Author: Yogi Nelson

Namaste Yogis. Welcome to the Blockchain & AI Forum, where your technology questions are answered! Here no question is too mundane. As a bonus, a proverb is also included. Today’s question was submitted by Mr. Bunch from UCLA, and he asked if the United Nations is united on artificial intelligence regulations?

Mr. Bunch, you came to the right place. Recently I traveled to Geneva, Switzerland and met with two U.N. employees. We discussed numerous topics, including your question. I’ll answer you by sharing what I learned.

In 2020, the U.N. decided to launch an AI initiative related to regulations of emerging AI technology. Markus Krebsz, Professor of Management, University of Stirling, was selected to lead the investigation. Professor Krebsz is also the Founder of the Human-AI Institute. Four years later, Professor Krebsz produced a report titled, U.N. Economic Commission for Europe Working Party 6. For the sake of brevity and sanity, I’ll refer to the report as WP6. Besides, WP6 sounds like a fun rock band, and good time! Lol. Let’s examine what’s inside WP6, commencing with the purpose, followed by findings and recommendations.
https://unece.org/sites/default/files/2024-07/ECE_CTCS_WP.6_2024_11_E.pdf

PURPOSE OF WP6
WP6 provides guidance to member states (the U.N. calls nations member states because they are all members of the UN) for an overall approach to regulations of products and/or services with embedded AI systems or digital technologies. The author reports the most appropriate use of WP6 include:
• Setting legitimate regulatory objectives
• Identifying and assessing risk
• Identifying relevant international standards
• Establishing mutually recognizable conformity assessment procedures
• Setting market surveillance and other enforcement mechanisms
• To promote convergence of national technical regulations among agencies currently in place or yet to be developed.

KEY FINDINGS & RECOMMENDATIONS
Krebsz starts with an obvious but nevertheless important disclosure—zero risk is impossible. There is no way to avoid making an error when establishing AI regulations. Hence, constant evaluation of AI regulations is prudent.

Second, certain products/services may be intrinsically high risk to human society. Therefore, governments must assess risk and choose appropriate conformity assessment methods, say Krebsz. In high-risk situation, AI cannot be allowed to override human control and human decision, according to WP6. Krebsz cites the example of sophisticated diagnostic system, where even if the medical equipment could generate algorithmic decision-making, human decision making must be included.

The third recommendation concerns AI technology embedded within products. Krebsz notes consumers may not be fully aware certain products contain AI embedded technology. Thus, consumers might be surprised to discover the technology reacting in an unexpected manner. The solution is robust testing under various scenarios, say Krebsz. Krebsz applauds the National Institute of Standards and Technology AI Risk Management Framework for their work on this issue.

WP6 supplies a series of AI bullet point recommendations and policy guidance, including but not limited to:
• AI embedded systems should not impinge of human freedom nor cause bias.
• AI technology should not further widen the digital divide as outlined in previous U.N. resolution.
• No AI should cause mental health problems.
• AI should be designed to encourage trustworthiness. In other words, transparency under the hood of the technology.
• Scenario and simulation testing and evaluation of residual risk should be part of a national AI policy.

Apparently, Professor Krebsz’s report was a smashing success! The member states representatives adopted WP6 and all signed the corresponding resolution, also written by Krebsz!

Time to leave you with a proverb from Mali, where they say: “The bird that flies off the earth and lands on an anthill is still on the ground”.

Until next time,

Yogi Nelson

Namaste Yogis. Welcome to the Blockchain & AI Forum, where your blockchain and artificial intelligence technology questions are answered! Here no question is too mundane. As a bonus, a proverb is also included. Today’s question, was submitted by Rolando from Panama City, and he wants to know about artificial intelligence (AI) governance?

Rolando, you came to the right place. Plenty of people are asking the same question. Recently I came across an KPMG article titled, “The Shape of AI Governance to Come”, in which KPMG explored fundamental questions related to AI governance. KPMG is a major business consultancy firm and recognized thought leader in the field of AI. Let’s see what KPMG said, beginning with the notion of trust.

According to KPMG research, for AI to become transformative, four critical trust factors must be addressed: integrity; explainability, fairness, and resiliency. These four trust factors make sense and are not surprising. The newsworthy discovery is that 87% of IT decision makers advocate regulations to achieve trustworthy AI governance. If regulations are the solution, what is the current state of regulations? KPMG spotted several trends.

First, the most advanced AI markets are putting in place regulations to protect and promote domestic research and development. Second, governments are establishing advisory committees consisting of industry, academic, and government officials to produce guidelines, regulations, etc. For example, Austria passed an AI ethics law, Singapore adopted a 5-year AI plan, and Japan also approved an AI technology strategy. Laws and regulations are a third trend. The implications of these trends could be far reaching and hence worth an examination.

Clearly, governments hope to realize long-term benefits through increased overall operational efficiency for both the public and private sectors. Second, governments expect to create intellectual property related to AI, and thus obtain patents, copyrights, and trademarks for various AI intellectual property. Attracting and keeping talent is a third implication. People will want to work in a country leading the way in AI, thus setting the stage for tax and immigration policy debates.

KPMG acknowledges AI moves faster than politics, hence KMPG provides six AI governance ideas businesses can deploy now until politics catches up.

2. 1. Develop AI principles, policies and design criteria, and improve controls within an environment that fosters innovation, flexibility, and trust while identifying unique risks. Inventory AI capabilities and use cases by understanding the AI footprint within the organization.
4. 2. Design, implement, and operationalize an end-to-end AI governance and operating model across the entire AI development life cycle, including strategy, building, training, evaluating deploying, operating, and monitoring AI. Establish separate governance committees and councils to discuss the unique risks and complexities associated with AI.
6. 3. Assess the current governance and risk framework and do a gap analysis to identify opportunities and areas that need to be addressed.
7. 4. Design a cross-functional governance to deliver AI solutions and innovation through guidelines, templates, tooling, and accelerators to deliver AI solutions quickly, but responsibly.
9. 5. Integrate a risk management framework to find and prioritize business-critical algorithms, and incorporate an agile risk mitigation strategy to address cybersecurity, integrity, fairness, and resiliency considerations during design and operation.

Time to go, but first a proverb from Suriname, where they say: You don’t need to see the top of the river to know which way the current is flowing.

Until Next Time,

Yogi Nelson