Back in April the Chinese police identified and captured Mr. Ao in a crowd of 60,000 people attending a concert. Mr. Ao was required by the authorities for economic crimes. The news spanned worldwide attention because of the ability of the Chinese government to spot a wanted man in a public crowd using their surveillance network, which by the way now has over 170 million facial recognition cameras, one of the largest monitoring systems in the world.
How China is using Artificial Intelligence to power its public security system is a clear example of the extent of the technology and how it enables a more efficient workforce. Despite the benefits, the technology also raises criticism on how deep privacy concerns can go.
Beyond the facial recognition capabilities and providing the police with tools to do their job more efficient, these AI technologies are harvesting tons of data, that can be used to extract meaningful information about the population and its dynamics; surveillance is just the tip of the iceberg.
The AI Hype
China’s surveillance systems, autonomous cars (Tesla, Uber, Google), virtual assistants (Siri, Google Assistant, Cortana), smart home devices (Amazon Echo and Alexa) are among the most prominent examples of how AI has penetrated our daily lives. On a smaller scale, Gmail’s ability to suggest answers to emails you receive, Google Photos’ ability to automatically organize and tag your pictures, and Google Docs’ speech-to-text capabilities are things we have used and incorporated into our daily lives. We often don’t really imagine the technology powering these functionalities.
AI is everywhere… and yet adoption is slow. Big companies have been pushing really hard on fine tuning their AI offerings and making it easier to integrate AI into anything. Long past are the days were you needed to develop your own model. These days, AI technologies are mature enough to just integrate them. Even though most executives believe AI will allow their companies to have a competitive advantage or enter new markets, only a few companies have extensively incorporated AI in their offerings or internal processes, and only a handful have a clear AI strategy in place.
Gartner predicts AI will hit the $3.9 trillion mark by 2022. So why, if AI is so promising, do only a few companies have a clear AI strategy? How come even fewer have incorporated the technology? The excitement about the potential of AI and what it unlocks becomes blurry when companies start to realize what is required to successfully deploy AI: lots of high-quality data, smart and trained people, and an experimentation culture.
It’s all about the data
Human intelligence is incredible, and it’s the human brain that makes it all possible. Today, no artificially intelligent system is able to replicate the functions of the human brain. Artificial general intelligence looks like a remote dream by today’s standards and technologies. What we are seeing is narrow intelligence, specific models applied to specific problems, which means no genuine intelligence, self-awareness or complex reasoning. If we intend to create models that relate the past with the present to infer possible alternatives, we have a long way ahead before making it possible.
Algorithms, just like the human brain, need to “learn”. The quality of AI solutions is as good as the data used to train them. Even though AI technologies are ready to be plugged to new and existing solutions, it is not a matter of just plugging them in and watching them go. In order to provide a high degree of accuracy, high-quality data needs to be used to train and validate the algorithms so that the result is not a “garbage in, garbage out” solution.
It’s all about the people
AI is a highly skilled business. Not only do businesses need to incorporate new skills, they also need to educate the current workforce and executives to enable AI to really cause positive business disruption. The AI field is multi-disciplinary. It requires business and engineering expertise across different areas that are not necessarily and readily available in organizations. Trying to deploy AI without the right team is a recipe for failure.
It’s all about the culture
As Peter Bebbington, CTO of Brainpool, puts it: “Business is uncertain, life is uncertain and nothing can give perfect prediction.” Uncertainty rules the world, and because of this no model will be 100% accurate. Imperfection is part of our lives, which is why AI requires experimentation to find the best solution. Experimentation isn’t always something that is well received by organizations, which generally require structured plans and predictability in projects and results.
A culture of experimentation rewards failure, this is the key to innovation. Failures need to be seen as learning opportunities, and should feed the continuous improvement process that empowers the AI strategy in an organization. Experimentation does not mean improvisation. Experimentation means having quick iterations that enable discovery and accomplishment towards a bigger goal. It means creating a culture of educated risk-taking that fosters creativity and innovation.
Where to go next?
To get ahead of the AI hype and start integrating AI into your organization, it is imperative that you educate your people on what your organization can achieve with AI and start designing an AI strategy that will boost all AI efforts within the organization. Start small, experiment, and grow your organization’s AI capabilities. The right person or partner will definitely play a key role in guiding a successful strategy and implementation, given that no AI capability exists inside the organization.
We can safely focus on integrating AI into our business to improve customer experience, generate new revenue, and enable cost-reduction, while others debate on whether AI will kill the human race or not.