By Nitish Deshpande
AI today is helping to improve efficiency, accelerating growth, and making better policies in areas such as agriculture and mining. AI is automating repetitive, standardized, and time-consuming tasks. When machines gather, and analyse data, the rate of human error diminishes. AI is complementing what humans can do. Humans are able to accelerate the time-to-discovery and shorten the timeline it takes to reach important milestones in areas such as healthcare and drug discovery. AI has the power to revolutionize the world we live in. But, you might have wondered why we need humans if AI can do everything a human can. Read on.
Science fiction is full of thinking robots and machines that have evolved into living and sentient beings. While the idea is entertaining, many consider the research and experiments are very ambitious and far-fetched. Humans are alive but machines are not. Still, there are those who believe artificially intelligent, self-aware computers will exist by the next 30 years. All that is needed to achieve this is to come up with a computer that is built with a pattern as that of the human brain. But, is it possible to emulate the human brain?
The human brain is a highly complex unit. It doubles and folds back upon itself to provide room for the neurons that process all the data that we absorb daily. There are a million neurons in the adult human brain each of which is connected to thousands of other such neurons that process data in a parallel method, making it possible to analyse and make decisions quickly. To replicate that complexity, computers would have to have crazy amounts of RAM which does not exist currently. Many believe it will be available in 10-15 years from now. Also, humans are able to take in complex data like colour, movement, sound, taste, smell, sense and process them simultaneously to arrive at an identification or to make a response. The computer processes each area individually, assimilates the data to identify the subject, then must still rely on pre-programmed data to respond. Human intelligence is better at multitasking, adapting, social interaction and self-awareness than AI. The general function of AI is optimization while that of human intelligence is innovation.
The process of thinking Involves Self Awareness. A computer has a bank of knowledge. It knows things, but does it know that it knows?
For example, think of it like a thermostat which senses, or takes in data, that tells it the heat in a room does not match the pre-programmed figure. It responds by analysing the data and adjusting the heating or cooling to bring the room to the accepted level. It knows that the temperatures must be synced, but does it know that it knows and does it know it is a thermostat? Humans rely on their instincts and upon past experience to decide upon a course of action.
In another example where an AI was modelled to predict whether the image is of a cat or not. The model predicted accurately most of the times when it was fed with coloured photos. But, when the same photos were changed by keeping the outline of the cat same but removing the details and filling it with a skin texture of elephant, the AI model wrongly predicted it as an elephant but the same picture when shown to a human to predict was correctly predicted as a cat. Why do you think this would happen? The AI behaves in a way that it is trained to and cannot adjust and learn easily like the human brain. AI relies on a lot of human input and even then, it gets a lot of things wrong. An accuracy of around 80% is celebrated in ML research. There is still a lot of room for improvement.
There is an important feature that the humans possess which is having emotions. Ray Kurzweil, who invented the text-to-speech synthesizer, believes that goal of machines having emotions can be reached within half a century. He believes that If brains function through transmission of electric impulses and data through the neurons, like algorithms, then emotions, intuition, empathy and even a sense of humour can be taught to a machine by emulating those algorithms. It is a matter of complexity. If that is true, then when we arrive at that level then machines will be sentient.
The debate about whether or not machines can be taught to reason and feel is, at times, heated. Do we know enough about how the brain functions to emulate it? Some say no. Still, researchers attempting to develop thinking computers are not likely to abandon their quests. So, the answer to the question “Can AI think?” can be answered only by Time to come!
By Nitish Deshpande
For those who are new to AI technology, Artificial Intelligence is intelligence demonstrated by machines, in contrast to the...
By Puneet Joshi
When digital marketers and customer service agents think of “chatbots,” they typically think of an automated text-based...