Beyond the buzzword: How 'AI' is losing its charm
Let me begin by saying that the potential of artificial intelligence is fantastic. There are issues that everyone is debating on, for example, whether AI will be the end of humans; but to be very honest, at this point, AI is used more as a buzzword to get attention rather than using it in an innovatively mind-blowing way.
This is evident across the world. A recent study by London’s MMC Ventures found that 40% of European startups claiming to use AI, did not really use it. The lack of understanding of AI is the major reason behind how some companies can get away with this.
So what is artificial intelligence exactly? There are two aspects: one is what stage of evolution is artificial intelligence at currently, while the other is what it can become with enough coding, data analysis, processing power and testing.
At this point, artificial intelligence is used in different stages of its evolution by companies. However, machine learning is one of the most commonly used aspects of AI today, from chatboxes to decision-making softwares.
Essentially, machine learning is a process by which computers learn from a large amount of data and predict outcomes. For example, when you go on Zomato chat, it automatically pulls up recent orders and asks you which one you need support with, it then asks you what the issues are based on whether it is an on-going or past order.
E-commerce platforms such as Amazon, noon, Namshi, and almost all others use machine learning as well to recommend products to customers.
It is also quite extensively used in the healthcare sector, especially in oncology. Machine learning is giving way to precision medicine by learning which medicines and treatments have worked for patients with different blood types, history, and such.
And yes, while they are undoubtedly a segment of artificial intelligence, they are essentially coded to predict data. To put it in extremely simple terms, it’s like excel but at a much much more advanced level, using millions of data points.
That’s not to say that machine learning isn’t fantastic, because it is. It can recognise images, voice, even grammar. But as you might see, it’s simply using the science of probability to come to these conclusions.
Now, coming to the hype around the words “artificial intelligence”, most people think about its potential future, which is its ability to make decisions based on data entered. Driverless cars are a perfect example of this. But as you might know, they are nowhere near ready to be implemented in the real world without supervision.
Which means that the time artificial intelligence is really interesting is when it takes actions based upon data that it automatically gathers. For instance, cameras today can predict temperatures and movement, but they do not do anything to help humans. When they notice fainting people or animals dumped on the streets, and alert emergency services, or even just automatically adjust the temperature in rooms based on how cold or hot people are feeling, is when AI is acting upon its potential and becomes actually helpful.
To reach that level of automation, AI needs massive amounts of data. But more than that, it needs freedom from human management. This opens up major debates, from trust to safety and everything in between. World leaders are still debating these issues.
However, the one thing for certain is that it will take time because the infrastructure and cybersecurity need to first be strongly in place for automation at that level to be allowed. But to develop that infrastructure, we need artificial intelligence at that level to be in practice, to show where the weaknesses lie. Somewhere, at some level, it needs to be put in practice to break this cycle of wait, and begin the cycle of AI progression.