1 Introduction

The growing literature on National AI Strategies (NAS) has paid vast attention to how nationally shared visions shape the way nations imagine their technological future [1, 2]. Such documents were seen as continuing a broader national narrative that gains new meaning in the contemporary technological environment, or those that envision the nation’s role in the ongoing AI race [3]. The Israeli case I present in this paper follows a similar trajectory but also highlights another side to this scholarly narrative that was not addressed. The paper shows how a deeply rooted vision of innovation both shaped the Israeli AI documents and made Israel choose a “national program” instead of a fully-formed and coherent “national strategy”.

The argument I present in this discussion is based on a synthesis between a historical national ethos and a contemporary technological and political vision that are prevalent in Israel. For religious, political and geostrategic reasons, the concept of time in Israeli politics is one of transience. This means that Israeli politics is mostly focused on the present and avoids planning too far into the future. In Jewish thought, the eschatological idea of redemption as being completely unexpected led believers to focus on everyday life and avoid thinking much about the future [4]. This had a direct influence on some elements in Israeli politics, and especially on the idea that reality can change dramatically at any moment. The revolutionary element in Zionist thought also had this kind of influence by articulating the notion that the future is always in the making and without a defined telos [5]. This led to what Eyal Chowers described as the “the valorization of risk and discontinuity” in Zionism and Israeli politics [5]. The sense of a constantly looming existential risk must also have influenced this concept of time in Israel, where the future might seem intimidating and survival necessary. This left Israel without strategic discussions or decisions on some fundamental issues such as a constitution or defined borders. As a government report suggested in 2012, different government offices have some strategic planning but they mostly focus on the needs of the office itself. In Israel, the report concluded, strategic thinking only happens during times of crisis and without long-term planning [6].

This view also shaped the idea of science and technology in Israel. As Noah J, Efron noted, in early Zionist politics “the pioneer-scientist was a hero to kids caught up in the drama of building a Jewish homeland” [7]. Scientific and technological revolutions fitted with Zionist ideas regarding the individual’s power in shaping history and contributing to nation-building. Since the early days of Israel, technological innovation has been seen as a way to gain a strategic advantage that will sustain Israel’s security [7]. This led to government investment in R&D and the training of new scientists which, in the decades to come, set the scene for Israel’s image as a “Start-Up Nation”. Getzoff explains that the meaning of “Start-Up” is “heavily associated with entrepreneurialism, innovation meant to disrupt the status quo of the capitalist economy” [8]. The meaning, then, of a “Start-Up Nation” in the Israeli context is telling. On the one hand, it explains how start-up tech companies are seen as helping Israel become stronger as a nation. On the other, it suggests that Israel itself is a “start-up”, motivated by the need for innovation and entrepreneurship but also one that is always in the making. As the late Israeli president and prime minister Shimon Peres wrote in his introduction the Senor and Singer’s book “Start-Up Nation”, Israel is “a country that is itself a perpetual start-up” [9]. The idea of the “Start-Up Nation”, therefore, encapsulates both the transient aspect in the Israel’s concept of time and the sociotechnical imaginary of innovation as sustaining Israel’s resilience as a nation.

This combination of a historical narrative and a technological ethos shapes how different actors see their roles and the decisions they finally make. As we shall see, the question of why Israel ended up, for the time being, with a national program instead of a national strategy, can be related to similar discussions on how the interests and power of institutions shape policy outcomes in different political settings [10]. Such works showed that a particular historical trajectory structured the power relations between institutions and shaped the available choices that could have been made. As we shall see in this case, this might explain the rejection of the first AI initiative by some government and public authorities and the formation of the second initiative. It might also explain why a “transient” program was preferred over the “stable” strategy.

We should also take into account the constructivist process that creates different kinds of imaginaries that shape social reality [11]. Imaginaries are social constructs that offer individuals a vision or meaning for their individual or social lives. For this paper, we should pay special attention to sociotechnical imaginaries that can be seen as “publicly performed visions of desirable futures”, as Jasanoff put it [12]. Although such imaginaries originate in different kinds of elites (political, technological or scientific) they are publicly held and can shape visions about the common good [12]. As we shall see, the vision of innovation in Israel will gain such meaning in the Israeli context. Innovation is not only seen as something that might drive science and technology forward but also has a broader meaning on Israel’s national resilience.

Based on these two deeply-rooted ideas of transience and innovation in Israel’s politics and society, I will argue in this paper that like other aspects of Israeli politics, Israel is not committing itself to a national strategy in order to remain agile and change its course when needed. This means, I argue, that the same ethos that turned Israel into a self-proclaimed “Start-Up Nation”, is the one that now shapes its AI policy. On the one hand, Israel is committed to scientific and technological innovation, but on the other hand, it does not want to restrain itself with a coherent national AI strategy. As such, the Israeli case continues the growing scholarship on NASs and shows how national visions shape national strategies. Yet the paper also offers a new perspective on this scholarship and suggests that under specific circumstances some countries might purposefully avoid having a coherent NAS and choose a different path in order to maintain both technological and regulatory agility.

To do so, this paper reviews the two documents that form Israel’s ongoing AI initiative. The first document, as I show, offered a detailed and comprehensive vision for Israel’s AI policy. Its main objective was, as we shall see, to gain a leading position in AI R&D and to encourage innovation as much as possible. This vision was based on the previously mentioned ethos which suggests that in order to maintain national resilience, Israel needs to be technologically and scientifically advanced. As I explain, the document was rejected by some public sector actors that did not agree with the document’s recommendations. This rejection coincided with the frequent elections Israel experienced during the past few years, which also delayed the adoption of this strategic document. Instead of adopting this document, it was decided to set up a new committee that relied on some of the work that was done in the first document but offered a more modest vision for Israel. This new report did not recommend broad institutional reforms and focused, instead, on specific elements that would benefit from government investment. The document was not seen as a “national strategy” but rather as a “national program”.

The paper outlines the background for each of these documents, explains their vision, objectives and recommendations and pays close attention to the ethics and regulation sections in each of them. The reason for focusing specifically on ethics and regulation is related to the argument I wish to pursue in this paper. It might be true that having a program instead of a coherent strategy can maintain Israel’s agility, as it did in the past in other contexts, and by doing so help it fortify its national resilience. Yet, in the case of AI and data science, such agility might not give sufficient attention to ethical and regulatory concerns and not see citizens as active participants in this ongoing process.

2 The first report—“The National Initiative”

2.1 Background

Israel began its journey on the National AI Strategy scene quite early. In May 2018 Israeli Prime Minister, Benjamin Netanyahu, launched the “National Initiative for Secured Intelligent Systems” which was vested with the responsibility of creating an Israeli National AI strategy [13]. The Initiative was led by Prof. Isaac Ben-Israel and Prof. Eviatar Matania who also led the formation of the Israel National Cyber Directorate a few years earlier. The Initiative gathered numerous experts from different fields (industry, academia and the public sector) and divided them into fifteen different working groups, each focusing on a different aspect of AI governance, regulation and use [14].

2.2 Vision, objectives and recommendations

The report begins by stating its vision: “To advance Israel’s resilience as a techno-scientific power and its national, economic and social security. To secure Israel’s future as a safe, open, democratic and knowledge-based society” [13]. The main objective, the writers explain, is to turn Israel into one of the five leading countries in the world, during the next 5 years. Other objectives include ensuring Israel’s technological resilience, ensuring Israel’s techno-defence power, promoting economic growth and promoting social welfare [13].

The vision and objectives of this report correspond directly to the ethos of innovation that was discussed earlier. Advancing AI technology in Israel is not only seen as a means to boost the economy but also as one that will enhance Israel’s national security. The writers explicitly address the contemporary AI “arms race”, as they refer to it, which sets the scene for the entire report. They suggest that Israel needs to proactively maintain its leading technological position in a way that would also influence its ability to sustain itself as a country. This creates a sense of urgency that distinguishes this report from other NASs since what is at stake here is not only societal well-being but mainly the existence of Israel as a country. For example, the first question that the writers asked themselves was “Which capabilities and technologies are critical for ensuring Israel’s national security and maintaining it as a leading high-tech power?” [13]. The second question addresses the economic and social benefits that can be gained from such technological leadership. Several actors, such as the defence sector, academia and the private sector, are mobilised for this national cause. These stakeholders, the writer hope, will create a new kind of eco-system that will be able to work together and develop AI in Israel.

The report suggests that AI should become a national priority in Israel, it being “a critical infrastructure for the future of Israel and a national priority” [13]. It also argues that AI is an “infrastructure of infrastructures” which means two things: first, leading in AI will allow leadership in future technological fields. Second, this national effort requires the cooperation of different sectors within the Israeli government, military and private sector. The writers were aware that such cooperation might be difficult but they stressed the importance of the strategic effort that would yield “a coherent plan where the different members of the Israeli eco-system will be able to take part and will improve themselves as a result of the cooperation with others” [13].

Based on the assumption that in the case of AI a strategic effort is needed, and aware that this might not be easily done in Israel, the authors of the Initiative report recommended creating a new “AI Directorate” in the Prime Minister's Office. This directorate was aimed to help integrate the work of government offices and the private sector, to manage the budget for implementing the program and to advise the government on AI-related issues. The writers suggested that the directorate should be a professional body and advised bringing people from the “techno-strategic worlds of AI” [13]. The writers considered this directorate as critical and explained that “dispersing the resources and efforts in this field without a systemic vision of a national administration will be inefficient, miss the purpose and will hinder the creation of the needed eco-system. Therefore, if it is decided not to create a designatory directorate within the Prime Minister’s Office, any further spending in this field should be seriously considered, as it might be futile” [13]. This demonstrates the strategic effort that the writers of the initiative believed was needed in order to create a synchronised eco-system that will help Israel advance more swiftly in AI-related fields. As we shall see in the next section it was this strategic effort and specifically the creation of a new AI directorate that attracted criticism and eventually torpedoed the initiative’s report.

The report itself is divided into three “layers”: (1) Critical Infrastructures; (2) Enabling Infrastructures; (3) Capacity building [13, 14]. Each of these layers is further divided into specific issues that should be dealt with separately:

  • Critical Infrastructures:

    • Computing Infrastructures—A local Israeli cloud; High-Performance Computing (HPC); Quantum computing.

    • Talent—Investing in Data Science and AI in academia; Expanding data literacy among k-12 students; AI education in schools; Attracting new faculty members by allowing them to have more control over IP (up to 80% of the rights and royalties, leaving 5% for their university department and 15% to the university); Creating new research centres.

    • Data infrastructures—Giving access to government-owned data; Establishing a national infrastructure for managing national data; Examining the potential of homomorphic encryption.

  • Enabling Infrastructures:

    • Ensuring cyber protection for all AI systems.

    • Forming a robust but minimal regulatory framework that would not restrain innovation.

  • Capacity Building:

    • Strengthening partnerships between universities and the private sector.

    • Implementing AI in the public sector.

    • Specific projects such as healthcare, agriculture, transportation, security and digital government.

The Initiative’s report is a comprehensive document, offering a detailed account of the different aspects that would need to be taken into consideration to develop the Israeli AI industry. For example, the report addresses related fields that can support the growing AI industry in Israel (IoT, robotics, data science, cloud computing etc.). It does so in order to assess how the Israeli private sector and academia are currently doing in these fields and highlight the required changes that would help both to prosper. It also pays a great deal of attention to the question of talent in Israel and focuses on the discrepancy between the success Israeli scholars have worldwide and the institutional difficulties they face in Israeli academia. Finally, it offers a detailed account of how specific sectors such as healthcare and agriculture will benefit from the AI revolution.

The report offered four specific recommendations:

  • Establishing AI as critical infrastructure for the future of Israel and considering it a national priority.

  • Formulating a national plan that will create a long-lasting ecosystem for AI.

  • Allocating 10 Billion NIS for R&D funding for 5 years (2021–2026).

  • Setting up a new AI Directorate in the Prime Minister’s Office.

2.3 Ethics and regulation

The Initiative's report included a detailed appendix on regulation and ethics, which was later expanded into an independent article [14]. The ethics and regulation appendix gives a comprehensive account of the principles that should guide decision-makers and developers in AI and encourages both the private and public sectors to embed these principles during all stages of AI development and implementation. The team that wrote this appendix also paid close attention to different kinds of regulatory mechanisms that might be useful to prevent harm but also to encourage innovation. They also mapped the different risks from AI and attempted to find the appropriate regulatory means to prevent them. The guiding principle that leads the main report is creating “the minimum ethical and regulatory framework necessary, in order to allow innovation and maintain Israel’s leadership” [13]. It also encourages developers to self-regulate where appropriate and to embed ethical principles in algorithms.

The ethical principles that are mentioned are:

  • Fairness

  • Accountability—which includes explainability and responsibility

  • Respect for human rights—which includes bodily integrity, privacy, protecting human autonomy and protecting civil and political rights.

  • Cyber and data protection

  • Safety

  • Maintaining a competitive market

The regulatory guidelines that are proposed [14]:

  • Aligning Israel’s regulation with International regulation and standardisation.

  • Mapping the actors to create adapted responsibility and incentive framework.

  • Adjusting the accountability mechanism to the dynamism of the AI area

  • Promoting normative clarity in critical stages of the AI value chain.

  • Constant review of the regulatory policy by the regulator.

  • Regulatory sandboxes.

  • The interface between the proposed principles and existing regulations.

  • Emphasise the role of the privacy authority in assessing the privacy implications of AI.

  • Emphasise the role of the competition authority.

  • Emphasise the need for interministerial coordination.

  • Authorities that are responsible for information resources should assess privacy concerns but also spaces for innovation.

The ethical framework relies on the EU ethical framework and the French National AI Strategy, although it did not endorse the vision such documents share [15, 16]. For example, the EU guidelines suggest that “AI systems need to be human-centric, resting on a commitment to their use in the service of humanity and the common good, with the goal of improving human welfare and freedom” (emphasis in original) [15]. Human centrism and the common good are the values towards which other ethical principles are oriented. The Israeli ethics appendix and the report itself, neither address this issue nor give citizens meaningful tools by which they make sure that their rights are kept during this process. For example, the writers note that trust is essential for the development of AI since the public will avoid using such tools if they do not trust them [13]. Yet the report does not offer appropriate mechanisms that will help citizens to build trust in such tools. The French NAS, on which the Israeli writers rely, suggests that because of the ethical challenges AI raises, “it would be prudent to create a genuinely diverse and inclusive social forum for discussion, to enable us to democratically determine which forms of AI are appropriate for our society” [16]. This logic, which exists in other NASs, and suggests that technology should be developed and implemented as part of a broader social dialogue between citizens, policymakers, developers and academics, is missing in the Initiative’s report.

To explain this view on ethics, we should remember that the main purpose of this initiative is to promote innovation and, consequently, fortify Israel’s resilience as a country. This vision is, therefore, essentially nation-centric which means that not much room is left for a human-centric approach that will allow a meaningful dialogue between citizens, developers and decision-makers. This logic was eloquently explained by Prof. Amnon Shashua, one of the main figures in the Israeli tech industry, who said in 2018 that he does not recommend strict regulation on data and privacy. When he addressed the need for more data in digital health he noted that “Israel is a small country, our survival depends on technological advancement, and to be honest, privacy is not that important” [17]. Such logic, it should be noted, does not lead the writers of the ethics and regulation appendix but it does exist in the main document. At one point the writers of the main document discuss the need for more data in healthcare and state that “privacy, regulatory and bureaucratic burdens regarding data sharing should be urgently removed” in order to encourage more cooperation between organisations, HMOs and research institutes [13].

Being that a discussion on ethics and regulation is essentially about priorities, it is interesting to follow the different perspectives that exist in the main report and its appendix. Following a clear trend in the Israeli discourse on science and technology, there is a clear tension between one view that sees AI as critical for Israel’s national resilience, and one that takes into consideration other civilian and social aspects. Eventually, a choice needs to be made regarding the role of citizens in the AI narrative that is being developed as part of broader strategic efforts. In the case of Israel, although there is a broad discussion on ethical and regulatory issues, citizens are not brought in and clear participatory mechanisms are not presented.

3 The second document—“Telem”

3.1 Background

Soon after the first part of the Initiative’s report had been published, several members of the Initiative’s steering committee objected to the conclusions and recommendations it made [18]. Coming from the Council for Higher Education, The Innovation Authority and the budget department at the Ministry of Finance, these members rejected the idea to create a new AI Directorate. As they put it, “The basic question—why government intervention of the proposed type is needed—has not yet been clarified and answered” [19]. Speaking anonymously to an Israeli newspaper, one opposer to the Initiative’s report called the creation of a new government directorate “a great error”. As they explained, “We already have an agency working on the matter, called The National Infrastructure Forum for Research and Development (TELEM), in which various government agencies work together […]. The agencies are perfectly able to work together; there is no need for a body that will only get into conflicts with the government agencies” [19].

This seems to be a pivotal moment in the story of Israel’s AI initiatives. The choice that was made by these government actors was to step back from a “national strategy” and move towards a “national program” that would be more modest, less expansive and without any major institutional reforms. The opposers, therefore, clearly defended the interests of the government bodies to which they belonged and preferred to avoid a comprehensive strategic plan that would include a new administrative body.

The Initiative’s report faced another problem. In June 2019 Prof. Ben-Israel and Prof. Matania, who led the Initiative, presented the report to Prime Minister Netanyahu. It was 2 months after the first 2019 elections, which did not allow either candidate to form a coalition government. As Prof. Ben-Israel would later recall, “We went to the prime minister […] he told us ‘wait another month, I’ll win the elections. Approving it now, without a budget—nothing will come out of it’” [20]. Netanyahu did not win the next round of elections, nor the one that followed it, and the report was not approved by the Israeli government.

To overcome the ongoing political deadlock and the opposition the Initiative's report faced from within, Telem decided in February 2020 to appoint a new committee that will “examine the need for government intervention in accelerating the development of AI and Data Sciences in Israel” [21]. The committee continued the work that was presented in the Initiative's report and was headed by Dr. Orna Berry. Instead of offering a comprehensive vision, as the Initiative did, it focused on four pillars: infrastructure, talent, regulation and access to data.

3.2 Vision, objectives, and recommendation

The premise for this report is the gap between Israel’s leading position in R&D and the private sector, and its relatively lower position in infrastructure and government readiness. The writers stress the sense of urgency regarding Israel’s position in the global AI race: “We were asked to bring this program to fruition because if such program is not implemented, Israel will find itself in the next few years facing a scientific, technological and economic gap that might become insurmountable, and a technological and scientific dependence that will hinder national resilience” [21]. The vision which the report sets forth is that Israel will become a leading power in AI and data science. Based on Israel’s existing capabilities and the challenges it faces, the report takes a more humble approach than its predecessor and suggests that Israel should lead “where it can”. Its specific objectives are as follows:

  • Creating the conditions for high-level scientific research in AI and data science.

  • Creating an eco-system that will enable the ongoing development and competitiveness of the Israeli industry, academia and public sector in AI and data science.

  • Maintaining this knowledge in Israel in order to allow a better understanding and development of critical tools for the defence sector.

  • Implementing AI tools in the public sector, thereby improving the service given to citizens.

  • Implementing AI in the industry and service sector.

The report identifies certain fields where government investment would be able to make a difference and put Israel in a better position. The first point the writers raise is the lack of appropriate computing infrastructure for R&D in academia, the private sector and the defence sector. Relying on the work that was already presented in the Initiative’s report, the Telem report raises the need for High-Performance Computing in Israel. Lacking such infrastructure, the writers argue, only deepens the gap between Israel and the world. They review different models that include government, public or regionally owned HPC centres that have been created in various countries, and conclude that such a centre is needed in Israel so that local academics will be able to promote local science.

The writers also suggested that a national cloud is a project worth pursuing since it is essential for developing certain aspects of AI and data science. As the writers note, there are already some ongoing cloud projects in Israel such as “Nimbus” a government program for acquiring a public cloud for government offices, “Timna” a public cloud that will gather medical data and “Psifas” that will contain genomic and clinical data. However, there is no program for a public cloud in Israel that, according to the writers, forces those who wish to use cloud services to transfer their data abroad.

The writers also argue that Israel should do more in making data available and allow more access, subject to privacy laws and data protection regulations. Israel holds a valuable trove of government data that relies on an ID number each citizen has. This means that when used properly, such data can become highly valuable for different government services. As the writers suggest, this is a national resource that is currently not ready for use but can be utilised for both R&D and government services.

Finally, the Telem report highlights a growing problem regarding human capital. As they suggest, Israel has clear advantages in AI research such as high-quality training, a relatively high number of leading scholars, a growing industry, and relevant training in the defence sector that offers a good starting point for those who later seek a university degree [21]. However, there are also clear challenges such as a small number of faculty members in these fields (less than 70), a spillover from academia to the private sector, and a global competition that makes it hard to maintain a leading position. To promote human capital in Israel, the writers recommend adding more tenured positions for scholars and allowing them to work more closely with the private sector. This will help train more students and perhaps also attract talent from abroad.

Based on this analysis, the Telem report recommended establishing a “national program for AI and data science in Israel”. The duration of the program is limited to 5 years and it will support academia, the defence sector and the private sector. The program is estimated at 5.265 billion NIS. Specific objectives include:

  • Investment in human capital. This can be done through different means such as: expanding the number of graduate students; expanding data science training to all university students; university and private sector partnerships; partnerships with the public sector; access to public sector data; recruiting more faculty members and university researchers; and giving more funding for infrastructure in universities.

  • Research and development. Sharing information and developing new tools based on partnerships between universities and the private sector.

  • Advanced manufacturing. Developing AI-based advanced manufacturing capabilities.

  • Management. Setting up a new managing body, civilian and military, that will synchronise the development and implementation of AI.

  • Supercomputing. This will be used for various purposes: training (academia and the private sector), inference (academia, private sector, public sector) and optimization.

  • Public Cloud. Purchasing cloud services from one of the companies that will set up cloud infrastructures in Israel during the coming years.

  • International Cooperation. This will allow expanding the current research budgets in academia.

Regarding the body that will manage the program, the report says that “a binding discussion has not yet taken place with the other partners in Telem” and it will be expanded in the future [21]. In other words, the committee avoided creating one unified and coherent body that will administer the implementation of the program and left it for a future discussion. Following Telem’s report, the newly elected government, led by Prime Minister Naftali Bennett, presented in August 2021 its “Plan for Promoting Innovation, Encouraging Growth in the High-Tech Sector and Strengthening Technological and Scientific Leadership”, also known as, “Resolution 212” [22]. As part of this government resolution, it was also decided to approve and begin implementing the first stage of the Telem report which included: developing human capital; creating a supercomputing centre for the public sector, industry and academia; promoting NLP capabilities in Hebrew and Arabic; and creating an enabling regulatory environment. The budget allocated for this stage is 550 million NIS. The resolution divides the responsibility between The Science, Technology and Innovation Ministry, Ministry of Defence, Ministry of Finance and The Innovation Authority, each assigned a different role. The resolution also makes clear that any decision that is made will not hurt the authority and independence of the Budgeting and Planning Committee of the Council for Higher Education or the Ministry of Defence.

3.3 Ethics and regulation

The Telem report offers an extensive review of various kinds of regulatory and ethical frameworks by the OECD, European Union, The United States, The United Kingdom, France and South Korea. The purpose of such a review, the authors suggest, is to learn from other countries and multi-national organisations how to handle the risks which AI and data science entail. As part of this review, the report also examines specific ethical issues that are related to data and the ways to solve them, like the British proposal to use data trusts or the French idea of considering data as a common good. It also reviewed specific regulatory frameworks such as the European GDPR and the Californian CCPA.

Unlike the Initiative’s report, the ethical and regulatory principles which the Telem report recommended are not written specifically for developers and decision-makers. As a result, both kinds of principles remain vague and do not offer tools that developers can use to assess potential harm. The ethical principles which the report outlines are as follows:

  • Transparency.

  • Explainability.

  • Privacy and Data Protection.

  • Cyber security.

As seen, the list of ethical principles chosen by this report is somewhat shorter than the one that the Initiative’s report offered. Also, explaining these principles in the Israeli context remains vague. For example, Israeli legislation lacks meaningful tools that give data subjects the “Right to Explanation”, for example. This means that mechanisms through which transparency and explainability can be maintained remain unclear in this report.

The guiding principle for regulation is similar to the one made by the Initiative’s report and suggests that an “enabling regulation” would suit Israel’s need for innovation. The regulatory principles are as follows:

  • Enabling regulation that will prevent “technological stagnation” and will include regulatory sandboxes.

  • Standardising AI so that ethical principles will be embedded during the development stage.

  • Purchase and sale policy, specifically regarding the defence sector.

  • Creating data centres that will make data accessible and enable more effective data sharing.

  • Managing data, cyber security and data protection that will allow anonymising personal data.

The report also recommends creating a sole regulator that will “unify the ownership over data” and consider data as a “public good” that can be used to benefit the public. As the writers suggest, “the regulator will create, enhance and share access to data through an enabling regulation” [21]. This point corresponds, according to the authors, to the French NAS which considers data to be a common good. Like the discussion in the Initiative’s report, the Telem report neglects to emphasise that the French approach is rooted in a broader regulatory framework, the GDPR, that gives data subjects clear rights. For example, when the French NAS addresses the idea of “data as a common good” it suggests that data of both public and private institutes should be made more accessible. It also argues that “Recent laws on individual’s rights to data portability could therefore be part of a broader citizen-based rationale, to enable the State and local authorities to recover data with the aim of developing AI-based uses for public policy purposes” [16]. As the French NAS also suggests, common goods are resources whose use and governance are decided by the community. This is an essential point that does not exist in the Israeli discussion on both AI and data.

It seems that according to the Telem report, turning data into a common good means giving more access to it, instead of considering the rights citizens have regarding such data. Like the Initiative’s report before, the idea of innovation that guides the Telem report also means that social concerns are downgraded, and meaningful involvement of citizens is missing. It also means that the decision on how to use personal data, even when held by the government, is not made alongside the active participation of citizens. We should remember that the need to balance privacy concerns and public benefits regarding data is not a new phenomenon. The need to keep an open dialogue between decision-makers and the public and the risks from not doing so have both been made clear in the late 1960s [23]. Such decisions cannot be made only by tech experts and decision-makers but need to be done through an open dialogue with the public. This is the point that is made in the French report when it explains what is the meaning of data as a common good is, and the one that is missing in Israeli discourse regarding this issue.

4 Discussion

Officially, Israel still does not have a coherent national AI strategy, but only a “national program”. This might seem like a semantic difference, but as I tried to show it bears a meaningful significance that needs to be explained. As I already claimed, the ethos of innovation is deeply rooted in Israeli culture and it guides both reports. This ethos is based on Israel’s ability to remain agile and quickly address unexpected circumstances in order to sustain itself. It is also the ethos that is usually used to explain the rapid growth of the Israeli high-tech industry. This ethos is related to a specific concept of time that has shaped Israel’s policymaking since its creation and limited its ability to think strategically.

In one of the sessions during the Israeli 2022 AI Week, Prof. Ben-Israel and Dr. Berry, the primary writers of Israel’s two AI documents, met to discuss the Israeli national AI program. They addressed the fact that, unlike many other countries, Israel has a “national program” instead of a “national strategy”. Both considered the lack of a coherent strategy as an advantage since it allows different elements within the public sector, industry and academia to compete and get better results. “It can be awful,” Dr. Berry noted, “but it can be utilised” [24]. Prof. Ben-Israel gave a more detailed explanation:

Israel was born during a war, which is still not over, but we don’t have a national defence strategy. We did very well without a strategy. This is part of, I think, Jewish culture - not even Israeli. We have the written Torah and we still live by what we call the “Torah by heart”, which is an interpretation of what, is believed, was dictated by God himself. This is very unique to Jewish culture and Israel used it. On the one hand, it’s more chaotic, the system is not as ordered as it is in other countries. On the other hand, it makes us more agile. In the case of emerging technologies, I see this as an advantage. Not as a disadvantage.

Adhering to the well-known tech maxim “it’s not a bug, it’s a feature”, the lack of a coherent national strategy is not coincidental but something that can be explained. According to these two leading figures, a lack of strategy allows agility, and agility can enable innovation. As I noted in the introduction to this paper, this view is deeply rooted in Israeli culture and is based on the notion that because anything can happen in the future, Israel needs to be innovative and agile in order to survive. But whereas it might work in some aspects of Israel’s national culture, and some would argue that it does not, it seems that when this idea is brought into AI and data science, things become a bit more complicated.

Dr. Berry and Prof. Ben-Israel are right to claim that this kind of ethos surely did contribute to Israel’s leading position in the High-Tech industry. Yet, it seems that in the case of AI and the data that fuels it, formulating a national strategy is more pressing. Having a clear regulatory framework that oversees and monitors how AI technology is developed and implemented, especially in the public sector, can be established only as part of a clear strategy that sets priorities and explains what is exactly at stake. It is also essential when the potential harm is greater than it is in other technologies, and where responsibility and liability become more important. Stuart Russel has recently noted in his 2021 Reith Lecture, that an AI race is not necessarily a good thing. As he put it, “If you’re in a race, then safety is the last thing on your mind. You want to get there so you don’t worry about making it safe” [25]. AI, as the Initiative’s report itself explains, might enhance existing inequalities or breach certain kinds of individual, civic and political rights. A national AI strategy is not a magic panacea to all of society’s ills, but it can create clear and unified rules for those who are already racing and help those who are lagging behind.

A lack of a coherent strategy also reflects another issue that should be taken into account. A national AI strategy can establish forums for public discussion, during which a shared national narrative on technology and innovation can develop. Such forums allow citizens to articulate expectations and concerns, that will then be embedded in a national AI strategy. A clear strategic effort can bring together experts, decision-makers and members of the public to envision together how society might benefit from the implementation of AI in society [26, 27]. This process is not without risks and it is not always successful [28], but it is necessary as a means to build trust between citizens, governments and the industry. It might be that a National AI Strategy is not essential for innovation, and governments can probably find ways to work around it and still maintain a leading position. But as I argue, a national strategy can empower society and facilitate an open dialogue between all relevant stakeholders in order to create trustworthy AI.

The social sciences, it should sometimes be stressed, are not about predicting but only about explaining reality. This paper aimed to explain the route Israel chose, but it might be that Israel will decide to adopt a fully-formed national AI strategy. The question still remains as to how such a strategy will see the role of citizens within it. The ethical frameworks in both documents give an elaborate account of the appropriate principles that should be utilised, and it would be futile to doubt the good and honest intentions of the writers. The concerns raised in both documents about a potentially stifling regulation make sense and should also be taken into account. But I would argue that while it continues to implement its national program, the Israeli government should also broaden its perspective and include more stakeholders in this process. As Eyal Chowers claimed, although Israeli politics is essentially republican it is hardly Aristotelian in that it prefers deeds over words [5]. In the case I presented here, such a view might negatively shape future strategic attempts and it should be taken into account. Policymakers should initiate a public discussion on AI, the ethical principles that should guide it and the appropriate regulation that will ensure both innovation and effective monitoring. Citizens should be kept as active participants in this process in a way that will also encourage them to share data in a trustworthy environment. This, I believe, can only happen as part of a more comprehensive strategic effort that will turn the implementation of AI into a truly national project.

5 Conclusions

The Israeli case seems to be unique in the literature on national AI strategies as it shows how specific national visions not only shape NASs but also make some countries choose a different course. As I argued, this decision continues a deeply rooted ethos in Israeli culture that is believed to have contributed to Israel’s leading position in science and technology. As I also showed, this ethos might contribute to innovation but it also bears important implications regarding ethical and regulatory issues. It might be that a “national program” might allow more agility and innovation as opposed to a coherent national strategy. But as I explained, such a choice might not provide a sufficient ethical and regulatory framework that gives citizens the appropriate means to oversee how technology is used.

The Israeli case I followed in this paper shows how a specific shared view about political reality might shape the formation of a National AI Strategy, of the lack of one. As I also showed, some public institutions were able to torpedo the strategic attempts made by the Initiative’s report and led to a more humble national program where responsibility was divided between more actors that maintained their authority. This might suggest that the research of NASs should also follow how different actors (public and private) shape NASs and which kinds of institutional interests are embedded within them. This might also stress more the political aspect of National AI Strategies, and open a new area for discussion in this emerging research field.