It is an unfortunate truth that most people around the world don’t have a real understanding of what artificial intelligence (AI) is. The same is true for many of the influential people who are in active discussions and working with companies where AI plays a significant part in our future. In fact, a recent survey found that most organisations don’t have a clear understanding of how AI or machine learning will help their businesses.
In general, people’s perception of what AI is comes from movies and media, with doomsday scenarios and super intelligent robots taking over the world. It’s not surprising given the unprecedented speed of development in AI, and the hype that has come with it. Our inclination as a species to seek out familiar references as a method to understand new concepts does not really help either.
The realities of AI are very far from what the general perception of it is. AI, in particular, deep learning, is absolutely amazing at solving very precise tasks. There are endless opportunities to improve the lives of people around the world, in all areas of society. There are millions of tasks where humans do an okay job today, but where we can use the cognitive power of AI to do greater things and solve new problems. Like finding and segmenting brain tumours, diagnosing skin cancer, optimising industry to decrease emissions, predicting weather patterns, developing autonomous cars that never crash and anticipating health needs. The list of possibilities is endless.
When we operationalise AI technology on a broad scale, the positive impact on society will be enormous. Far greater than anything humanity has ever seen before.
So, what is everyone getting wrong?
On the one hand you have the general public over simplifying and hyping up AI. On the other hand, education around AI is currently dominated by AI experts who come from research, and focus on algorithms, technological advancements, methods, and then publishing their new findings. They are some of the sharpest minds on the planet, and are doing a fantastic job of uncovering new science, techniques, and methods to solve humanity’s problems.
But, when asked to explain the topic of AI they often fail to convey what the future has in store, and how it will affect our lives. Instead, they focus on their field of expertise, the science. Not the end use. This is understandable given their technical backgrounds. However, these AI experts are regularly invited to business summits and political meetings to explain what AI is. Very often this results in someone on stage explaining the concept of convolutional layers to an audience who really want to know how it affects them, their lives and their businesses. Think about if you were to see a car for the first time in your life and went on to ask the question “what is that?”, and then the mechanic starts to explain the concept of combustion engines, it probably wouldn’t make you particularly excited about taking that car for a drive.
Professor Damian Borth from St Gallen University also explains this well. People outside of the academic world usually do not care about how things work. At the very least, they do not see this as the first step in their process of learning. They care about how their lives will be affected. They need to understand how it will affect them on a personal level.
But, do we need to have a full understanding of AI?
The short answer is yes. Currently, AI technology is only being operationalized by a few large tech companies and governments around the world. This is happening faster than any technological shift humanity has ever seen before and the vast majority of people don’t understand how AI will impact them.
For people all over the world to enjoy the positive change of scientific advancements, understanding how AI will affect them is crucial. Without this knowledge, we cannot have productive discussions or make informed decisions. Leaving AI breakthroughs solely in the hands of a small group of AI experts is preventing the technology from becoming widely available for everyone across the world, in healthcare, education and infrastructure.
So, how do we tell the AI story?
For as many people as possible to understand what AI is we need to convey what the future holds. For everyone in society. For you, me, our children, friends, family, colleagues, and neighbours.
In every ‘AI story’, there are five main chapters:
1. The problem
The thing we want to solve; improve efficiency, predict something, find something.
The thing we want to solve; improve efficiency, predict something, find something.
2. The data
The available data used to train an AI model.
The available data used to train an AI model.
3. The AI model
When trained and deemed to be ‘successful’, i.e. able to see patterns, come to conclusions and predict outcomes better than previous methods, or humans.
When trained and deemed to be ‘successful’, i.e. able to see patterns, come to conclusions and predict outcomes better than previous methods, or humans.
4. The deployed AI model
When an AI model has been trained, it needs to be put into production, in a stable environment where it can be hosted, maintained, retrained, etc .— but most importantly called upon by the actual products or services using the AI model. For example, a weather or traffic prediction service, or as part of a radiotherapy process identifying tumours faster and with higher accuracy.
When an AI model has been trained, it needs to be put into production, in a stable environment where it can be hosted, maintained, retrained, etc .— but most importantly called upon by the actual products or services using the AI model. For example, a weather or traffic prediction service, or as part of a radiotherapy process identifying tumours faster and with higher accuracy.
5. The effects
The final impact or outcome produced by the AI model, and the service or product for businesses, society and people in general.
The final impact or outcome produced by the AI model, and the service or product for businesses, society and people in general.
A rounded view of AI
Currently, most stories are told only with chapters two and three, the data and the AI-model.
Unless we talk about the problem we want to solve, what service we need to build, and how they will affect people — people outside research and academia will never be able to understand the importance of AI.
All major discussions about the future, regulations and ethics for AI and how society and businesses will be affected are useless if you do not understand all of the five chapters above. So, the next time you discuss AI, start from the end of the story with chapter five. Ask the following questions: How will AI affect us? What problems can it solve? What services are needed to make it happen? If we can make AI understandable and accessible for everyone then the possibilities for humanity are truly endless.
No comments:
Post a Comment