Rise of Artificial Intelligence
Artificial intelligence was first introduced in 1956 at Dartmouth University conference and soon gained momentum, with several efforts being undertaken worldwide. However, the interest in AI started to fizzle out slowly due to lack of sophisticated computers capable of processing enormous data. Hence, the AI segment lost interest as well as financial backing followed by a period of stillness, which continued from the mid-1970s to the late-1990s- an era which has been dubbed as ‘the AI Winter.’ This AI Winter continued until the interest in AI was rekindled. American companies became interested in AI once again and the world got AI-based technologies like the IBM’s Deep Blue and Watson. The world of AI has moved fast after this, with several of our gadgets -which we use on a day-to-day basis- now powered with AI-based functionalities.
AI-based technologies have proved their mettle and there is no slowing down in the adoption of AI solutions. In fact, scientists and developers say that what we now know about AI and its potential is just the tip of the iceberg and the technology’s full potential is yet to be explored.
Undoubtedly, the future of AI is bright and the golden period of the technology has just begun. It is increasingly used across sectors including manufacturing, retail, healthcare, finance, recruitment, education, transportation and even entertainment. We have AI-powered radiology equipment, autonomous cars, home automation system and so much more – all given to us by the virtue of AI.
The AI-powered systems that we are using now have seen a short time-to-market, unlike previous innovations which took decades to enter the market. A software dubbed Stats Monkey (now called Narrative Science) was developed back in 2009 by researchers and students at Northwestern University’s Intelligent Information Laboratory for automating sports writing. Six years later, Associated Press deployed an AI system in 2015 which delivers over 4,000 quarterly earnings stories, almost 15 times more when compared to human effort, according to AP’s stats.
The powers of AI are not unheard of and major companies including Microsoft, IBM, Facebook, Google and Amazon have invested significantly in AI, bolstering the sector further.
The sector is poised for growth and some factors that are driving its growth today are:
We have been using a lot of digital systems today, with every click, share, like, comment, swipe and tweet generating adding to the pile of data. Both offline and online activities are noted down and shared in the form of data, forming huge datasets that include structured and unstructured data from disparate sources. These datasets are now coming to use, with the advent of AI and its subsets- machine learning (ML) and deep learning (DL), which use the datasets for training their algorithms to perform tasks, intelligently.
Furthermore, the availability of cloud infrastructure has made access to data easy, with governments, educational institutes and businesses now storing huge datasets on cloud and accessing them easily to derive insights using AI, ML or DL technologies. They are also using these datasets to train new AI models, leading to the creation of next-generation AI solutions.
Next-generation Computing Architecture
Remember the time before the AI Winter we discussed earlier. Well, that time the computer, even the super-computers, were incapable of processing huge datasets to train computer vision and AI and that affected AI and its development tremendously. But with the recent advancement, we now have the powerful Graphics Processing Unit (GPU), which has found its application into the world of AI. Typically used in high-end gaming PCs and workstations, these GPUs are far more powerful than the CPUs and their use in the AI field have helped in powering and accelerating the algorithm training. GPUs are an essential component of every AI-powered solution now, ranging from virtual machines to consumer devices.
Another innovation that is helping the development of AI is the Field Programmable Gate Array or FPGA. These programmable processors could be customized to perform a specific type of task and are greatly used in niche computing tasks.
Bare metal server is another modern-day tech that is allowing the scientists and developers to run high-performance computing jobs in the cloud and is helping the growth of AI.
Deep Neural Networks
Neural networks form the core of deep learning and the advancement of Artificial Neural Networks (ANN) is boosting the growth of AI, as these ANNs, which are loosely based on the neural network of a human brain and their functions, help in delivering more accurate and enhanced throughput. These ANNs learn from the datasets to form their own patterns and algorithms to produce much enhanced output. They continuously learn, rectify and update their understanding to produce better output.
The benefits of AI are not unknown anymore and their usage in our day-to-day lives to make tasks simpler has increased our appetite for more advanced AI-based solutions.
According to a recent survey by the Yale University and the University of Oxford, AI researchers projected that machines will become better than us at translating languages by 2024 and they will be better at writing high-school essays by 2026, driving trucks by 2027, writing a book by 2049 and performing a surgery by 2053.
Well, the possibility of actual realization of these projections couldn’t be ruled out. However, the increasing penetration of AI into our personal and professional lives has also created fear. Fear of destruction by self-thinking AI systems and robots and fear of losing jobs to the AI-based machines.
The researcher community in divided into two cohorts, with one citing the concerns as real and looming and the other having a more optimistic views on the rise of AI.
Instances like the fatal accident by a driverless Uber car in March 2018 that killed a woman on a street in Tempe, Arizona, or the much-hyped conversation between Facebook AI Research-developed chatbots Alice and Bob, who were found chatting to each other in their own coded language, and the hysterical reaction to this episode are examples of how fearful we are of the power of AI.
However, several governments, companies and even data from certain reports have despised such negative notions around AI.
One of the reports by the World Economic Forum has indicated that usage of machines and algorithms at the workplace is expected to displace 75 million jobs while creating another 133 million new roles by 2022. This highlights how the job requirements will evolve with the rise of AI.
Moreover, organizations and governments are trying to develop new ‘Explainable AI’ solutions in a bid to allay the apprehensions that many AI users have regarding the functioning of AI and the process through which AI arrives at the output. Explainable AI will be capable of explaining their rationale and highlighting the understanding of the issues at hand and the process through which it arrived at the conclusion (output). This will not only allow the users to vet the process and determine the credibility of the output, but will also let them identify bias or shortcomings -if any-in the AI models and rectify the same.
The concept of Explainable AI is being backed by the US Defense Advanced Research Projects Agency.
All these efforts will certainly help in driving the growth of AI, while the fear of destruction of humankind by AI could be allayed to some extent by the projections of several noteworthy scientists and visionary leaders, including Amazon’s Jeff Bezos who are of the belief that a fully capable artificial general intelligence- which could actually cause self-initiated annihilation of human race- is far from being developed, and even if it is developed, extermination of human could not be its first and only agenda.