Analytics 3.0 is the process of creating algorithms for machine learning which allow computers systems to operate as a human. The reasons for building such systems are numerous: programming computers to perform tasks which humans can do would enable people to spend their time in more rewarding endeavors, freeing up the power of the human mind to tackle more complex tasks instead of those which can be performed by computers.
Also, automation of tedious or repetitive processes reduces the likelihood of human error, while also improving the speed at which those tasks are performed. In turn, this allows for greater productivity and improved quality of life.
The potential applications of machine learning may be numerous. A well-publicised potential benefit includes:
- The possibility of autonomous vehicles
- which promises advantages including reduced incidence of road traffic accidents caused by human error 
- reducedlabor costs of hiring workers to drive vehicles [2,3]
- increased mobility for disabled people, the elderly and children [4, 5, 6]
- a myriad of benefits related to ease and speed of transport
Indeed, the amount of time gained by busy workers not having to drive vehicles and instead using their time for other purposes such as work or relaxation could significantly increase productivity as well as improve quality of life
Medical research and practice
Another potential utility of machine learning is the application of machine learning processes to medical investigation and provision. Modern medical institutions generate substantial amounts of data for both research and clinical applications, and automation of the processes involved in collecting and acting upon such data may greatly facilitate the best possible medical provision .
With the advent of genotyping and DNA analysis as a standard process in medical research, making sense of the large quantities of data that are generated has become an ideal application for machine learning processes .
Climate change modeling
Climate change presents a significant number of threats to a large number of people across a wide variety of locations. Climate change modeling can help address this problem by providing accurate insights into the effects that changes in climate will have different geographical regions as well as the populations which exist within them.
However, accurate predictions of changes in climate must make use of a large dataset of climate information with a large number of variables. Machine learning can help to make sense of that data in real-time, allowing for accurate forecasts and the appropriate responses to be made .
With an ever-increasing range of threats to personal safety in the modern world, the application of machine learning to security devices and processes may allow governments and organizations to detect anomalies which would otherwise be missed by humans in a similar role .
The development of computer systems which can respond in real-time will help address security concerns more quickly and safely. Internet and credit card fraud detection are other potential applications of machine learning processes .
Natural language processing and linguistics
Natural language processing is another area where machine learning techniques can be utilized. Analysing the large datasets of spoken and written language is a task suitable for machine learning technologies, and development in this area could lead to computers which analyze spoken and written language more intelligently, and importantly, more humanely .
Hostile artificial intelligence
Despite the numerous potential benefits that machine learning provides, there are some limitations and implications which deserve some attention. First of these is the commonly-held belief in the possibility of artificial intelligence acting outside of the interests of humanity.
This belief, however, is based on the principle that machine can experience some consciousness, and with it, the ability to choose to act in a manner which is against the interest of humanity. This scenario is unlikely if we consider the machine to be entities which do what we program them to do. A more likely scenario is one in which the people who build this artificial intelligence do not fully account for all the variables that the machine is faced with, leading to unpredictable results.
This, in turn, would give an appearance of the device acting outside of the interests of the people who work with it. So, while the machine is working outside the benefit of those it serves, it is nonetheless doing what it is programmed to do. A solution to this issue may be in the recognition that the environment which the machine operates in is perpetually changing and that the device must adapt to this change, which consequently creates a need for additional input from the programmers of the computer.
An additional problem associated with the use of machine learning processes is the possibility of a centralized database concentrated in the hands of too few people. This, in theory, could lead to abuses of power.
Also, the concentration of datasets into the hands of a small number of organizations could result in monopolies, and potentially to prices which are inclusive of only those able to pay for the technology, such as high net worth individuals.
However, the solution has the opposite problem, in that if the data which machine learning algorithms rely on are free to be manipulated by a variety of courses, it would be tough to create standardized software solutions which work well together. This could lead to vulnerabilities in the software or otherwise reduce its effectiveness.
Responsibility for failure
While machine learning technologies such as autonomous vehicles promise a great deal, there are some serious reservations regarding who should take responsibility for any failures which occur. While reductions in the incidence of road traffic accidents are likely based on this technology, even a single safety failure could lead to significant legal implications.
It remains to be seen at what point a passenger of a vehicle becomes a driver. With the gradual implementation of autonomy that some organizations are incorporating, there needs to be a thorough discussion about the responsibilities of the driver and what components of the vehicles they are responsible for, as well as which ones they are not.
Also, the organizations which are developing this technology need to address the issue of who is to blame should one of the technologies which they are responsible for fails and leads to an accident.
- Autonomous Cars Could Save The US $1.3 Trillion Dollars A Year”. businessinsider.com. 12 September 2014. Retrieved 3 October 2014.
- Mui, Chunka (19 December 2013). “Will The Google Car Force A Choice Between Lives And Jobs?”. Forbes. Retrieved 19 December 2013.
- Cowen, Tyler (28 May 2011). “Can I See Your License, Registration, and C.P.U.?”. The New York Times.
- tenquist, Paul (2014-11-07). “In Self-Driving Cars, a Potential Lifeline for the Disable.” The New York Times. Retrieved 2016-10-29.
- Curry, David (2016-04-22). “Will elderly and disabled gain most from autonomous cars?”.ReadWrite. Retrieved 2016-10-29.
- James M. Anderson; Nidhi Kalra; Karlyn D. Stanley; Paul Sorensen; Constantine Samaras; Oluwatobi A. Oluwatola (2016). “Autonomous Vehicle Technology: A Guide for Policymakers.” RAND Corporation. Retrieved 2016-10-30.
- Yang, Yuedong; Gao, Jianzhao; Wang, Jihua; Heffernan, Rhys; Hanson, Jack; Paliwal, Kuldip; Zhou, Yaoqi. “Sixty-five years of the long march in protein secondary structure prediction: the final stretch?”. Briefings in Bioinformatics. doi:10.1093/bib/bbw129.
- Larrañaga, Pedro; Calvo, Borja; Santana, Roberto; Bielza, Concha; Galdiano, Josu; Inza, Iñaki; Lozano, José A.; Armañanzas, Rubén; Santafé, Guzmán. “Machine learning in bioinformatics.” Briefings in Bioinformatics: 86–112. doi:10.1093/bib/bbk007.
- Alpaydin, Ethem (2010). Introduction to Machine Learning. London: The MIT Press. ISBN 978-0-262-01243-0. Retrieved 4 February 2017.
- Sarikaya, Ruhi, Geoffrey E. Hinton, and AnoopDeoras. “Application of deep belief networks for natural language understanding.” IEEE/ACM Transactions on Audio, Speech and Language Processing (TASLP) 22.4 (2014): 778-784.
Analytics 3.0, machine learning algorithms, applications of machine learning, autonomous vehicles, machine learning, machine learning processes, automation, Climate change, Climate change modelling, Natural language processing, artificial intelligence, machine learning technologies