Machine Learning vs AI: What's the difference?
                                                                            By Priya (Editor)

Artificial intelligence (AI) is the simulation of human intelligence processes by machines to complete tasks usually undertaken by humans. These machines are programmed to mimic and learn the actions of humans, and learn with experience to think and act rationally and humanely.

There are 3 types of AI, with each of the 3 different entities built for a specific purpose:

• Artificial Narrow Intelligence (ANI) - This is the only kind of AI that actually exists today. These Artificial intelligence systems specialise in one area to solve one specific problem. Examples of this type of AI can be seen through smart assistants that I'm sure we all use on a daily basis, such as Siri, Alexa, Google Assistant and Cortana. These smart assistants are artificial intelligences that uses advanced machine learning technologies to function.
  
• Artificial General Intelligence (AGI) - AI which has a human level capacity of mental skills, including perception, attention, memory, decision making, and language comprehension. Unfortunately, this entity of AI is merely a theoretical concept, and is rather challenging to develop given the resources we currently have access to.
• Artificial Super Intelligence (ASI) - Although this seems rather far fetched, an Artificial Super Intelligence (ASI) system would be able to surpass all human capabilities, such as making rational decisions and building emotional relationships. Once we see the development of AGI, AI systems would rapidly be able to improve their capabilities, with the jump from AGI to ASI being possibly as short as a nanosecond, due to the speed Artificial Intelligence would learn!
  

Diagram to show the stages and future of AI

Machine Learning (ML) is a sub-domain of Artificial Intelligence that teaches a machine how to make inferences and decisions based on past experiences. It is a method of data analysis that identifies patterns to reach a possible conclusion without having to involve human experience.

Here are some applications of Machine Learning:

1. Drug Discovery - By now we have all become aware of Pfizer, the pharmaceutical company that has helped to develop one of the most commonly adminstered vaccines against COVID-19 today. Pfizer is using IBM Watson, a supercomputer that involves artificial intelligence (AI), on its immuno-oncology (a technique that uses body’s immune system to help fight cancer) research.

2. Machine Learning in Retail - The moment you start browsing for items on Amazon, you see recommendations for products you are interested in as “Customers Who Bought this Product Also Bought” and “Customers who viewed this product also viewed”, as well specific tailored product recommendation on the home page, and through email. Amazon uses Artificial Neural Networks machine learning algorithm to generate these recommendations for you.

3. Conversational chatbots - These chatbots are popularly called virtual assistants. Such chatbots are contextually ‘aware’ and improve over time as they leverage NLP, ML, and NLU (Natural language understanding) to learn from user inputs. Apple’s Siri, Amazon’s Alexa, and Google’s virtual assistant are examples of conversational chatbots.

Overall, it is clear that artificial intelligence and machine learning are quite different, but together they can have an incredible effect on the future of medicine, retail and the functiionality of the future.

AI and its Development of Medicine                By Anya Year 10

Artificial Intelligence (AI) is the ability of a machine to perform tasks which are typically performed by intelligent beings such as humans. The most well-known examples of AI among the general public include virtual assistants and self-driving cars. Machine Learning (ML) is a ‘subset’ of AI, and it involves the development of computer systems that can learn and adapt without specific instructions to follow; instead, statistics and algorithms are analysed for patterns in data. For example, ML is often used in image and speech recognition. The complexity of these algorithms is progressing rapidly and therefore they are being used increasingly to further human capability in a variety of fields, one of these being scientific research and development. Generally, these uses involve simulations and calculations based on data which has been gathered. This article will provide a view of how AI and ML are influencing and improving one of the ways we evolve medical science.

The use of ML is often seen in forming accurate medical diagnoses. For example, when taking MRI scans of a patient’s body (e.g., at the heart or lungs), the MRI must be done quickly, to prevent the scan from producing a blurred image as a result of unwanted movement of the patient’s body. However, this rapid capture can result in compromised image quality and imperfect data sets of the scan, therefore rendering it useless for accurate diagnosis of a condition which may be difficult to identify.

This is where image reconstruction becomes useful: a process which utilises ML to create better quality images. Traditional methods of ‘analytic’ reconstruction require many accurate measurements of the patient, whereas ‘iterative’ reconstruction (the version involving AI) creates multiple ‘guesses’ of the correct image and, through comparing data sets, can arrive at a conclusion. The system is not perfect with regards to efficiency (compared to how fast it has the potential to be), but it can produce images quickly which scientists can analyse and utilise in the future.

This leads to medical image analysis. It is hoped that the algorithms that the machines contain will become developed enough to identify pathological features such as neurological conditions which are much more difficult to spot otherwise. By comparing images and datasets to find unique patterns, AI can help to generate conclusions which can help to save lives in the future. This can also automate the medical diagnosis process, as well as improve its accuracy in identifying problems in complex structures. There are many other advantages to this method of diagnosis: namely the fact that it is a non-intrusive method of diagnosing diseases like cancer very early on, for example, because computers can help us to better thoroughly distinguish between cancerous and healthy tissue in ways which are not immediately obvious, like the texture of the tissue.

Despite all its uses and its enhancing effect on research, machine learning has many flaws. One researcher, Ali Rahimi, has gone so far as to calling it a version of alchemy because it’s unpredictable and, to an extent, it’s a constant guessing game for data scientists reasoning why one algorithm may work over another. Since much of the work done involving AI is experimental, ideas about and the practicality of complicated algorithms are constantly advancing but are also a highly debated topic, because one of the only ways faults can be eradicated and the code understood is through extremely rigorous testing procedures such as sliced analysis. Overall, machine learning creates significant difficulties in research fields because results are often not reproducible (meaning they’re often inaccurate) and scientifically valid, but the next generation of ML models are being developed to minimise these issues. The holistic analysis which ML provides for analysing large data sets is still greatly advantageous.


References:
https://physicsworld.com/a/a-machine-learning-revolution/
https://www.science.org/content/article/ai-researchers-allege-machine-learning-alchemy
https://towardsdatascience.com/the-machine-learning-crisis-in-scientific-research-91e61691ae76
https://futurehealthcare.software/2021/08/10/automated-medical-image-analysis-using-ai-the-why-the-how-and-the-what/

---

‘Women In Tech’ Series – Dr Nuria Oliver PhD

What do the following societal challenges have in common?
 
-          When will the next Covid wave hit us?
-          What are the triggers for mental health?
-          Which area of Africa will Malaria affect next?
-          How can visually impaired scientists understand complex data sets  using music?
 
Actually, they have two things in common. Firstly, that technology is a critical tool in answering them. The second is that in this edition’s ‘Woman In Tech’, Dr Nuria Oliver has helped solved all of these problems using techniques in ‘Big Data’, Computational Science, Machine Learning and Artificial Intelligence (AI)!
 
Nuria Oliver is a formidable computer scientist and a renowned expert in AI. She has a PhD from MIT and is a world expert on computational models of human behaviour, human computer-interaction, intelligent user interfaces, and mobile computing. She has published over 150 scientific papers and book chapters that have been cited over 10,400 times, and has received numerous awards including most recently from the King of Spain.

Not only is Nuria named as an inventor in 41 patents, but she is also on a personal mission to use her skills to address some of the big issues faced by citizens, government and businesses. Importantly, Nuria is also passionate about ensuring that the world of technology is explained in simple, everyday language so that young people (especially girls) are drawn into this fascinating topic - so that they too can one day make a difference to the communities that they live in.

Take a look here at some of Nuria’s amazing research papers.
 
I was fortunate enough to spend some valuable time in discussion with Nuria where she touches on what inspired her to move into the field of Computational Science and AI; and she also goes on to explain some of the work she recently led (such as the predicting the spread of Covid in the early days of the pandemic) and she offers some brilliant advice to those who are considering technology as a career.

Machine Learning in Radiology: The Third Eye in Diagnostic Imaging Interpretation                                                By Shriyaa Year 11 

Interpretation of radiological images has always been the forte of radiologists for more than a century. Machine learning, however, has become the disruptor of this monopoly since the last decade. Its remit spans the entire spectrum; from requesting an image to acquisition to interpretation. The question: Is machine learning a third eye which augments the viability or perhaps threatens the existence of radiologists?
 
Machine learning is nothing but a method of data science that provides computers with the ability to learn without being programmed with explicit rules. In radiology, the computer is taught the nuances of X-ray images and their relevance by repetitive learning. Once an algorithm develops sufficient confidence to interpret an image, it is then fed new images to identify the abnormality in comparison to a human eye. The advances in machine learning and artificial intelligence have increased the accuracy several folds in the past few years. It is no longer in the fringes of medicine, but a vital armamentarium for a radiologist to augment the specificity (false positive) and sensitivity (true negative) of diagnostic interpretation.
 
Table 1: Clinical Applications of Machine Learning in Radiology

Machine Learning, for a start, has advanced substantially within the realm of healthcare. Currently, it is utilised by both computer science experts and clinicians in order to transform the way healthcare is being delivered to the population. The widespread implementation of Machine Learning holds the potential to increase the value of current diagnostic tools and imaging techniques such as x-rays and computed tomography (CT) scans by increasing imaging quality to aid more accurate diagnoses and thus drastically improving the quality of patient care and safety.

The above image illustrates the utilisation of medical imaging feature extraction. One of the vast applications of Machine Learning in radiology is the automated identification of pneumothorax - a condition where air collects between the lungs and the chest cavity. The incorporation of algorithms alongside featuring of relevant findings in picture archiving and communication system (PACS), a medical imaging technology that provides convenient access to scans and images, demonstrates one of the various methods by which machine learning can be efficiently integrated into the field of radiology.
 
Whilst Machine Learning has enabled data scientists and clinicians to harness its extraordinary potential, one should be aware of medicolegal paradox of machine learning when used in clinical decision making. Radiologists, for example, take ownership of medical diagnoses and treatments provided by clinicians based upon their own image interpretation. On the other hand, computers cannot be held accountable; they continually learn and relearn interpreting X-rays in ways that are still not fully understood. Due to these fallacies, machine learning cannot be expected to replace radiologists. Instead, machine learning techniques are expected to aid the expertise of the radiologists and improve their diagnostic efficiency – a third eye to identify patterns and associations that may normally evade the human eye.
 

References

Chang, Y., Paul, A.K., Kim, N., Baek, J.H., Choi, Y.J., Ha, E.J., Lee, K.D., Lee, H.S., Shin, D. and Kim, N., 2016. Computer‐aided diagnosis for classifying benign versus malignant thyroid nodules based on ultrasound images: A comparison with radiologist‐based assessments. Medical physics, 43(1), pp.554-567.
 
Do, S., Salvaggio, K., Gupta, S., Kalra, M., Ali, N.U. and Pien, H., 2012. Automated quantification of pneumothorax in CT. Computational and Mathematical methods in Medicine, 2012.
 
Setio, A.A.A., Ciompi, F., Litjens, G., Gerke, P., Jacobs, C., Van Riel, S.J., Wille, M.M.W., Naqibullah, M., Sánchez, C.I. and Van Ginneken, B., 2016. Pulmonary nodule detection in CT images: false positive reduction using multi-view convolutional networks. IEEE transactions on medical imaging, 35(5), pp.1160-1169.
 
Stamm, J.A., Korzick, K.A., Beech, K. and Wood, K.E., 2016. Medical malpractice: reform for today's patients and clinicians. The American journal of medicine, 129(1), pp.20-25.
 
Wang, S. and Summers, R.M., 2012. Machine learning and radiology. Medical image analysis, 16(5), pp.933-951.
 
Zeng, J.Y., Ye, H.H., Yang, S.X., Jin, R.C., Huang, Q.L., Wei, Y.C., Huang, S.G., Wang, B.Q., Ye, J.Z. and Qin, J.Y., 2015. Clinical application of a novel computer-aided detection system based on three-dimensional CT images on pulmonary nodule. International journal of clinical and experimental medicine, 8(9), p.16077.

AI and Mechanical Learning                                 By Mahi Year 9

Microsoft Teams, the online platform used during lockdown at NLCS, as well as across the globe

The benefits and disadvantages of AI and Machine Learning, focusing on how Microsoft Teams was used in Lockdown.

During March and June 2020 and from November to early December 2020, England experienced national lockdowns. This resulted in the introduction of “online school”, with many schools worldwide using 3 main programmes: Microsoft Teams, Google Classroom and Zoom. Microsoft reported that more than 230,000 educational institutions were using their programme “teams” for remote or hybrid learning in the wake of the Covid-19 Pandemic. NLCS employed Microsoft Teams and all of us have experienced and seen the disadvantages of teams and online learning first-hand as a result of the lockdown and the Covid-19 Pandemic. The coronavirus pandemic has forced students and educators across all levels of education to rapidly adapt to online learning. The impact of this — and the developments required to make it work — could permanently change how education is delivered. While online and distance learning has been used before to maintain continuity in education, such as in the aftermath of earthquakes, the scale of the current crisis is unprecedented. Speculation has now also begun about what the lasting effects of this will be and what education may look like in the post-COVID era. For some, an immediate retreat to the traditions of the physical classroom is required. But for others, the forced shift to online education is a moment of change and a time to reimagine how education could be delivered.

The benefits of Microsoft Teams has been widely seen, especially first-hand by students. Within Teams, educators can quickly converse with students, share files and websites, create a OneNote Class Notebook, and distribute and grade assignments among other things. Furthermore, OneNote Class Notebooks and assignment management allow educators to organise interactive lessons and provide feedback. Many students were stuck abroad due to the first or second lockdown. Online School allowed them to participate in class, and did not prevent them from learning or allow them to fall behind in academic terms. What I mean by this, is that Microsoft Teams is accessible from across the globe, allowing more connectivity, faster. Examples of the benefits of Microsoft Teams have been seen throughout this year, in which some foreign speakers have been able to give talks to NLCS Students through teams, without the hustles of international travelling, something which wouldn’t be possible had we never been through lockdown.

Although Microsoft Teams has its benefits, it also has its disadvantages. Many offices and schools still use Microsoft Teams as an alternative use to school or office work. Even throughout the post-lockdown time, many schools have not fully reverted back to their old ways. Furthermore, AI and Machine Learning is not perfect - The technology market has a tendency to crash. It can break down at any time and can be hacked or exploited. This can cause delays and problems for many people, and presents a problem towards the prospect of AI and Machine Learning. Online School can also make it hard to interact, as many students have psychologically adapted to learning in person. Various further educational innovations have occurred to make the universal adoption of remote learning a possibility. One key challenge is access. The lack of Internet connectivity in some locations, especially rural ones, and the competing needs among family members for the use of homework technology. However, creative solutions have been found. For example, school buses have been used to provide hotspots, and consequently improve network connections, and class packets have been sent by mail and instructional presentations aired on public broadcasting stations. The period of Coronavirus has also seen increased adoption of electronic resources and activities that can now be integrated into online learning experiences.

As the Coronavirus pandemic has been relatively recent, The effects of AI and Machine Learning are still being defined. The concept of AI Learning Machinery is experimental and many now use it as an alternative to set homework or send out notifications to a particular class, and NLCS still widely uses Microsoft teams to distribute work. The human race had to experiment and try to adapt to this new work environment in response to the Covid-19 pandemic, which included technological and administrative systems for implementing online learning, and the structure that supports its access and delivery. Before the pandemic, the primary purpose of online education was providing access to instruction for those otherwise unable to participate in a traditional, place-based academic programme. As its purpose has shifted, its audience, as well as the wider learning spectrum, has changed. It will be interesting to see which aspects of emergency remote teaching remain in the next generation of education, when the threat of the pandemic is no longer a factor. The flexibility and learning possibilities that have emerged from necessity are likely to shift the normality of the way in which students and teachers learn and communicate.

---

Here are some links relating to"AI and Machine Learning" - be sure to check them out!

https://www.bbc.co.uk/news/av/technology-59954359 
A smart spoon and home security tech

BBC Click’s Lara Lewington takes a look at some ingenious, yet rather uncoventional, innovations from this year's annual CES convention, the world's favourite and most renowned event that's a hub for technologies of every possible nature imaginable. Want to see some of this tech for yourself? Watch this short video to see her testing a few of these devices for us, from a machine that condenses plastic bags together for efficient reusing, to a spoon that claims to make food taste "better"... To see these technologies for yourself (and more!) click the link or the picture to head to the BBC website.

Click the link (or the picture above!) to head to the video
Source – BBC News

 2. techhq.com/2022/04/what-happens-when-ai-and-ml-turn-rogue-data-poisoning/
Ever heard of data poisoning...?

AI will bring massive benefits to society. However, criminals also use AI to cause cyber-security risks to businesses and their customers. Every AI needs lots of ‘training’ data from which it continually learns, so that eventually the AI can perform complex tasks. However, some hackers have found a way to ‘poison’ this training data such that the AI could end up learning to do some rather strange things. Click the picture to read more!

3. https://interestingengineering.com/7-ways-ai-will-help-humanity-not-harm-it
AI: Here to help humanity

As per the example above more and more by businesses, governments as well as ordinary people have growing concerns that AI could be harmful. Hopefully if you read this interesting article you’ll be much more assured!

4. https://open.spotify.com/show/1VXZAMyo0nzrW2neDtk6fa
The Metaverse has arrived!

Have you heard of the metaverse? The metaverse is a digital reality that combines aspects of social media, online gaming, augmented reality (AR), virtual reality (VR), and cryptocurrencies to allow users to interact virtually. Check out this short podcast trailer for an excellent series called ‘Welcome to the Metaverse’. If the iPhone and social media totally changed the internet 15 years ago, in less than 2 minutes, you will get to hear how our use of the internet will completely change - all over again.

5. https://screenrant.com/scariest-techno-horror-movies/
Prepare to be scAIred!

How do you like a good horror movie? If you do, (and you obviously also love all things tech!), then check out the 10 most scary techno-horror movies...

Thank you so much for reading this issue, which was focused on "AI and Machine Learning" - and thanks again to those that contributed. The articles were all fantastic! Next term, the theme will be Space Tech, so I hope lots of you will get involved. If you want to contribute, (or if you have any questions about the topics mentioned in this issue), please email me ([email protected]), but I will be sending out an email to the school soon about the next edition.

If you have time, I would really appreciate it if you could fill in the small survey below. It won't take long, and I would love to see your suggestions on the different topics I can include and the areas I need to work on in future editions.

Thank you!