An initiative that I think is great and that must be commended by the Institute for Human-Centered Artificial Intelligence at Stanford University is the sharing of videos spanning the entirety of their fall conference on AI Ethics, Policy and Governance. The two long videos, each one lasting a few hours, is an important contribution for anyone interested in this area or the ongoing discussions at one of the leading institutions for artificial intelligence research on the planet. The discussion with Michael Kratsios is from the second day of the conference. It is from early in the video from the second day, so it should not be hard to find. However if you are wondering what else is happening during these two days you can look through this public agenda.
It might be good to note that the Trump administration has been slow to fill roles within its tech Policy apparatus. Kratsios was confirmed as CTO in August after the position sat vacant for more than two years. However he has been working as a Deputy Assistant to the President of the United States for some time.
Who is Michael Kratsios?
Michael is introduced at the conference as the following, so I will repeat:
"Michael Kratsios the Chief Technology Officer of the United States. Michael was unanimously confirmed by the US senate in August 2019 to serve as CTO. He is just the fourth person in the United States to serve in this role. As US CTO Michael serves as the president’s principle advisor on a very broad portfolio of tech challenges ranging from quantum computing, 5G and rural broadband, autonomous vehicles, commercial drones, STEM education, advanced manufacturing as well as AI. Prior to becoming CTO Michael served as Deputy CTO in the early stages of administration. Prior to entering government he was a principle at Thiel Capital and served as Chief of Staff to entreprenur investor Peter Thiel. He graduated from Princeton with a degree in political science."

He is in a discussion moderated by Eileen Donahoe, Executive Director, Global Digital Policy Incubator, at the Stanford Cyber Policy Centre. She focuses on developing global digital policies to address human rights, security and governance challenges. Before Stanford she was the first US ambassador to the UN Human Rights Council during the Obama administration. She also served as the Director of Global Affairs at the Human Rights Watch. She serves at the boards of numerous organisations including the national endowment for democracy and the World Economic Forum council of Human Rights.

Free-Market Approach to Innovation in AI
We could sum up Michael’s perspective as the following. The regulations need to be tailored and sector specific. He mentions the free-market approach to innovation for the United States to stay in the lead:
"The way that I view the world is that there are technologies that are born free or technologies that are born in captivity." – Michael Kratsios the 29th of October 2019 at HAI fall conference.
Reducing regulatory risk associated with investments. He is asking how to create an environment especially for military technology. Michael is managing a large portfolio of investments, and he prioritises – through the role of convener building high-level strategies and directives.
Michael saw the opportunity to be ‘entreprenerial in nature’ therefore recruiting takes time finding individual portfolio managers. They recruit for technical expertise to drive towards policy. According to Michael a third of his time is spent on policy; a third is meeting with stakeholders; and a third is getting the word out (which he considers part of his engagement with HAI).
For him there has been a continuation of policy-making from the Obama administration to the Trump administration. Potential policy-impacts. R&D strategic plan on AI within fundamental research with allocations of money. They built an launched a national strategy by executive order.
Cleaner Regulations
At some point Michael mentions the notion of creating cleaner regulations. What I get a clear sense of is making regulations that will work better for businesses. He is not using the word of ‘deregulation’, however that is what he seems to be alluding to in his regulatory approach of technology.
"Deregulation is the reduction or elimination of government power in a particular industry, usually enacted to create more competition within the industry."
Because there is no such thing as clean regulation. By referring to this I think he may be referring to the notion of fairness and expediency, being able to implement solutions in the field of AI.
Combined with the free-market approach it seems he argues on behalf of ‘making it easy’ for large technology companies to contribute to the US economy or common projects such as defending the country. He mentions that drone testing by Amazon and Apple had to be done abroad, and that he has been wanting to bring it back to the US.
HAI, State Technology and US Values
Michael referring to his relationship with HAI says: "For us it’s having smart people like you guys communicating with regulators across all of our agencies so we can make a smart set of rules."
The moderator: Eileen Donahoe asked the question of digital authoritarianism. Michael answered:
"If the US can continue to be a leader in artificial intelligence we can ensure that the values that we hold so dear are the ones that are going to be underpinning development globally."
Therefore displaying the antagonistic relationship between US and China is important for Michael, and reflect the current Trump administration. They want to drive scientific discovery through a vibrant private sector funding of R&D by private and public funding. He mentions that this is different than what their ‘adversaries’ are doing, instead of a centralised plan.
The US CTO Supports the OECD Principles of AI
Michael Kratsios talks a lot about going to the OECD, and support their principles. This is incredible in my opinion because it opens the door for a lot more focus on a planet-centered approach to AI. The first principle in OECD Principles on AI is:
AI should benefit people and the planet by driving inclusive growth, sustainable development and well-being.
There is not much like this principle in the current executive US strategy, which means this may be used as an argument for driving new solutions or changing the focus in the US (although I think it will be an uphill struggle).
Questions to Michael
Of course I was not able to cover every point in the talk by Michael, and this was partly the point of creating a short summary. However he took several questions that I took note of and will try to communicate.
Public Data Cloud
One of the leaders of HAI, John Etchemendy, talk about a public cloud that provides free access to public-interest researchers. This possibility mentioned by John opens up for a lot of possible applications and as well possible misue as ‘public-interest’ is a very pragmatic approach in the United States. Public-private partnerships are great, however it seems a lot of the time public-interest in the US can almost directly be translated into private-interest. Michael comes from a mainly private background with Peter Thiel, and Palantir (that Peter owns) have gained several new defence contracts.
AI and Immigration Policy
Q: Your relationship to immigration policies?
A: One major pillar of US innovation is immigration. We believe in a merit-based innovation system. Politically challenging, and requires parters on the hill, it requires struggle to come to consensus. Improve or move towards merit-based system. It requires two separate lotteries, switch the ordering of a lottery. People who get access that have advanced degrees, they have a better chance to come and study.
In answering this question Michael dodges the discussions on the Immigration and Customs Enforcement (ICE) and the protests in large technology companies. However he says it is challenge with ‘partners on the hill’, so his priorities may be different although he must fulfil a mandate.
Innovation and Values
Q: US leadership in AI can have cost to society suffering certain harms, what costs should individuals bear in order to maintain US leadership in AI?
A: We are lucky here in the united states… The alternatives are too dire, we need to have interactions towards values that all of us hold dear are implemented in the future.
Investing in Advanced AI and R&D
Q [Fei-Fei Li]: Upstream issues and the pipeline issues, relating to STEM education, no matter how much we invest in advanced R&D we need that pipeline of talent. What is this administrations effort in boosting STEM education? One of the biggest issues is the lack of representation in the pipeline of talent. We had wonderful presentation on the lack of representation. How is this administration also working with congress, thinking about investing in diversity and inclusion?
A: US passed Americans compete to produce a five year plan on STEM education. Lots of our agencies across government, getting these together to drive STEM research. One of the pillars was the exact thing we were talking about, how to push underrepresented minorities into the STEM field. Layered onto this, we want to collaborate more, and we can connect you to the people. The Federal government can play a role in driving agendas, but most of the work is done at state local levels. How we can build those and be helpful.
Infrastructure and the Data-janitorial Groundwork
Q: We do not have a modern digital infrastructure. There was a research cloud mentioned (mentions E-Estonia), how do you provide datasets?
A: One of the least partisan issues have been modernising government technology infrastructure. One of the first stats is the total cost for IT services per employee is around the neighbourhood of 40’000 per year per employee, and for private it is 4,000. We can do a better job of spending those dollars. Programs like the United States Digital Service, to bring in talented engineers from Silicon Valley; another program is the Presidential Innovation Fellows program. PIF program, folks are able to come into the government, where they are able to work on efforts. Trying to push around centres of excellence, we have 25+ agencies, and there is no reason to do the same over and over again. We are in the early stages of launching an AI in government centre of excellence.
Protecting Democratic Elections
Q: How as CTO what specific initiatives are you personally spending your time and budgetary support on to protect democratic elections?
A: That is run out of DHS, someone that works closely with state governments, state-local efforts. To the extent that anyone I can put you into contact with the National Security Coucil.
Michael is CTO, yet he is not responsible for protecting elections or election fraud. It could be argued US is more concerned with Defence than internal democracy in their strategy relating to technology. As the comedian John Oliver said in a recent episode of Last Week Tonight digital voting is not unproblematic and often outdated.
More Cross-Sector and Cross-Disciplinary Work
Q: There is a growing recognition that to succeed at the normative and innovation side there has to be collaboration between academists and industry. The HAI program is a good indicator of that trend. Having a mind for the technological challenges as well as the normative societial challenges. Cross-sector and cross-disciplinary work.
A: One of the biggest reflections is the R&D strategic plan on AI, one of the eight pillars that covers these issues of ethics that we should think about it not as plainly technological questions, but kind of a larger social science question. When we have the high-level discussion of people that meet quarterly on AI, cross-agency collaborations. The big manifestations is the regulatory guidance memo.
Parting Advice
Q: What parting advice may you have for people wanting to get involved in AI tech policy?
A: Doing public service is important, an opportunity was presented to me, we live in by far the greatest country in the world. The freedoms the power, so many people do not have this, we have a very special country here. We have amazing national labs, an environment that if we continue to foster it we can drive the changes here. I would be more than happy to talk to you and make that a reality.
Recent Moves in US Tech Policy
Besides the conference we can consider the hiring by Michael Kratsios. According to NextGov writing on the 21st of November 2019, there has been two changes to consider:
- Dr. Lynne Parker and Winter Casey would move up the ranks within the Office of Science and Technology Policy. In their previous roles, Parker spearheaded the development of the administration’s national AI strategy and Casey worked with allies to build consensus on tech issues. They are both now Deputy U.S. CTOs.
- Eric Burger would join OSTP as assistant director for telecom and cybersecurity. He is responsible for the telecommunications and cybersecurity portfolios. Perform coalition building, coordination, and policy work related to the United States’ use of the radio spectrum, communications networks, and cybersecurity.
Conclusion
It seems clear that US high-level strategy will continue to favour a free-market approach in technology that will be heavily reliant on the opinions of the large technology companies. Since administrators and public officials are coming from the private sector, especially from investment, VC and ‘startup’ type business it is not hard to see the type of power that their existing relationship in technology will have towards these new collaborative efforts in the US. Michael is taking steps to strengthen the development of US policy relating to technology, and we will have to see where that takes us, seeing as US is according to many sources currently the leading nation in the field of AI.
This is #500daysofAI and you are reading article 188. I write one new article about or related to artificial intelligence every day for 500 days.