The Implications of AI Taking Over the World, as told by Microsoft’s President Brad Smith

In April I had the opportunity to attend Columbia University’s Data Science Day, a day focused around the hottest research in the data science field and its impact on different industries. The keynote speaker — Brad Smith, Microsoft’s president and chief legal officer — stressed the importance of the ethical considerations of artificial intelligence. He emphasized seven key issues with AI that have the most power to dramatically affect the world.

(https://www.wired.com/story/microsoft-wants-stop-ai-facial-recognition-bottom/)

1. Fairness

Throughout history we have passed laws and regulations to ensure humans treat one another fairly. Now that we are utilizing artificial intelligence and it is becoming embedded in every aspect of our lives, we must also ensure that technology follows the same principles responsible for these laws and regulations.

For example, the Equal Rights Amendment was passed by Congress in 1972, giving women and men the same legal rights in the United States. This law should equally apply to the possible hidden biases in algorithms. If an algorithm weeding through job applications happens to choose more males than females for an open position, this should be seen as a violation to this law.

2. Reliability and Safety

Each industry has different safety precautions that they take to ensure the public can rely on the products they produce and the services they provide. For example, the automobile industry must test each new model of car a certain number of times to ensure they are safe to use. The same kind of thing must be done with AI. We must test the models that are being created to make sure they produce the results that we intended.

3. Privacy and Security

We are currently in the middle of a data revolution. Data has become a valuable asset and its importance will only continue to grow as technology grows. But with the need for data also comes the possibility for the invasion of privacy. Data scientists need to ensure that the intentions for the data they are collecting are made clear and they are only collecting it from someone with their consent.

Europe recently passed the General Data Protection Regulation, requiring companies to ask for permission to use data and to be completely transparent on how they are using it. Smith suggests embracing the idea of “data minimalism”, meaning data should only be used for what it was provided for and then be deleted.

4. Inclusiveness

AI needs to be created with all different types of people in mind. It must consider the needs of those of different genders, ages, races, sexual orientations, and those with disabilities. This is similar to the idea of fairness in that various biases are possible if these technologies are not made with all types of people in mind.

5. Transparency

If AI is going to be incorporated into every aspect of our lives, each person needs to understand how it is being used. This means technology companies must take the initiative to publish their algorithms in a way that’s easy for the general public to understand. A person who has no background of technology should be able to learn how AI is affecting them and what data the model is collecting.

6. Accountability

We must ask ourselves, who will be held accountable for mistakes that AI makes? Is it the machine itself or the human behind the technology? Laws and regulations should be put in place so that the people creating these machines are held to a high standard of ethics. When actual people are held accountable, these other factors discussed are more likely to be considered and prioritized during the engineering process.

“We are the first people on the planet to give machines that kind of power. If we fail, we do a huge disservice to all future generations.”

Smith ended his lecture saying, “we are the first people on the planet to give machines that kind of power. If we fail, we do a huge disservice to all future generations”. If we do not take the time to consider these seven factors now, there is potential for a lot of damage down the line. As AI becomes integrated into our daily life, it is essential that we are all aware of its effects and how it’s behaving.