Photo by — @soheili

If I Should Write an Introduction to Artificial Intelligence Policy…

How would it be and would it be alright to write it?

Over the course of the year I have come to learn several interesting things about the field of artificial intelligence (AI). One of those is that AI is far more than the technical aspect of researching or developing AI. Even when it comes to the application of AI there is a lot to learn beyond the direct interaction with a user or citizen.

This short piece is, as you might understand from the title, not anything definite.

How could one introduce AI policy?

If I was to write an introduction to AI policy where would I start?

I was talking to a consultant who has been working with the pragmatic workings of policy most of his life, and he had an idea.

We discussed the descriptive and prescriptive parts of a strategy, and that in a pragmatic sense it was questionable how a strategy was ‘used’. In the operational sense if it is policy or an AI strategy it could be well-written, yet the question remains how — in the operational or governmental — this strategy is to be implemented.

I have seen institutional promises of either funding numbered in PhDs (German national AI strategy) or particular projects such as a regulatory sandbox (Norwegian national AI strategy), so these aspects do play out to some extent. Norway recently announced their regulatory sandbox, and it would be possible to measure, at least to some extent, whether the German state has fulfilled their promise of funding another 100 PhDs.

Anyhow, the consultant thought I should see how many times each strategy had been cited. In this sense it would be possible to get an overview of, at least publicly, how the strategies had been discussed. If one did this one could examine other aspects in relation to the amount of public discussion surrounding an AI strategy. Still, this might be moving more into the details of a particular case, as national AI strategies are only part of the overall picture when it comes to AI policy.

If I was to introduce AI policy there is the chance that I might have to go into a description of AI and then policy.

However, I define myself somewhat as a social anthropologist and not necessarily a political scientist. I could pull on a Wikipedia definition:

“A policy is a deliberate system of principles to guide decisions and achieve rational outcomes.”

This would not be very convincing. Not that it would have to be.

Perhaps, it would be fine to begin with honesty?

One concerned citizen attempting to navigate in the world and developments within artificial intelligence. Seeing what is discussed and thinking about what the overall discussion is lacking.

I could have said diversity, but that has become painfully clear this last year with the Black Lives Matter movement and the ‘pause button’ on facial recognition by several large technology companies.

Then again this is an issue that remains and is not to any great extent close to being discussed enough in policy. At least not in most contexts, the economy, jobs and industry takes the front seat. I would hate to be a ‘back seat driver’, the person in the back of the car asking the driver to slow down, and yet it seems somewhat responsible.

An introduction to a particular kind of AI policy or a general kind of AI policy?

I wrote a sketch late at night. It was also on the topic of ‘how to write an introduction’.

“AI AI AI What Happens Next?
An introduction to artificial intelligence policy in 2020

How could we make an introduction to AI policy?

Although this book is made with good intentions you may find it lacking. If so, there is plenty of room for amendments. This is an open-source book project as an introduction to help you gain a rudimentary understanding of international AI policy. By picking up this book or accessing it online you will have to come to terms with the core purpose of the project: this is meant as a learning process and as an iterative undertaking to learn. This writing was done to build conceptual frameworks, knowledge and to begin developing competencies towards this field. There are likely more capable minds, however no such book was published. Part of the reasoning behind this project is not only to talk of the policy that is written or performed, rather to consider what is missing in AI policy that could need more careful attention.

This book was written as a personal project begun by a naive Norwegian writer. In this book there is a prior acknowledgment that the climate crisis has reached large scientific consensus and needs to be addressed. The field of artificial intelligence and those who hold interest in policy will be held to account if the overall effort is detrimental to this process. If one believes that artificial intelligence can work for our shared ecology this remains largely to be proven in the decades to come. This project is undertaken with critical perspective alongside the encouragement and wish that AI can be a contribution to reaching our shared goal of increased sustainability.

-Alex Moltzau

0 /// A note on intelligence and lack of understanding

(mapping out what i currently discussed, yet additionally what is missing)

(urgency on important issues, biodiversity, sustainability etc.)

(racial inequalities and facial recognition technology in policing — techlash)

1 /// Perspectives on International AI Policy

The bipolar perspective on AI policy

(political realism?) (the policy shift in documents from the Obama administration vs. Trump)

(Kai-Fu Lee) (AI Super-Powers)

Beyond a bipolar perspective of AI policy

(There is much to be seen in between the colours and flags)

(international resources and digital infrastructure)

The multifaceted European strategy and national AI strategies

Multilateral collaborations on AI

Large technology companies and their political leaders

(market-entry, practices by technology companies in infrastructure)

(The discussion between Elon Musk and Jack Ma) (Jack Ma in the United Nations)

Where are decisions made and who makes these decisions?

(human-in-the-loop and coded gaze)

2 /// General concerns in AI Policy

International security policy

(AI Arms race)

(Campaign against killer robots)


(Factories, mining, oil, robotics)

Economic policy

(the outlined increase and its proponents — consulting)

Labour concerns

(x) (?) (NAV)

Environmental and resource constraints

(devices, recycling etc.)

(forests ++)

Health and wellbeing

(Tradeoffs and need for innovative education)

(Healthcare professionals + critical thinking abilities related to technology)


(experiments by Salesforce)

Energy Policy

3 // National Priorities in Strategies


The digital infrastructure

AI Policy in operational systems

OpenAI and reinforcement learning.

Facebook and international governance

Google Scholar and research policies

YouTube’s policies


Governance is not decoupled.

Nick Clegg.

Danish diplomat hiring.”

As such, in my rough notes there is a lot to introduce.

It becomes a mix of technology policy in general that connects with the topic of artificial intelligence, because it cannot be disembedded from the overall technology industry when it comes to particular priorities within society.

Writing an introduction may be for the experts. I thought to myself that since I am new to this field that it would be great to learn from summarising the surface that I have scratched by writing about (or related to) AI for more than a year, steadfast every single day.

If I Should Write an Introduction to Artificial Intelligence Policy…

It would actually be alright to write it.

Write it intended as a sharing experience.

Yes, I think I would want it to be an open project intended for learning.

This is #500daysofAI and you are reading article 386. I am writing one new article about or related to artificial intelligence every day for 500 days.

AI Policy and Ethics at Student at University of Copenhagen MSc in Social Data Science. All views are my own.