Artificial Intelligence and Governance: Quo Vadis? (Part I)

Abraham Lincoln once famously said: “You cannot escape the responsibility of tomorrow by evading it today”. Upton Sinclair thought that “People can’t understand new ideas if their livelihood depend on the old ones”. Both quotes can be applied to the stand of AI in industries and governmental sector. Over the short term, AI will automate tasks that were done by humans before. Next waves of AI will disrupt traditional economies, companies; the way people acquire jobs and getting educated. Without effective governance to mitigate risks to our current ways of business and life, inequality will worsen. Safety and control will grow in importance as AI becomes embedded into any process and any product.

In this part of the “A.I. and Governance” I will address most common concept in A.I. every non-technical executive and board member should know and understand.

___________________________________________________________________

On January 16, 2018, the chief executives of the world’s largest public companies received a letter by Laurence D. Fink, founder and CEO of the largest investment firm in the world — BlackRock. He informed business leaders that their companies had to do more than make profits. They needed to contribute to society if they wanted the backing of his firm.[i] Such a position hopefully demonstrates an emerging shift in corporate governance. Fink addressed social changes due to retirement, infrastructure, automation, and retraining of the employees so they are fit to work with new technologies. Automation, retraining and infrastructural challenges are clearly connected with the rise of Artificial Intelligence.

I interviewed 60 top executives and board members while writing “The Artificial Intelligence Imperative. A Roadmap for Business” (Praeger, 2018). In sum, traditional boards (and companies they oversee) are not sufficiently prepared to address and fully benefit from AI.

Figure 1: Traditional Companies are Poorly Prepared to Benefit from AI

From the demonstration of an ‘artificial intuition’ with DeepMind AlphGo Zero to reviewing legal documents, screening through heaps of medical images to detect diabetic retinopathy more efficiently than humans can, AI is progressing at a very fast rate. According to a 2013 McKinsey Global Institute report, it is the attitude of business leaders and executives that will dictate the path of AI development and advancement in their companies.[ii] Since most directors see their business models might be disrupted sooner than later by Internet players, they are more willing to hire directors with background in IT, implement technology committees, and have updates from CIOs or digital officers on a regular basis. Artificial Intelligence, however, is so far-reaching, that the knowledge and experience needed goes beyond occasional technology updates, traditional risk management and historically backward looking governance frameworks of a board. If implemented consequently, AI touches every aspect of organizational culture, and amplifies weaknesses; if too many conflicting goals compete within silos, AI is used just for cost-cutting, and employee communication around automation and augmenting of tasks is mismanaged.

Figure 2: Examples of 2017 AI Advances and AI M&As

What is Artificial Intelligence? It covers a group of technologies and scientific disciplines that focus on automation, acceleration and extreme scalability of human perception (e.g. the capability to see, or to understand and speak a human language), decision-making and reasoning.

AI Concepts Shaped by Social Sciences

Artificial Narrow Intelligence (ANI) is an AI specializing on just one task, e.g. scheduling a meeting, recognizing patterns on radiological images, filtering spam in email accounts, or remotely predicting when a rail track requires maintenance. ANI is considered to be a great productivity factor and is generally perceived to be harmless. Andrew Ng, an Chino-American AI researcher, practitioner, and Investor believes, that we need to dedicate most resources to ANI, since many technologies are ready and the impact on bottom and top line of companies will be visible in short time.

Artificial General Intelligence (AGI) should be able to “mirror” the behavior and capabilities of a human to solve problems, comprehend abstraction and complexity, learn from experience and find the best way to cope with a new situation. Most AGI researchers try to emulate the human brain. Physicist Max Tegmark and Elon Musk warn, that besides focusing ‘just’ on creating such an AGI, we need to figure our alignment of its and ours goal, and dedicate a lot of resources to AI safety. Tegmark with his Life Institute and Musk with research organization OpenAI[iii] are among other smaller initiatives to make our future safe.

The University of Oxford philosopher Nick Bostrom defines Artificial Super intelligence (ASI) as “an intellect that is much smarter than the best human brains in practically every field, including scientific creativity, general wisdom and social skills”.[iv] The timing for arrival of such an ASI is disputable, with predictions ranging from 2024 to 2060.

AI Technologies

Most progress was made around statistical approaches in AI, represented in different tribes of Machine learning. These approaches are focused on mimicking the human ability to guess and represent over 65% of what AI science is about today. Within ML Deep learning is attracting most investments and research talent.

Figure 3: Deep Learning within AI

Machine Learning enables systems to automatically improve their performance at a task by observing relevant data. It has been the key contributor to the AI surge in the past few decades, ranging from search and product recommendation engines, to systems for speech recognition, fraud detection, image understanding, and countless other tasks that once relied only on human perception skills and judgment.

Recent progress in ML was possible due to the availability of vast data sets, progress in semiconductors (especially development of GPUs — Graphic Processing Units), the commoditization of cloud technologies, and the increased participation of universities and top technology firms in open research.

Figure 4: Enablers of Machine Learning

.

To Be Continued. In the next part of “A.I. and Governance” I will talk about how machines learn from data, talk about perception technologies in AI, and bottlenecks, e.g. biased data and algorithms.

[i] Andrew Ross Sorkin, “lack Rock’s Message: COntribute to Sociatey, or Risk Losing Our Support, NYT, January 15th, 2018.

[ii] James Manyika, Michael Chui, Jacques Bughin, Richard Dobbs and others, “Disruptive Technologies: Advances that Will Transform Life, Business, and the Global Economy”, McKinsey Global Institute Report, May 2013.

[iii] Elon Musk left the board of Open AI in February 2018 to avoid conflict of interests with development of AI at Tesla. He continues to finance and advise the organization.

[iv] Nick Bostrom, “Superintelligence”, 2014, page 22.