Four Possible Approaches for the New Norwegian National Billion NOK Investment in AI Research
The AI Advisory Expert Group in Norway delivers its second note on structuring the AI billion
Further down in this article you will find a translation from Norwegian to English of the second note to the government by the AI Advisory Expert Group appointed by the Norwegian Research Council in 2024 to advise on how to structure the one billion NOK in funding for AI research in Norway.
I also previously translated the first note from the expert group into English and it can be found here:
The overall process to shape the Norwegian AI Billion has been fairly inclusive:
- 200 participants who attended the input meetings on 6th, 7th and 8th November
- Nearly 120 actors have submitted written input.
- Recordings of the prepared presentations at the input meetings can be found on their website.
Although there were concerns raised whether the composition of the expert group could have been even more considerate of diversity and existing talent in the Norwegian AI community. I raised this previously in an article:
Underneath, following the title of the ‘overall measures’, you can find the translation of the second note by the AI Advisory Expert Group in Norway. Like last time I may make some formatting choices to highlight text with bold, cursive or ‘quotes’. I also added ‘for the investment in AI research’ in the titles. I also added a numbering to the titles.
1. Overall measures for the investment in AI research
Expectations
The national investment in artificial intelligence has three integrated tracks:
- Consequences of artificial intelligence and digital technology
- Digital technologies as a research area in itself
- How digital technologies can be used for innovation, including in research
In our first note we described the current situation; where there is the greatest need for research and research-based competence at KI, where Norway has special prerequisites for helping to move the research front — and what expectations we should have for the AI initiative.
We described a rather ambitious footprint where an ambition for the AI initiative should be that it is:
- Long-term: The effect of the billion goes far beyond the next 5 years and the billion.
- Fruitful: The effect of the billion gives benefits outside of research as well.
- Groundbreaking: The AI billion produces important, ambitious and influential research that
places Norway on the map of AI research-intensive countries. - Interdisciplinary: Large parts of the research are interdisciplinary.
In this memo, we go into more detail about some general measures to ensure that these expectations are met. We will first outline some key considerations that must be taken care of, before we discuss some possible ones main approaches that meet this in different ways.
2. Key considerations for the investment in AI research
The AI investment in research must be seen in the context of other initiatives. Norway has a national AI strategy that provides a national framework for AI both for business and the public sector.
This emphasizes the need for development of infrastructure, regulations, expertise and research. On the research area also includes the AI billion in a larger context where it is added to other grants for AI research. The Research Council already has a significant AI portfolio today corresponding to NOK 700 million per year (in 2023).
The AI billion will be distributed over five years and will thus increase the effort to around NOK 900 million per year given that other funding remains at the current level. In the following it is the special research effort on AI, the AI billion, which will be in focus.
In order for the AI billion to have a significant effect in line with the ambitions mentioned above, special attention should be paid
emphasis on the following considerations in the design of overall measures:
Clear commitment: Today’s AI portfolio in the Research Council is spread across a wide range of areas and theme. In order for the AI billion to have a great effect, it should be given a clear and more concentrated arrangement there resources are coordinated.
What is financed through the investment should therefore be included in a larger one connection where the various parts are in contact with each other, triggering new synergies and contributing to common goal.
Collaboration: The initiative’s three tracks cross subjects, sectors and national borders. Succeeding in the venture will require significant cooperation. It should therefore be designed in such a way that it facilitates interdisciplinary work cooperation, for international cooperation, and for cross-sectoral cooperation. This includes collaboration with stakeholders outside academia and links to other measures/initiatives.
Connection between the tracks: The main structure of the investment is the three tracks: consequences, technology and innovation. These are closely related and must also be studied in conjunction. It is important that the investment is designed in such a way that research is planned within each of the tracks, and research where the tracks are seen in context.
The investment must grow: The government has set aside one billion for the investment. As explained in the first note the need is significantly greater. The investment must be arranged so that it can also activate larger investments from other funding sources, such as the EU (Horizon Europe). It should be designed so that it can be easily strengthened by larger allocations from the state budget, and that the facility motivates funding and in-kind contributions from business and the public sector.
Both rapid and long-term: Technology development is going fast and AI is being used in an increasing number of areas. It is great expectations for the investment from business and the public sector, which need research results, and from the public who want more knowledge. In the research environment, there are expectations for the future announcements. It is therefore important that the investment is designed so that it can get started relatively quickly and that it continuously produces results that can be put to use. At the same time, ground-breaking knowledge development is required investment over time, and provision must be made for long-term research efforts.
In addition to this, the investment must follow good principles for responsible research and innovation (responsible research and innovation, RRI). Artificial intelligence offers many possibilities, but it is also risks linked both to technology development and to how the technology is used.
3. Possible approaches for the investment in AI research
Which approaches can realize the venture’s ambitions and take care of the considerations described above?
In the following, we outline four main approaches to using the AI billion.
These describe how different approaches can meet the ambitions of the venture and take care of the considerations in different ways. In a design of the investment, the approaches can be further developed, combined, and/or act as inspiration.
Approach 1: One nationally distributed centre
One possible approach would be for the billion to be used to establish a national center for research on artificial intelligence. Such a center can provide a clear investment with a large national and international presence
visibility, and will enable interdisciplinary activity that sees the three tracks in context. It can be geographically distributed with research activity all over Norway, but with joint professional management.
The design of such a center could be modelled after the Alan Turing Institute in Great Britain, or Pioneer Center for AI in Denmark. Central elements in such a center may be one director for each track, and one
collegial management group consisting of representatives from AI-intensive research institutions.
The management group should represent the breadth of professional approaches and have representatives from the technology subjects, social sciences, humanities and law. The directors and management team defines and updates interdisciplinary research topics and designs calls for proposals (larger projects, centres, international collaborative projects, etc).
The calls are announced through the Research Council who then assess the applications through ordinary panels. The center does not have its own staff, all of them employment is at participating research institutions. Up to 10% of the allocated billion will be used for the center construction, while the rest will be announced as projects under the centre. There will be one aim that other sources should also invest in research through the centre.
The advantages of such a national center are, among other things, that it will provide a clearly concentrated investment great national and international visibility. The center can build systematic collaboration across disciplines, actors and sectors and network and meeting place activities at national level. It will be able to promote short distance from the research front to setting up the center and announcements.
Such a combined effort will be able to be attractive to top international researchers. It will be able to ensure that the three tracks are kept together and seen in relation to each other. And it will be able to facilitate the development of strong research groups with ground-breaking research at many Norwegian institutions. Such a center will able to create and run arenas where different research environments, private and public actors meet, and for interdisciplinary and cross-sectoral activities around the relevant AI themes (within all three tracks). If the center is successful, it will be able to attract long-term funds and thus be viable even after the first 5 years.
One of the disadvantages of such a centre, which has not previously been established in Norway, is that it will take time to get one established and it will therefore take some time before good research activity is achieved in the centre. First must the center is established and then projects must be defined for tendering. To avoid too long before start-up of a center can be combined with some project announcements in advance of the establishment of the centre. Another challenge at such a center will be to ensure good cooperation between a center that formulates calls for tenders and arrangements for the billion, and the Research Council that evaluates
applications and allocates funds. This will require a clear assignment of responsibility.
By establishing a national centre, there is a further risk that a strong professional management may cause someone of the tracks are prioritized over the others, or that some academic facilities fall through. This can particularly be a problem within track 1 of the investment (consequences of the technology) where a broad disciplinary approach is needed. Countering this will require good professional grounding, clear management, and statutes that ensure breadth of investment. Channeling all the funds through one center also means that the risk increases
if the center is not successful. The center should therefore be evaluated after three years for continuation.
Summary: An approach with one national center will enable a clear investment with opportunity for good visibility both nationally and internationally. It will provide good opportunities for multidisciplinarity, to see the investment’s tracks in context, and concentration of efforts will give great results opportunities for ground-breaking results. At the same time, there are risks associated with the design of the center and about the center succeeds. Establishing a center will take time and it will take time to get research activity started in the centre.
Approach 2: Multiple centres
Another approach is a solution with several parallel centers. This enables several larger investments with various thematic effects. The Research Council’s centres for environmentally friendly energy (FME) are an example of one such approach and can serve as a model. In the last FME call, up to 1.2 billion NOK was announced for the establishment of between six and ten centres with support of NOK 120 to 200 million per centre. The call for proposals set out requirements for thematic priorities and how the centres should be set up, including requirements about collaboration partners.
Such an approach in the AI field can, for example, be combined with requirements that all the centers that are established must be interdisciplinary and that each center must cover a minimum of 2 out of 3 tracks. Together, the centres will be able to cover all three centers and provide important thematic inputs, for for example in health, public administration, green transition etc.
The advantages of such an approach are that it can ensure interdisciplinary research within and across of all tracks in the venture. It is based on a well-known model, which will be easy to manage by the Research Council, and relatively easy to administer for the institutions. It provides a clear arrangement, at the same time that the risk is reduced by the fact that there is not one national center as in approach one, but several centres that are independent of each other. The multi-centre solution also lays the foundation for several businesses can have ownership of the initiative through center participation and management.
The disadvantages of a multi-centre solution include that it is less concentrated than a one-centre solution. This can give less strategic power nationally and internationally than one large centre. Furthermore, establishing centres will delay start-up and should possibly be combined with other announcements prior.
Summary: A multi-centre solution à la the FMEs will provide the opportunity for a clearer arrangement than one simple portfolio model (see approach 3, below), but management will be more distributed than a one-centre solution. It can facilitate interdisciplinary and inter-institutional cooperation on a larger scale ventures, and involve partners with the potential for the venture to grow, beyond them originally funded the centres. Establishing even relatively simple centres will take time.
Approach 3: Portfolio of different types of projects and initiatives
Another possible approach is a mixture of projects of different sizes and facilities, research schools, centres, and more. Research can be facilitated through various types of targeted calls for proposals within the initiative’s three tracks, interdisciplinary collaboration, and research that integrates perspectives from more than one track. The entirety of the investment can be safeguarded by the billion being managed by one portfolio board in the
the Research Council and that a portfolio plan is drawn up that promotes the entirety of the investment. The portfolio board can use expert committees in the design of calls for tenders, among other things, to ensure that they are broad enough professional competence to look after all three tracks.
The advantage of such a model is that it is well-proven. It is flexible and can include different types means. It will be able to be seen in connection with and help to strengthen the Research Council’s existing ventures. And, this is an approach where you can announce funding calls quickly and get started with the investment.
A main objection to such a model is that it does not provide an overall footprint of the investment. It will be distributed and can provide less national coordination and a less concentrated effort than that which can be achieved through the other approaches. Challenges with this model will be securing cooperation and dialogue between different types of projects financed through the investment. In order for such an approach is not to be split up, it will be important that it is managed collectively by one portfolio board. Furthermore, it will be important that there is a requirement that a part of the portfolio is made up of centers or projects which includes more than one track. Arenas can also be established across the projects in the portfolio to promote joint efforts and understandings.
One risk is that there may be a large gap between responsible portfolio management and the research front. This can be extra challenging because the academic breadth with three tracks is demanding to cover. It will then be important to ensure good advice on setting up calls for tenders and developing the portfolio. Use of expert committees or other advisory forums will be able to supplement the portfolio boards.
Summary: in light of the criteria, such an approach will be the one that will get started the fastest. It facilitates collaboration and growth opportunities in the same way as the Research Council’s existing portfolio. The model is less coordinated than the others the approaches. This can possibly be compensated with the establishment of transversal arenas within the venture. The main objection to this approach is the lack of clear focused
footprint. Here the other approaches are more clear.
Approach 4: Competence hub
A fourth approach is a competence hub, or a virtual centre. This can be inspired by the Center for digital life where a competence hub/virtual center structure has been established to facilitate cooperation and coordination between research projects. The research projects are then announced and awarded by the Research Council and can be a mixture of larger projects that cut across at least two of them three areas, and projects of a more traditional size that can have a sharper focus.
Inspired by the Center for digital life, a possible competence hub can be announced through a separate call for proposals with the aim of including all key environments within AI research in Norway. Up to 10% of the funds can go to the hub, which will have the following key tasks: Visibility of the entirety of AI research in Norway (lighthouse effect), dialogue with the Research Council on the alignment of calls for proposals, research schools and joint PhD networks, guidelines for RRI, design of data handling plans, etc. It must be set aside sufficiently funds for and demands are made that associated projects spend time and resources towards the central hub.
An important advantage of such an approach is, among other things, that it combines a clear commitment throughout a common hub with bottom-up approaches through the projects that will be assigned to the centre. Through the hub, it will be possible to promote collaboration between projects, subjects and sectors and ensure close connection between the investment’s tracks. The hub will also be able to promote international visibility and add facilitate cooperation that generates investment from other parties. If the hub is successful will it should also be easy to extend as it takes a limited part of the investment’s funds.
A disadvantage of such a hub is that it also takes time to establish. This can be counteracted by starts with the first project announcements before the center is established. In any case, experience from establishment of the Center for digital life shows that it takes time to make the hub work well and create added value across the projects. It will be important to work well with the development of the hub so that projects and other centers see it as attractive to participate and contribute to common goals. It will not have the same signal value as a center built around top researchers (ala Kavlisenteret).
There is a risk that several environments will compete to become a hub without wanting to cooperate. By the establishment of the Center for digital life, the Research Council had to take action to bring about cooperation. It is also it is important to recruit good coordinators.
Summary: A competence hub/virtual center will be able to give a national footprint, make it visible the venture nationally and internationally, and facilitate significant collaboration across the venture projects. It will be possible to get started quickly with project establishment, but it will take time establish the hub itself. There is a risk associated with whether the hub becomes attractive enough for it to attract enough activity and participants.
4. Comparison of the four approaches
The various approaches differ along several dimensions. The bigger a centre/project is, the more it can do interdisciplinary work and research are promoted across the tracks within the same research environment. Correspondingly, yes more organisationally concentrated, the investment is designed in the form of one center (national or hub), the more will coordination could take place within the framework of the centre, while approaches 2 and 3 would be able to benefit collaboration and arenas for sharing knowledge and expertise across the projects/centres. The approaches will also be able to be combined. This applies in particular to approaches 2, 3 and 4 where you can various possible combinations.
All approaches will be able to meet the venture’s ambitions, but to varying degrees. The central considerations we recommends should be used as a basis, but will be handled in a slightly different way:
Clear commitment: The more concentrated the commitment, the clearer the arrangement will be. It the most obvious arrangement will be the one that gathers the entire investment in one centre. It will give the bet strong visibility nationally and internationally and provide shared access to updated research results. A portfolio of different types of projects and centres could have the advantage of a larger catchment area, but
thus also running the risk of a less clear footprint.
Collaboration: All approaches can promote collaboration through requirements in the call for proposals that emphasise for example interdisciplinary work, cross-sectoral and international collaboration within projects/centres. However, projects/centres of a certain size will make it easier to realize such collaborations within the projects/centres than smaller projects. National cooperation is the basis for them the national centers (approaches 1 and 4).
Connection between the tracks: Correspondingly, all the approaches will be able to facilitate that the tracks seen in context through research projects and centres, but this will be easiest to promote within the same project/centre when these are of a certain size.
The investment must grow: There is no definite answer, as to what will best facilitate growth beyond research and beyond 5-year period. Centers of any size (options 1, 2 and 4) will have a good starting point for establishing long-term collaborations. At the same time, we know that smaller projects with requirements for cooperation and co-financing have a good multiplier effect.
Both fast and long-term: The larger the structures to be established, the longer it will take to get in time with the investment. At the same time, the larger projects/centres will better facilitate the long-term research and development.
5. Recommendations
In our first note, we recommended that the AI investment should have as an ambition to be long-term, fruitful, ground-breaking and interdisciplinary.
In this note, we point out that the following considerations must be taken into account in order to realize this ambition: The stake must be clear; it must facilitate cooperation between subjects, sectors and above national borders; the tracks in the investment must be seen in context; the venture must be designed so that it can grow; and it must reconcile the need to get started quickly with long-term knowledge development.
We have outlined four possible approaches that can meet this in different ways:
- One nationally distributed center;
- Several centers;
- Portfolio of various types of projects and measures; and
- Competence hub.
We have shown how all approaches have clear strengths, and that all have some disadvantages that should be compensated for if they are to be realised. The approaches are not designed in detail and will be able to be further developed, combined, and/or serve as inspiration for further work.
In the expert group, there are different views on what would be the best approach to take care of the concerns we have outlined.
Part of the expert group believes that a nationally distributed center would be the preferred one the approach. These emphasise that the AI billion provides a unique opportunity to create one concentrated investment with a clear footprint where Norway is placed on the map as an AI nation which emphasizes Nordic values in our AI investment: trust, democracy, legal certainty and welfare. One like that central investment will facilitate collaboration both between the tracks and be easy to collaborate with
with other actors nationally and internationally.
Others in the expert group prefer a combination of several approaches. These assess the risk in the case of a bet on one center as hay. Both the prerequisites for national cooperation between all relevant research environments and the ambitions for such a large academic breadth in one centre, have proven demanding in other contexts. International research centers with AI as a theme either have a technological focus with some social science research and little or no humanities or legal research, or a humanities or social science focus with little or none technological research. This part of the expert group does not point to one particular approach as one preferred alternative, but that the investment can integrate elements from several of the approaches such as overall makes a strong investment. To ensure that the investment has a clear footprint and to promote cooperation between the various parts, it will be important to have comprehensive, transversal measures or arenas.
The entire group is united in the fact that regardless of which approach or approaches are chosen, it is important that the ambitions for the investment are high and that the considerations we have outlined in this note are taken care of.
Submitted to the Research Council of Norway on 20 December 2023
The members of the expert group: Tanja Storsul (leader), Arnoldo Frigessi, Cathrine Pihl Lyngstad, Eirik
Andreassen, Ieva Martinkenaite, Inga Strümke, Jill W. Rettberg, John Krogstie, Klas Pettersen, Olav
Lysne, Petter Bae Brandtzæg, Signe Riemer-Sørensen and Tobias Mahler.
I hope this translation was helpful, leave me a comment here and tell me what you think Norway should go for or reach out to me on LinkedIn.
This is also part of my personal project #1000daysofAI and you are reading article 525. I am writing one new article about or related to artificial intelligence for 1000 days. The first 500 days I wrote an article every day, and now from 500 to 1000 I write at a different pace.
