These days, Artificial Intelligence (AI) has become the hottest topic in the tech industry. Some even argue that AI has snowballed into an industry itself and is ready to become one of those industries that permeate into every sphere of our lives – socio-economical, civil, and political. While the big-ticket items like Machine Learning (ML) or Deep Learning (DL), tend to hold a lot of attention and are the four words that everyone knows about, Natural Language Processing (NLP) has steadily become a greater part of our lives without as much of a splash.
In the simplest definition, an NLP is a branch within AI that focuses on giving computers the ability to understand text and/or speech similar to how humans can. While AI as a concept has been around since the early 1900s, and was championed by mathematicians like Alan Turing and Noam Chomsky, a lack of processing power and limited thinking prevented meaningful progress at the time. This problem plagued NLP especially. As our ability to process numbers faster increased, existing NLP models which were built around handwritten rules were made obsolete by the development of ML algorithms and Neural Networks that were able to finally bring the intelligence in the AI aspect of NLP.
As computer scientists gained access to more data and resources, NLP began infiltrating our lives. Grounded in feed-forward neural networks, tech companies began employing NLP to add small features to their products such as text-to-speech, speech-to-text, typing predictions, and spellcheck. In 2011, NLP took the biggest step forward in recent history with the reveal of Siri, a new automated assistant that anyone could ask a question to, and would then respond. This product, which was now available to consumers worldwide, was a wildfire and was suddenly the talk of the town. But despite all the advancements in the early 2000s, it was still very grounded in a Voice-Command structure that relied on the algorithm looking for yes/no responses and was able to perform direct actions.
As the tech industry and demand for smart gadgets grew, home assistants like Google Home and Alexa started to become commonplace. Becoming an integrated member of the household, the tech industry, in a wonderful moment of collaboration, began creating home appliances and electronics such as washing machines that were compatible with the assistants. The successes of these virtual assistants spurred on the leaders in this industry, leading to the creation of sophisticated bots like the Google Duplex, revealed in 2018.
Google Duplex took chatbots to a whole new level, using a natural-sounding human voice instead of a monotonic and robotic one (as per industry norm at the time). It was marketed as being able to make appointments, confirm timings, confirm capacity, and much more, essentially taking over the traditional calling duties that a receptionist would have. As with any disruptive technology, it has taken time for it to be adopted. In 2021, Google Duplex is finally being used across the US and has since been removed as a Google Pixel-only feature.
This raises significant concerns about the quality of service, the impact this will have on the workforce, and begs the question: just how ethical such technology is. The industry has been in the press constantly for mishaps with user data – Google and Facebook being the two companies whose business model revolves around collecting and using user data. There are already concerns about how virtual home assistants essentially give these companies a backdoor into private conversations and there is still great doubt on the simplest of things like whether when we switch off an assistant, it really does stop recording our data. Duplex, however, kicks into overdrive, as now we are reaching technology that can mimic human behaviour and potentially come close to passing the Turing Test. In short, the Turing Test is a method of inquiry that Turing designed to check whether computers are capable of thinking like humans.
This is yet another way for companies to collect data on us, and this one perhaps more sinister than others, given that we could be under the presumption that we are having a direct conversation with a human, when in fact it could be a highly sophisticated chatbot at the other end. In the pursuit of advanced technology, we are rapidly reaching a point where the negative implications that such pervasive and lurking technology needs serious consideration, and possibly regulation.
Author: Kunal