Have you ever wondered whether the technology that improves your customer experience is impinging on your privacy at the same time? Many artificial intelligence algorithms are based on consumer and vendor data, often collected without the knowledge of the affected parties.
When researchers first began to develop AI in the 1950s, their goal was to teach machines how to mimic human intelligence. In recent years, this approach has been abandoned, in favor of relying almost entirely on user data.
However, 80% of companies that rely on artificial intelligence fail because they have security and legal problems. Some have been sued for fraudulently gaining access to proprietary information. Even tech companies who are careful about security may unknowingly leave themselves open to security breaches and legal complications. There have also been cases in which rumors about potential security issues have resulted in negative publicity and subsequently damaged a company’s reputation.
Tech companies have begun to address these problems, but tweaking the algorithms doesn’t solve the root of the problem.
A better solution to the security and legal problems of AI is to create systems that don’t rely on customer data. Instead, these systems are trained to work on synthetic data, which is reproduced data based on real-world statistics. It was first developed in 1993 and has since been refined.
Technology developers who have gone back to the original way of thinking about artificial intelligence are currently creating synthetic data systems. For instance, DARPA (Defense Advanced Research Projects Agency) is investing $2 billion in the creation of models that mimic core domains of human cognition. And Google’s parent company, Alphabet, has been working on an initiative called Project Loon since 2013, which uses Gaussian processes (probabilistic models that can deal with extensive uncertainty, act on sparse data, and learn from experience) in order to launch balloons in the atmosphere which provide internet to underserved populations around the world.
In 2018, Facebook announced plans to open two new AI labs that rely on synthetic data, in order to protect its users’ privacy. This move followed a 2015 lawsuit against the company for violating privacy laws. Earlier this year, Facebook agreed to pay $550 million in damage to users whose privacy had been violated.
Synthetic data systems don’t collect any private information, so there is no danger of data breaches and no potential lawsuits looming on the horizon.
Ethical artificial intelligence has advantages beyond customer protection. Since these systems don't need to pull data from clients, they have a much faster integration than a system that is based on private information. The technology can be used quickly and easily by the end-user, without the need to provide data prior to its implementation.
This works best when the process doesn’t rely entirely on artificial intelligence. Instead, it follows a multi-disciplinary approach. For example, in order to automate vehicle inspections for rental car companies, the system combines key engineering elements to create unique identifiers for each damage type and measurement, so that damage can be assessed accurately and quickly.
More specifically, the Click-Ins system adjusts 3D models of vehicles to be absolutely precise. The system is trained to compare the model with photos of the actual vehicle, even when the pictures are of low quality. It can ignore bad lighting, shadows, and dirt to analyze how the photo is different from the car’s model.
This system translates into faster integration, high ROI, and company savings. It doesn’t rely on company or customer information, so the client only has to provide the most basic information for the system to work. The solution is ready to go and can be implemented almost immediately, saving time and manpower.
Ethical AI mimics human intelligence and utilizes synthetic data in order to eliminate the ethical and technical disadvantages of customer-dependent AI and creates a solution that works fast and well. As more companies invest in this type of artificial intelligence, the future of tech will be positively impacted and end-users will benefit from more secure and cost-effective systems.