Will computers take over the Knowledge Work?

Advances and development in machine learning, computing power and big data may enable computers to be more intelligent and capable than human beings. Will computers take over the jobs of knowledge workers viz. doctors, lawyers, engineers, automobile drivers, analyst etc. in future?


It was a milestone in the field of computing and artificial intelligence when IBM’s Deep Blue, the chess
playing machine, defeated World Chess Champion, Garry Kasparov in May 1997. With a power of evaluating 200 million moves per second, never ever in history a computer with such a computing power had faced a chess champion. The winning of the computer in the chess match untapped a plethora of unimaginable possibilities that would be possible with high computing power, artificial intelligence and huge amount of data.

Since then, new developments have happened in the field of computing, machine learning and natural language speech processing coupled with the availability of huge amount of data not only from internet but as well as from sensors around us as Internet of Things (IoT) devices enter into our day-to-day life. All these developments throws up one important question: Will computers take over once perceived as only human-capable, “KNOWLEDGE WORK”?

KNOWLEDGE WORK BY COMPUTERS

IBM WATSON at Jeopardy!

In 2011, IBM Watson-an artificial intelligent computer system- won 1st prize in a highly anticipated quiz “Jeopardy!” against its former winners, Brad Putter and Ken Jinnings. Unlike any other normal quiz, the clues are provided in the form of answers and the participant has to answer framing a question and topics covered are vast ranging from literature, arts, history, business, sports, science, culture, etc. The clues maybe in form of puns, twisters, wordplays, etc. which are generally harder for a computer to analyse. The first person to press the buzzer gets to answer the question. The questions were posed in natural language to IBM Watson. Watson completely outperformed the former champions by being first to answer in majority of the questions.

Watson had a data of over 200 million pages in its 4 terabyte disk. Watson applied advanced algorithms of natural speech processing, information retrieval, machine learning, knowledge representation, automated reasoning and machine learning technologies for data mining.
 
Deep Blue defeating World Chess Champion in 1997 and Watson outperforming former champions in “Jeopardy!” point to a very likely role the computers may play in our lives in future. Both, Chess and Jeopardy, involve using cognition, reasoning, guessing, strategy and most importantly person’s own experience. Knowledge workers use learnings from their own as well as others’ past experience to gain expertise in their work. It is this capability that computers may develop over time with advanced algorithms of machine learning and huge amount of data that will enable them to do the “Knowledge Work” that once was thought to be possible only by humans.

 Knowledge Workers

The term “Knowledge Workers” was coined by Peter Drucker in 1959 to refer to workers “who developed and used knowledge at “workplace”. The word “Knowledge” here means the expertise that has been gained by the person through experience and uses that knowledge at work in the organisation. Doctors, engineers, architects, painter, musicians, lawyers, analysts, etc. are all knowledge workers. They use the specialized knowledge gained by them at workplace. The experience of these professionals is non-dispensable and cannot be substituted by machines. For example, doctor’s advice and prescription cannot be taken over by computers; artists use their creativity and imagination to paint which again cannot be easily imitated by computers.

How our future will be like?

Automation has already taken over the manufacturing and many transaction work (for example, ATM replaced bank tellers) leading to loss of jobs. But the advances, in turn, has created need of skilled/ knowledge workers in other domains; computer industry, healthcare, service, etc. Along the side with automation of physical jobs, the use of computers and growth of internet and IT industry has helped employees to improve their efficiency. Many of complex calculations, analyses, modelling, etc. (for eg., modelling of cars through CAD/ CAM) can be done very easily on easily on computers which if otherwise would have taken a number of days or even months. Simulation and testing of products on computers have helped manufacturers to reduce manufacturing costs, make virtually zero-defects product/ service and greatly improve efficiency. Thus, computers completely automated some of our physical jobs and in others, it has helped the employees and companies work efficiently.

Doctors, engineers, software developers, teachers, architects, lawyers, analysts, etc. use computer to work efficiently. It is this knowledge worker’s job that was once thought to be non-replaceable by computers. But as times are changing, the confluence of computing power, machine learning, big data and improvements in natural speech processing may take over the knowledge worker’s job!

Imagine the following scenario in 2025:
Biosensor on your body notices a change in your blood pressure and it alerts your “Virtual Physician”- a computer program on the cloud. As you do your daily chores, your vital signs are being monitored. The computerised physician uses these parameter values and refers to hundreds of thousands of journal cases, your family’s history of hypertension, your diet and exercise routines and vital signs of other men of your age and reaches the conclusion and suggests: “You don’t need drugs, but you do need to stop eating fast food and skipping the gym”. Imagine the time and all hassles you have been just saved from.

Consider next scenario where autonomous self-driving cars will run on streets and highways: 
Google’s driverless car has been running on the streets of California and Nevada for a couple of years and has logged more than 500000 miles with just one accident (that too was because of human error). Intelligent autonomous vehicles can be used to automate many of the tasks. One example could be a convoy of autonomous trucks on highway for transporting goods. Such processes would eliminate need of drivers.

Not just these, but, computers will take over many other knowledge work that is being thought to be only human-capable. Even now, some of the computer programs are already entering into mainstream of our life to help in work as a normal person would do. For example, sensors-fitted cricket bats, hockey sticks, boxing gloves or golf stick are already available. The values measured by sensors are transmitted to a computing device. The software can analyse theses values and generate insights detailing the weakness, strengths and other aspects of playing style of player. The insights generated by the computer program are more accurate and helpful than those given by a game coach. In future, these sensor-fitted devices will the sole coach of sports players. Computers will also render the jobs of teachers useless. Already, robots are teaching students in some of the schools In Japan. The robots are better able to deliver education by teaching at individual student pace and level.  

As of today, we already seeing the some forms in which computers are helping do the knowledge work. One of the best and most common examples of computer doing the knowledge work is Apple Siri and Google Now.

APPLE SIRI AND GOOGLE NOW

Apple Siri is an intelligent personal computer assistant available on iOS platform that takes commands in natural speech to answer questions, make recommendations and perform actions. In some cases, Apple Siri is known to return some brilliant, clever, humorous and intelligent answers that normally a person would not have answered. The application is the best example how computers are thinking in a better way than a normal human could have. Consider the following conversations between person and Siri:

Conversations:


































The above results show that how easy it is obtain information with Siri to questions such as “What is the average weight of male hypertension patients?” which require so much of research on internet if done manually. Also, other answers to questions like “Planes overhead?” are almost impossible to obtain if tried by a normal person. The above results also show that Siri is able to solve mathematical problems. And in response to another question, “Why fire trucks are red?”, Siri returned an absolutely brilliant and humorous though illogical answer to the question. Such answers would have required a lot of effort if it had to be done by person itself. Google Now is a similar application to Siri available on Android platforms.

Such cognitive complex analyses, subtle judgments and creative problem solving of the computer programs has been possible because of huge amount research gone into a computer science field called Machine Learning.

MACHINE LEARNING                                                                   

Machine Learning, a branch of artificial intelligence, deals with construction and study of systems that learn from data. The term “Machine Learning” was first coined by Arthur Samuel in 1959 as a “Field of study that gives computers the ability to learn without being explicitly programmed.”
A simple example of Machine Learning would be a computer program that gradually learns to distinguish between spam and non-spam emails from past experience of whether the user marked it as a spam or not. Another example would be an optical character recognition in which printed characters are recognized based on examples of previous printed characters. The machine learns itself from the past data/ experiences. As program gathers more and more data, it becomes better and better in its intended task from its experience.
The biggest success of Machine Learning and also the most importantly required criterion for a system to learn by itself comes from the wave of “Big Data” that World is experiencing in this age.

BIG DATA

According to IBM, we create 2.5 quintillion (1 quintillion=10^18) bytes of data every day. This data comes from various sources such as data from sensors present in industries and devices; social media sites; photos, videos and audio tracks; transaction records and cell phone GPS signals to name a few. This all is BIG DATA.
The phenomenal growth of internet has played a huge role in generation of big data. Some information on amount of data being generated on internet every minute are as follows:
·         48 hours of new video uploaded on YouTube.
·         204166667 email messages being sent by the user.
·         2 million searches on Google.
·         700000 shares on Facebook by users
·         $ 272020 being spent by consumers on web shopping and the list goes on.
Apart from internet, another major source of data has been possible through rapid adoption of Internet of Things (IoT) concept by the market forces.

Internet of Things (IoT)

An IoT device is an object embedded with a sensor to collect its intended parameter and has the ability to communicate the data either through wire or wirelessly, thus forming a node in the information network. The information transmitted by the sensor maybe uploaded in the internet by one of the nodes that is connected to the internet for access to the information from anywhere in the world.
IoT will significantly affect the process flow in every sector of business. Its applications in some of its sectors include:
Healthcare: A person’s body will be fitted with sensors that will monitor body’s critical parameters like beat rate, blood pressure, etc. Sensors transmit the measured values to the person’s doctor wirelessly on internet (for eg., through mobile phones). In case of emergency, it can automatically trigger an emergency call to the nearest hospital for ambulance. The remote monitoring of health parameters through sensors on patients body is one of the applications IoT.
Manufacturing: IoT is already making a huge impact in the manufacturing industry with wireless sensor modules to monitor parameters like temperature, pressure, etc. for worldwide monitoring and also tracking of inventories, percentage completion of products and improving efficiencies of the processes by gaining insights from data generated by the sensors.
Agriculture: Internet of Things (IoT) aided agriculture provides farmers with data of the farms such as temperature, moisture and conditions of the crop and soil. These parameters help farmers improve agricultural production and crop yield. One specific example of IoT for farming consists of a network of sensors that is spread out in farm. The sensors communicate independently to each other using ultra low power Zigbee modules, an ad hoc network technology. The coordinator node in Zigbee (master node) is connected to a GSM module that sends the data collected along with suggestions to farmer’s mobile through SMS.
The applications of IoT in various industries are immense and unimaginable. An estimated 30 billion devices will be connected to the internet by 2020 according to the ABI research. The growth of IoT market will contribute towards generating huge data which will help systems learn by itself (Machine Learning) which in turn open up possibilities of automation of knowledge work by computers.

Computing Power

Today, an iPhone 4 costs only $400 compared to $ 5 million that it cost for the fastest computer CDC 7600 of equal performance in 1975. Apple’s iPhone 5 is the world’s first phone to have 64-bit processor which makes it as fast as a desktop computer/ Laptop. Computing power has been growing exponentially doubling every two years. Size of the processor is also getting smaller and smaller. Further developments in field of computing in form increased speed, reduced cost and compact size will be paramount for the machines mine the huge amount of data (2.5 quintillion bytes/ data) that is being generated every day. As of 2013, World’s fastest processor, Cray XT5 Jaquar, can process upto 2.3 quadrillion (1quadrillion=10^24) calculations per second and costs around $ 3 million. The tremendous ability to process data at high speeds will determine the possibility of automation of knowledge work.

Economic Impact

The computers will completely change how knowledge work is done by making more efficient and fast. Currently, global employment costs are $33 trillion a year which would reach $41 trillion a year of which knowledge work employment costs would be $14 trillion on current trend. According to McKinsey Global Institute’s (MGI) report titled “Disruptive Technologies: Advances that will transform life”, automation of knowledge work will have an impact of estimated $5.2 trillion to $6.7 trillion impact on economy by 2025 through improved productivity.

A Final Word

Computers taking over many of knowledge work will raise a number of legal and ethical questions. Who will be responsible in case of inappropriate treatment prescribed to patients by computers? What will be the priority of an autonomous car in event of possible collisions: safety of the passenger or safety of pedestrians? Who will be liable in such situations? Undoubtedly, the computers will, one day, have processing power far exceeding the capacity of human brain. But, it is only for the society to decide the roles that computers will play in our life. 

0 comments:

Post a Comment

Speak your mind here.