We Need to Be Smart to Create Intelligence

As a society, we have an ambivalent relationship with artificial intelligence (AI). While we love the benefits it offers, we are also afraid of it, because it represents the great unknown. So, newspapers are full of articles about computers beating people at chess or Go. Just recently, a Tesla accident, when the car was working autonomously, raised numerous questions about the safety of AI-controlled cars.
But these are early days for AI. We will invite AI into new areas of our lives as we realize the benefits it could offer. I believe that in the future, we will be working with machines to solve some of humankind’s greatest challenges, such as disease, poverty or even wars. Still, too often we are frightened by words such as Singularity – the moment when computer intelligence will surpass human intelligence, even though there is no real proof that a moment like that will ever arrive or even have the consequences we imagine.
In the short term, AI will be built into ever more products. Just look at all the girls – Cortana, Alexa and Siri – that are powering today’s experiences. They are actually a good example of what AI can do already and will be able to do even better in the future. They provide a friendly interface and user experience to interpret our intent and mine the available knowledge. In the future, they will have the ability to reason over large amounts of data and do pattern recognition more quickly.
However, to minimize any fears of machines “eclipsing” humankind, we do need to have a reasonable and fact-based approach to AI. At Microsoft, we’ve defined several principles that we should discuss and look at when we work on this exciting area.
First of all, AI must be designed to assist humanity. This is the number one goal, if we want to ensure that cars driven by AI are safe and respect human autonomy. Also, machines can be more than useful in dangerous work, such as mining, where human lives are at risk and the introduction of intelligent technology can easily reduce the risks.
The next step is transparency. AI should be transparent and people should be able to see how it works and what its rules are. The fact is that technology will know a lot about us, sometimes even things we don’t know ourselves. This is why it transparency is needed – we need to have the same level of knowledge about machines, so we can ensure that they are used ethically and responsibly.
AI must be used without impacting the dignity of people. It should preserve culture and diversity, which is why more than just the tech industry should be working on AI. We need more diverse populations to work on it to ensure that different cultural and ethical norms and expectations are integrated.
Then there’s security and privacy. AI needs to be programmed to protect the privacy of people and earn their trust. There is also the need for some level of accountability so that humans can undo unintended harm. We must design these technologies for the expected and the unexpected.
Overall, I am excited about the field of AI. Machines can help us achieve so much more – they can perform dangerous work and analyze data to provide real-time feedback and information about the world around us. They can help us understand the world and people around us better. Driverless cars and similar technologies are just the start as some level of intelligence will be integrated into every product and service we use. And to realize this potential, we need to be smart today and define the principles and goals that will guide this revolution.