on, there is very little that we actually know from what the others are doing. There is a huge appetite and interest for having a diverse group of people – entrepreneurs, innovators and policy makers in the room to learn from each other. We should coordinate AI initiatives to benefit from each other’s innovation and expertise, establish bilateral, regional and international collaboration formats to address issues related to AI such as safety, bias, explainability etc.
We commonly use the analogy of data as the new oil, electricity – or even air, without which a business suffocates. However, unlike commodities, data is non-rivalrous and renewable: same data can be collected, copied, or shared simultaneously and across borders, and its consumption only increases its value. Artificial intelligence impacts greatly on our ability to export since it improves our ability to understand, predict and better serve customers from other markets, cultures and languages. And if AI impacts on our exports, it also impacts on our economy. If e-commerce were a national economy, it would already be a member of G7 and yet grow four times faster than the Chinese economy.
Access to data is essential for artificial intelligence, as the accuracy and usefulness of algorithms depend on data available. Open access to data – free from collusion by multinationals, trade discrimination and provided under secure forms – is essential for the economy, not least for SMEs. Global service-providers are training algorithms based on globally available personal and non-personal data, SMEs will need to have access to data, computing possibilities, investment opportunities and talent to upgrade their business models for future needs.
Participants took note of the latest developments across the world, including non-discrimination rules on data trade in EU bilateral FTAs, Comprehensive and Progressive Trans-Pacific Partnership (CP-TPP) and the recently published US-Mexico-Canada (USMCA) agreements. Trade agreements complement domestic rules: Privacy laws only determine who can take data out our countries, not whether we can use data from other countries. Moreover, trade rules guarantee non-discrimination in countries that are unlikely to be deemed as equivalent to our privacy rules, or with whom we may not share some fundamental values.
The free trade agreements of today are designed to empower businesses and bind governments from discriminating trade partners. Meanwhile, privacy laws empower citizens and bind business from abuse. Future trade agreements may be able to square this circle, and also empower users to access online services and apps of their choice.
Trade discrimination in data flows and digital trade should also be addressed in the context of WTO E-commerce Work Programme that links to the modernisation of the organisation. As data and AI become essential features for trading in traditional goods and services, some countries could potentially reverse the liberalisation achieved in the WTO. Given inadequate progress at a multilateral level in the WTO, the world will turn to bilateral and regional efforts.
Negotiating trade agreements are not so much about the commitments themselves, but defining reasonable exemptions from these commitments. Trade agreements allow the countries to set whatever standards they wish on privacy and security, while binding its signatories against arbitrary discrimination (through unilateral data localisation, disproportionate security provisions or disclosure requirements on algorithms or source codes, etc.) without legitimate justifications.
Public authorities have a key role in demystifying and encouraging people to use solutions based on artificial intelligence. Public datasets should be open to public and private sector developers, and investments into this field should be encouraged in e.g. autonomous-driving vehicles, AI-assisted medical diagnostics, improving the productivity of our farms and fisheries. The number of applications will also grow exponentially with Industry 4.0 and Internet-of-Things (IoT) etc. One participant brought to the discussions how the full commitment to openness and data sharing as the default option in its public policies, enhanced by principles of making government services “cloud first” and opening the government technology stack to private sector developers.
In the context of EU trade policy, the ongoing trade negotiations with Indonesia, as well as the advanced economies of Australia and New Zealand (and soon with the UK) may provide an opportunity to revisit its digital trade rules. Participants also noted the intention of European Union and Singapore in developing voluntary cooperation on interoperability frameworks for electronic identity and trust services such as e-signatures that will facilitate transactions, e-commerce and trade across borders.
It was repeatedly underlined that cybersecurity, operational resilience, education and e-skills of the people are fundamental to citizens and consumers to reap the benefits of the digital economy and for their continued trust in all the aspects of our digital societies, from our trading system to our electoral systems.