By K. Ananth Krishnan, executive vice president and chief technology officer at Tata Consultancy Services (TCS)
Like many innovations, artificial intelligence first came into public view as science fiction. Mary Shelley's "Frankenstein", Karel Capek's "Universal Robots of Rossum" (where the word "robot" came from) and other works conceived of artificial beings with the ability to think like humans.
Today, the exponential growth of computing power is bringing these and other imaginations to life: from cars that drive themselves to smart assistants, self-learning networks and dozens of other less glamorous, but no less important applications. And we are still in the early days.
Artificial intelligence is the next wave of technology transformation, and no business executive can afford to ignore it. According to Gartner, smart agents will facilitate 40% of mobile interactions by 2020. The market for cognitive and AI solutions will show a compound annual growth rate of 55.1% between 2016 and 2020, according to IDC, from almost 8 billion dollars in revenue to 47 billion. In the Tata Consultancy Services (TCS) Global Trends Study on Artificial Intelligence, more than 150 of the largest corporations spent an average of $ 150 million each on AI initiatives in 2015.
However, it can be difficult to have a solid understanding of what exactly AI is and how it can best help your business. In addition to being technologically complex, AI, like many of the products it generates, is evolving and changing rapidly. It is, at the same time, much more and much less than any of the utopian promises or dystopian visions that usually surround it.
It is vital that companies are lucid and strategic when navigating this new scenario. Senior executives are being bombarded - from the IT room to the boardroom - with questions and demands on how they are following the topic and how they plan to move forward. It is easy to get into the AI wave and feel compelled to act simply because the competition is doing it, but you cannot prepare companies for the future with just one stroke.
Here we offer a practical and realistic framework for how to think about AI:
1. AI is a great idea. You have to think small.
It is easy to get stuck in the big visions of a company transformed by AI. For most companies, we are far from that reality. Executives should have a very thorough focus on identifying specific problems or needs that would benefit most from AI's ability, and not think about whether it would be possible to replace half of their workforce with bots, for example.
In today's business world, the most successful applications of AI are often those aimed at solving difficult but very mundane problems with a significant return on investment. Consider, for example, the classification of a support ticket. By allowing massive processing of data from multiple sources, AI can transform data from support tickets into invaluable information for business intelligence.
It also helps to think through the basic objective: is it improving productivity in the company? Has customer service improved? Or is it expanding its market and sales potential? A single AI solution will not meet all of these needs
2. The algorithm is the brain. Focus on the data, which is the blood that irrigates the brain.
With AI's ability to relate and analyze unstructured data stores (think of direct customer feedback, plus social media comments), the emphasis should be on identifying and capturing the best possible data from all relevant sources.
Companies have always collected data through a variety of applications, such as CRM, Business Intelligence and now social media. This will only grow as organizational boundaries merge and the ecosystem becomes smarter with sensors, smart factories and cities, connected devices and more.
The good news is that there are methods and techniques available that can convincingly store this data, analyze it and build forecasting models around it that will learn and improve. The better the data, the better the results.
3. AI consists of machines. Think of people.
Successful AI applications do not necessarily replace human labor with machines, but increase and improve collaboration between the two.
Yes, certain tasks done by people now, such as low-level customer interaction, can be automated through bots or other AI applications. According to Forrester, 25% of all tasks will be aimed at software robots, physical robots or customer self-service automation by 2019. However, the same study states that 13.6 million jobs will be created using AI tools over the next decade.
Maximizing the capabilities of AI creates a need for more knowledge creation and organization by people. This knowledge, in turn, feeds the machines to help teach them. The ability of a machine to have human-like cognition requires huge amounts of data and training. For example, it is not enough to provide an autonomous car with a route map. A successful computational model would include as much information as possible about how a human being understands a route, including knowledge about curbs, traffic lights, likely obstacles during different weather conditions, details of the road surface and much more.
For the car's system to have the ability to "reason" and make appropriate decisions as circumstances arise and errors are corrected, it depends on receiving continuous streams of data. All this "learning" requires a different relationship between machine and man, which goes from being a mere operator or administrator to becoming a continuous teacher. This, in turn, has profound implications for the workforce and the way it is distributed.
4. AI is growing at breakneck speed. Put your foot on the brake.
While intoxicating, the elixir of mass data analysis can raise ethical issues and risks for your customers and your business. Do not put these questions aside, or leave them for other people. The academy is already in the early stages of developing structures that can help. Researchers at Carnegie Mellon University have developed a Quantitative Influence Index. This index can analyze the weight given to a set of factors when reaching a machine decision. For example, the index can reveal the weight given to age and income when making a loan decision. Such structures can make the functioning of the AI system much more transparent, accountable and ethical. This development could facilitate change management and the acceptance of AI systems in companies.
5. It is difficult to quantify the ROI of AI. Do not worry about this.
You may not know the numerical value of your AI initiatives for some time. It requires continuous attention and calibration. Create other markers of progress, success and failure, including the ability to do new things with intelligence and an anticipated analysis that AI can provide.
Today, the exponential growth of computing power is bringing these and other imaginations to life: from cars that drive themselves to smart assistants, self-learning networks and dozens of other less glamorous, but no less important applications. And we are still in the early days.
Artificial intelligence is the next wave of technology transformation, and no business executive can afford to ignore it. According to Gartner, smart agents will facilitate 40% of mobile interactions by 2020. The market for cognitive and AI solutions will show a compound annual growth rate of 55.1% between 2016 and 2020, according to IDC, from almost 8 billion dollars in revenue to 47 billion. In the Tata Consultancy Services (TCS) Global Trends Study on Artificial Intelligence, more than 150 of the largest corporations spent an average of $ 150 million each on AI initiatives in 2015.
However, it can be difficult to have a solid understanding of what exactly AI is and how it can best help your business. In addition to being technologically complex, AI, like many of the products it generates, is evolving and changing rapidly. It is, at the same time, much more and much less than any of the utopian promises or dystopian visions that usually surround it.
It is vital that companies are lucid and strategic when navigating this new scenario. Senior executives are being bombarded - from the IT room to the boardroom - with questions and demands on how they are following the topic and how they plan to move forward. It is easy to get into the AI wave and feel compelled to act simply because the competition is doing it, but you cannot prepare companies for the future with just one stroke.
Here we offer a practical and realistic framework for how to think about AI:
1. AI is a great idea. You have to think small.
It is easy to get stuck in the big visions of a company transformed by AI. For most companies, we are far from that reality. Executives should have a very thorough focus on identifying specific problems or needs that would benefit most from AI's ability, and not think about whether it would be possible to replace half of their workforce with bots, for example.
In today's business world, the most successful applications of AI are often those aimed at solving difficult but very mundane problems with a significant return on investment. Consider, for example, the classification of a support ticket. By allowing massive processing of data from multiple sources, AI can transform data from support tickets into invaluable information for business intelligence.
It also helps to think through the basic objective: is it improving productivity in the company? Has customer service improved? Or is it expanding its market and sales potential? A single AI solution will not meet all of these needs
2. The algorithm is the brain. Focus on the data, which is the blood that irrigates the brain.
With AI's ability to relate and analyze unstructured data stores (think of direct customer feedback, plus social media comments), the emphasis should be on identifying and capturing the best possible data from all relevant sources.
Companies have always collected data through a variety of applications, such as CRM, Business Intelligence and now social media. This will only grow as organizational boundaries merge and the ecosystem becomes smarter with sensors, smart factories and cities, connected devices and more.
The good news is that there are methods and techniques available that can convincingly store this data, analyze it and build forecasting models around it that will learn and improve. The better the data, the better the results.
3. AI consists of machines. Think of people.
Successful AI applications do not necessarily replace human labor with machines, but increase and improve collaboration between the two.
Yes, certain tasks done by people now, such as low-level customer interaction, can be automated through bots or other AI applications. According to Forrester, 25% of all tasks will be aimed at software robots, physical robots or customer self-service automation by 2019. However, the same study states that 13.6 million jobs will be created using AI tools over the next decade.
Maximizing the capabilities of AI creates a need for more knowledge creation and organization by people. This knowledge, in turn, feeds the machines to help teach them. The ability of a machine to have human-like cognition requires huge amounts of data and training. For example, it is not enough to provide an autonomous car with a route map. A successful computational model would include as much information as possible about how a human being understands a route, including knowledge about curbs, traffic lights, likely obstacles during different weather conditions, details of the road surface and much more.
For the car's system to have the ability to "reason" and make appropriate decisions as circumstances arise and errors are corrected, it depends on receiving continuous streams of data. All this "learning" requires a different relationship between machine and man, which goes from being a mere operator or administrator to becoming a continuous teacher. This, in turn, has profound implications for the workforce and the way it is distributed.
4. AI is growing at breakneck speed. Put your foot on the brake.
While intoxicating, the elixir of mass data analysis can raise ethical issues and risks for your customers and your business. Do not put these questions aside, or leave them for other people. The academy is already in the early stages of developing structures that can help. Researchers at Carnegie Mellon University have developed a Quantitative Influence Index. This index can analyze the weight given to a set of factors when reaching a machine decision. For example, the index can reveal the weight given to age and income when making a loan decision. Such structures can make the functioning of the AI system much more transparent, accountable and ethical. This development could facilitate change management and the acceptance of AI systems in companies.
5. It is difficult to quantify the ROI of AI. Do not worry about this.
You may not know the numerical value of your AI initiatives for some time. It requires continuous attention and calibration. Create other markers of progress, success and failure, including the ability to do new things with intelligence and an anticipated analysis that AI can provide.
EN 
PT 
