Home
About the Team Team Profile
Ideal Jobs Industry Data
Interview 1 Interview 2
Autonomous Vehicles Blockchain Cybersecurity Machine Learning
Project Idea Tools
Group Reflections Personal Reflections

Machine Learning



Machine learning has transformed the world, from automated transport and taking over dangerous jobs to environmental protection, improved elder care, and much more. This essay provides an overview of Machine learning. It starts with a synopsis of where Machine Learning started and how it relates to Computer Science. It then looks at state-of-the-art technologies and innovations used in the development of Machine learning, followed by a summation of the socioeconomic impacts on society, as well as a personal reflection on how Machine learning affects my immediate environment. Finally, the work concludes, Machine learning is not only the way of the future but could spur on the Fourth Industrial Revolution.


Machine learning is how a machine learns and improves through experience and the analysis of data sets. The way in which researchers approached the development of Machine learning was to take core elements of human learning and transfer that to machines. The various modes of learning that they transferred to machines are "feature detection, unsupervised learning, deep learning, supervised learning and teaching," Faul (2020, p.5&6). When looking at where Machine learning sits in the subject area of Computer Science, we see it as "a branch of artificial intelligence (AI) and computer science which focuses on the use of data and algorithms to imitate the way humans learn, gradually impr`oving its accuracy" (IBM Cloud Education, 2020, para. 1). In other words, it is "the capability of a machine to imitate intelligent human behaviour" (Brown, 2021, para. 8).


Figure 1: Euler Diagram showing how AI, Machine Learning, Neural Nets and Deep Learning interact by Yulia Gavrilova, 2020.

Figure 1: Euler Diagram showing how AI, Machine Learning, Neural Nets and Deep Learning interact by Yulia Gavrilova, 2020.


Innovations are continuing to push Machine Learning to new frontiers, and each one of these state-of-the-art technologies that we will be investigating, builds upon a previous generation of knowledge, laying the groundwork for future revolutions in this field. One example of recent innovation is the GPT-2 language model which was recently superseded by GPT-3. This language model can operate independently with only a small amount of input from a human user and has been launched by the company OpenAI, whose mission it is to ensure that artificial general intelligence (AGI) benefits all of humanity (OpenAI, 2015). Though it is an open-sourced project, researchers are limiting what the public can access due to the possibility of using it malevolently. The latest version GPT-3 has been particularly noted by Dale (2021, p.113) for its ability to "produce apparently novel text that reads as if it was written by a human." This could revolutionise how creative works are written, it could largely take over the role of content writers, it could produce code that is useable. It can even write very realistic news articles (Dialani, P 2020).


Further innovation from MIT-IBM's Watson AI lab includes the development of "quantum algorithms" which would facilitate Machine learning on quantum computers. This is of importance, as it would allow quantum computing on "highly complex data structures" far beyond what we can currently do with a "conventional computer"(Temme, K & Gambetta, J, 2019).


In the world of hardware, the development of the tensor processing unit (TPU) by Google to accelerate machine learning workloads, (Cloud TPU, n.d.) a TPU can process linear algebra equations and be hosted on the cloud which allows people all over the world to access it (The Research Nest, 2020).


Machine learning is developing at an astonishing rate and has enabled a plethora of innovations. Common applications in Machine learning include machine translation such as Google Translate and Microsoft Translate. Image recognition software that can serve purposes such as high-end security for online transactions (Analytics Insight, 2021), preventing, disrupting, and investigating crimes (NSW Police Force and Facial recognition, n.d.) or smartphone applications such as the Google Lens, which allows you to translate information in real-time, copy and paste directly from your smart phone to computer, identify plants and animals or help you with your homework. (Google Lens, n.d.). Further uses are self-driving cars and self-flying drones, which use a variety of cameras, sensors, algorithms, and image recognition to allow these autonomous vehicles to operate (Kharawal, 2020).


Further advances in this sphere that are likely to eventuate are developers using quantum processors to develop apps "with a quantum advantage" as said by Ferris (2019, para. 18). Trustworthy AI projects are working on a system that allows Machine Learning projects to actually make fair and smart decisions by building trust into the AI systems (Ferris, 2019). Additionally, as hardware becomes more sophisticated, improvements in the miniaturisation of electronics are made, and improved deep learning and Machine Learning algorithms are researched; we should see further developments with Machine Learning Technologies such as Generative AI, which can create content with data input from things such as text, audio, or images. We will likely see further use of decentralised data to enhance Machine Learning iterations to streamline queries, medical diagnostics, and device collaboration (Preetipadama, 2020). We may see a revolution in the role of 'Tiny Machine Learning' (TinyML) and the function it serves in linking devices on the Internet of Things (IoT), (Springboard, n.d.). The results of this collaboration between TinyML and IoT can see many current issues resolved, such as enhancing our data security and privacy, more secure data infrastructure, lower costs in regard to data streaming and faster inference (Springboard, n.d.).


When looking at how Machine learning is changing the world, we need to consider its likely impact on our life and society. A valid area of concern for many is the mass unemployment caused by automation of jobs. In fact, according to Petropoulos & Bruegel (2018, p.120) the jobs that are most at risk are "middle-level jobs that require routine manual and cognitive skills." There are also concerns that, as Machine learning improves and becomes more sophisticated, the "inequality between the 1% and the 99% may widen as workforce automation continues" (Williams-Grut, 2016). The challenge will be to retrain people currently working in a lower-skilled profession and upskilling them for more technical jobs that require a much higher skill level.


Additional societal impacts could be in trying to ensure that the technology remains fair and unbiased. The nature of Machine learning is that the machine is constantly 'learning' from vast amounts of data and models. If a model were to display an inaccurate view of the world from the data it is processing, it may lead to further inequalities of marginalised groups in society. An example of this could be that it may do racial profiling based on speech (Kelly, 2021).


On the other hand, research also shows that though there is the potential for mass unemployment and the potential for exacerbated inequalities, there is also the possibility of a future where people work cohesively with machines and AI technology. In the World Economic Forums future of jobs report 2020, it was found that though Machine learning and AI is taking over millions of jobs, it is also creating millions more (Caine & Firth-Butterfield, 2020). Moreover, there were also discussions regarding how AI can be part of the solution in upskilling employees to prepare them for the future of work and their part in the Fourth Industrial Revolution (Caine & Firth-Butterfield, 2020).


It is not only in our jobs that Machine learning technology will have a profound impact; it will impact everything from digital infrastructure to our cities and communities via 'digital twin' technology. (Batty, 2018). A 'digital twin' is linked to a real-world object and can be utilised to provide a 'virtual environment' for "infrastructure planning, management, construction, manufacturing and even healthcare" (NSW Digital, n.d.). What we will probably see more widespread in the future is the use of this 'digital twin' technology coupled with AI advancements to build 'smart cities. The aim of these 'smart cities' is to develop better systems for "sustainable living, increased comfort and productivity for citizens" (Syed et al. 2021).


When considering the impacts of this technology in my daily life, there are many instances where mundane and repetitive tasks may become automated. Things such as doing taxes, book-keeping and even transport may all be affected by the developments in Machine Learning. As a student, machine learning may help to create a tailored study plan based on my "academic history" and "attendance" (Sinnott, 2018). It could even bridge the gap in areas of my learning where I need a fuller understanding and could aid the learning processes as well as streamlining aspects of the tutors' job (Sinnott, 2018). How I interact with my health care provider could change with the use of sophisticated analysis and cross-referencing, it would change the patient-doctor relationship to a patient-doctor-AI relationship. Driverless busses would change the way my friends and family travel within cities and might make the roads safer. Researchers at Curtin University found that "90% of all car accidents are caused by human error. Driverless technology can help to reduce the number of accidents on our road," (Curtin University, n.d.).


Family members who are unable to drive as they age; will be able to take advantage of the advancements in driverless technology to retain their autonomy. We may see retirement homes become a thing of the past and have family members able to retire in their own homes with the assistance of IoT. Autonomous alert systems that could let a caregiver know if an abnormality has occurred, (Srinivasan et al., 2020), in-home robots to assist with tasks such as cleaning and mobility (Elite Data Science, n.d.)


In conclusion, this essay started with a definition of Machine learning, followed by an outline of various state-of-the-art applications and a look at potential future developments of this technology. It also considered the impact on society as well as a personal reflection on how it would affect my life and the lives of those around me. As technology progresses Machine Learning will become further ingrained into our daily lives whether that be at work, at home, with friends and family or alone. It is the technology of the future that will hurtle us forward into the Fourth Industrial Revolution. Whether that is positive or negative will largely depend on how we as responsible citizens of the world use this technology for the betterment of humanity.



References: