Microsoft AI is being developed in stringent conformity to AI ethics
At the end of November 2018 the Chief Information Officers (CIOs) in East Africa assembled for a two day CIO100 Symposium at Enashipai Resort and Spa, Naivasha, to talk about among other things Artificial Intelligence. Microsoft, being one of the sponsors of the event, took the opportunity to talk about Microsoft’s philosophy on AI. Microsoft’s presentations were made by several speakers including Sebuh Haileleul, Country Manager, Microsoft East Africa and Christopher Casarrubias Director, Emerging ICT Markets Access Partnership and were based on a White Paper the company recently published which basically outlines how Microsoft AI is being developed to amplify human intelligence, instead of competing with it.
In a nutshell, whether AI is deployed for enterprises to effectively and efficiently manage business processes, or to Industries to automate production lines, or to Agriculture to solve the ever growing food demand, Microsoft’s philosophy is that AI must be deployed in such a manner that Fairness, Reliability & Safety, Privacy & Security, Inclusiveness, Transparency, and Accountability are observed. These key tenets form the backbone of any AI project that Microsoft sets itself to do, be it implementing AI for office operations and efficiency through Azure Cloud solutions, or implementation of AI in farms through integrated technologies like Farmbeats.
While developing and deploying its AI by following ethical values including respecting privacy and security of those who provide data, Microsoft is not lost to the fact that for AI to flourish, data is needed – big data. And it is not just any data, but quality data. Like any human or any other life form, junk data will simply create junk AI, so the quality of data that is being fed into the machine learning algorithms that ultimately translate to AI must be very clean and healthy. We already know of situations where unhealthy data has led to AI labeling black people as monkeys, and a Microsoft’s own Twitter chatbot Tay being taken down for being “a white-supremacist mouthpiece” who among other things claimed that the Holocaust was a fiction, blamed 9/11 on President George W. Bush, and described President Obama as a monkey. Thus, for AI to meet the criteria of Fairness, Reliability & Safety, Privacy & Security, Inclusiveness, Transparency, and Accountability, that AI has to be fed properly and appropriately with good healthy
The other thing that Microsoft is not blind to is the fact that AI isn’t supposed to solve only the enterprise problems. Other real world problems specifically those that abound in remote rural areas need tackling too. The problem is that collecting data in remote rural faces two humongous challenges: 1. There are hardly any humans who live in remote areas, and 2. The few who live in such places do not have access to the Internet, the primary means through which data for AI is collected. These two problems make it hard for anyone willing to develop AI capable of solving perennial problems such as drought in one season and floods in the next, scarcity of food, non-performing agriculture, plant diseases and pests, etc, to collect the needed data that would help the AI understand just what the problems are. Foremost, companies like Microsoft are in the business of making profits, so they actually do not have incentives to venture into such remote areas to spend the needed billions of dollars to come up with sustainable solutions for those perennial problems.
The bottom line mentality has however not stopped Microsoft from doing what needs to be done. In Africa for example, Microsoft partnered with Techno Brain to launch a data collection platform to help farmers improve their yields. The digital platform currently collate data from farmers in Tanzania and Malawi in a pilot programme. The data being collected include GPS coordinates, expected rainfall and weather patterns, land type and soil nutrition. This data is then processed via Microsoft’s AI powered cloud (Azure) to create insights that help farmers in Africa make better farming decisions.
This proactive measure by Microsoft was echoed by Christopher Casarrubias Director, Emerging ICT Markets Access Partnership during the CIO100 Symposium when he said, “As countries around the world become increasingly aware of the potential economic and social benefits of developing and applying AI, policy frameworks and principles to support the AI ecosystem are on the rise. While there are real challenges to overcome, the potential for growth and development is too important to pass up. With the right policies and collaborations across stakeholders, Africa can leverage AI to benefit all of its communities. “Through discussions with other industry leaders at the event, Microsoft aims to bring across its three key focus areas in preparing for the impact of AI on the continent: Preparing today’s students for the jobs of tomorrow and assisting workers in preparing for a transforming economy”, sentiments that were emphasized by Sebuh Haileleul, Country Manager, Microsoft East Africa, who said that while forward-thinking policy makers and the like have already started embracing and promoting the need of a vibrant AI ecosystem on the continent, structural challenges still exist that hinder the realisation of this. “Microsoft and Access Partnership recently collaborated on a white paper, entitled: Artificial Intelligence for Africa: An Opportunity for Growth, Development and Democratization. It demonstrates that, for a healthy AI ecosystem to exist, education systems need to adapt quickly, broadband coverage must expand rapidly, and ethical implications regarding the fair, secure and inclusive use of AI applications must be addressed through collaboration and engagement to ensure trust in AI.”