Human Element and AI
AI (Artificial Intelligence) is no longer movie special effects or dreams of the future. This development in technology is now a reality and it is the new hype all across the globe. We then wonder, would these digital assistants replace the human labor force? Thankfully, we are still in the development stages where investors are eagerly funding research and development. Replacing the workforce may eventually happen, but not for the next few generations.
Developers claim that AI is currently used to improve on human decision-making abilities by capitalizing on machine-learning, data science and a combination of algorithms to create augmented intelligence. One such example is the implementation of Hart (Harm Assessment Risk Tool) an AI system, by the Police force in Durham USA. This system helps the police to identify whether a suspect should be kept in police custody by classifying the suspect according to high, medium or low risk, based on data from police records between 2008 and 2012. In 2013 when the system was tested, its accuracy of high-risk suspects was at 88% and 98% for low-risk suspects.
While this application of AI proved to be beneficial, there were limitations that needed human intelligence to prevent unfair profiling. For instance, because the data provided to the AI system was through the internal database, human intervention was still needed to search through larger external databases such as the national police computer system. What would happen then, if the AI system had access to all databases?
The GloVe algorithm was found to have over 840 billion trained words from the internet and according to a study by Princeton university, these AI bots become biased based on discriminatory views of gender and race and essentially, a reflection of their creators, which is evidently problematic. According to Narayanan, the assistant professor of computer science at Princeton University, “We have a situation where these artificial-intelligence systems may be perpetuating historical patterns of bias that we might find socially unacceptable and which we might be trying to move away from.”
Programming AI systems without bias is sometimes overlooked by developers but it is an area that must be fully looked at. It is similar to mentoring a child, what we put in, molds them into who they are. While we can override code, we still need to be careful of asserting our own influences. Nevertheless, a report by Accenture in 2016 claimed that AI would boost labor productivity upwards of 40% by 2035 since it would change way work is done and reinforce the role of people to drive business growth.
AI is inevitable for the future, but we must not fear it. Instead, we must encourage its use to increase our efficiency and economic growth. Special attention has to be placed on research and development before it can be fully implemented. However, the human element must still exist because it is our own intelligence we are using to programme these systems and none of us are 100% perfect.Published: Feb 10,2019 11:55:00 AM IST