Artificial Intelligence has now cemented itself as a common feature not only within business and technology sectors, but also in our everyday existence. Whether it’s in the form of smart email categorisation or a ride-sharing app like Uber, it seems that our lives are now seamlessly intertwined with our technological counterparts.
This is a concern for some; Elon Musk famously remarked that AI was ‘a fundamental, existential risk for human civilisation’. While this belief is perhaps more in keeping with the stories portrayed by Science Fiction, it’s a fear that has been echoed by many, especially those who believe that AI could begin to put people out of work. This is a very real threat for those who work ‘repetitive’ jobs, such as factory operators or cashiers.
Overall though, AI is not the terrifying risk that some would lead us to believe. Well, not yet anyway. As technology is still considered to be within its earliest development stages, there’s no real risk of AI becoming too clever for its own good right now- if anything, we still have a long way to go before technology catches up with human cognition. So with this in mind, when does Artificial Intelligence fall short in today’s current climate, and what can be done to improve machine learning?
It still needs supervision
One of the major flaws of current AI is that it cannot always recognise and process certain actions in the same way as our own cognitive skills. As such, we need to remember to work cohesively with technology rather placing too much confidence in AI, as this is when problems begin to occur.
For example, when Uber’s self-driving car caused an accident that resulted in a death of a pedestrian, it was found that the AI driving the car had failed to acknowledge that it should have made certain choices when presented with specific data. In addition, the person overseeing the AI (the ‘safety driver’) hadn’t been paying enough attention, so both machine and person were at fault.
After the incident, Uber decided to replace its safety drivers with ‘mission specialists’ who would have more specialised training in working with AI cars. This emphasises the need for technology and people to work together, i.e. ensuring specialist human monitoring becomes an integral part in overseeing Artificial Intelligence systems.
Accidents like this inevitably reaffirm a natural distrust of AI. It cements the idea that Artificial Intelligence is a threat, rather than a form of technology that relies upon human interaction. Until we stop thinking of AI as a superpower and start recognising the importance of our responsibility in its development, AI will fail to live up to its full potential.
Issues with Customer Service
In the current climate, Artificial Intelligence is not at a stage to manage as a bespoke solution to businesses. It’s something that’s certainly worth considering further down the line, but right now it’s just not viable for wide-scale practical use.
One way for businesses to take advantage of technology in the meantime is through automation. Many companies have started to make use of bots/chatbots to streamline their customer service process, allowing their staff to utilise their skills and free them from particularly repetitive tasks. Automation used in conjunction with customer service specialists can help companies reduce operation costs, and save time. However, this isn’t always a guarantee overall, as automations can often break or work insufficiently.
It’s also worth considering automation from a customer perspective. If you are dealing with an issue with your bank or utility provider, would you prefer to deal with a person or an automated service? It’s now becoming so common for brands to favour automation in an effort to save money that customer service is becoming neglected as a consequence. Current trends demonstrate that consumers are fast becoming fed up with dealing with bots, and instead prefer a human touch when dealing with complaints or issues.
It’s therefore important for businesses to remember that while automations are helpful in streamlining processes, personalisation and human interaction are crucial for making customers feel valued and establishing brand loyalty.
Machines are learning, but cannot understand
Artificial Intelligence by today’s standards is merely known as ‘narrow AI’; this means machines are designed and trained to perform a menial task through processing of data. However, these tasks are not always carried out seamlessly by AI, as the numerous examples show.
Gmail now has a brand new ‘Smart Reply’ feature that suggests replies to your emails. While this sounds like it could save time in the long run, the sad reality is that the AI cannot understand why certain responses aren’t always appropriate. It has tried to suggest ‘I love you’ as a response to many different types of emails, including work emails, purely because the machine has recognised it as a common phrase. AI can process data and identify a common response, but it doesn’t understand what these responses actually mean.
For Artificial Intelligence to be successful, it not only needs to accurately interpret patterns and feedback on correct and incorrect outcomes, but also understand the social and emotional implications of its decisions. This is currently impossible with narrow AI. Technology would need to develop further into Artificial General Intelligence, i.e. the ability to think and process information the way humans do, for this to be a possibility.
Given the current limits of technology, the likelihood of machines establishing their own cognitive function is a long way off from now. And if this particular scenario does play out, there will be obvious risks and dilemmas if machines can begin to think for themselves.
What are your thoughts on AI? Do you feel confident that Artificial Intelligence can be improved and used efficiently? We’d love to hear your thoughts on this one- leave us a comment below!