Key skills required to become an AI expert
Source: Education Times
TNN | Posted on Saturday, September 18,2021 11:50
It is important for a learner to become an expert in a niche technology instead of being a jack of all technologies. Like every industry, Engineering education and technology is also market-driven in nature. Due to huge market share and human resource demand in many industrial sectors, top educational institutions across the world started offering undergraduate and postgraduate degrees with specialisation in technology such as Artificial Intelligence, Data Science, Cybersecurity etc. For example, Bennett University (BU) in Greater Noida just started offering 10 much-in-demand specialisations for a BTech degree in Computer Science Engineering. Among them, Artificial Intelligence (AI) is one of the most sought-after areas for most Engineering students and IT professionals across the world due to its wide applicability in different Engineering disciplines.
Always in demand
A massive migration of students, researchers and industry professionals to any Engineering field will eventually lead to overpopulation of engineers in a decade of time. This happened many times in the past for various Arts, Science, and Engineering fields where surplus availability of workers was evident. It also creates fears among learners at times who have chosen AI technology as their career choice. Luckily, the demand for AI is not going to shrink in the near future since its potential is yet to be explored in various ways. There are a few unique skills an individual can develop over a period of time to stand out from a large number of AI engineers in future.
Theoretical and practical aspects
Clearly, mere degrees and grades can never ensure the employability of any individual. Unlike many IT domains, AI mainly requires problem-solving and analytical skills. Hence an expert in AI requires knowledge in both theoretical and practical aspects of the field. The theoretical part of AI needs knowledge of different AI models, awareness of different research problems and their AI model or algorithmic solutions related to diverse data formats available as table, image, video, audio, text, and sensor signals.
The practical knowledge of AI requires hands-on experience in handling large scale data, mastering recent software tools or packages and finally building and improving the AI models. It is important to work on mini-projects as well as semester-long projects with different levels of complexity. As training large AI models require huge computation power, learners need to access online or on-campus labs with supercomputing facility. At Bennett University, we have a powerful NVIDIA supercomputer DGX V100 which can train multiple AI models to serve a large number of students and research scholars simultaneously.
Due to recent advancements in AI-related programming languages and software tools, large AI models can be built even using simple lines of code. Even though building an AI model needs minimal effort in coding, training them on Big Data in a high-performance computing infrastructure to achieve the desired performance in terms of accuracy or running speed requires a lot of effort. An AI engineer cannot treat the model as a ‘black box’ and try random steps to improve their performance where model training and improvement requires in-depth mathematical, analytical, and debugging skills.
Updating knowledge on recent AI models and tools available in the market becomes essential for those who wants to stay relevant in the AI industry. Even though being updated is a must-have skill in any IT industry, AI requires additional care on being updated as the intelligent software or apps used by customers go through solution updates very frequently. The main challenges encountered during the process of updating, include studying, consolidating, and finally making sense of a huge number of software tools, research solutions published regularly. As the programming and software tools for any AI domain go through major changes every year, it is recommended to spend more time on gaining analytical skills through experimentation rather than just focussing on a few software tools or packages and trying to master them. All these require decision-making while learning, like what to keep aside, what to study, and how much depth to study, by which the learner mainly needs to ‘learn how to learn’. Overall, excelling in AI requires wide awareness of AI technology and in-depth knowledge in the specific application domain of the learner’s or company’s interest.
Building a Greener Web: Six Ways to Put Your Website on an Emissions Diet
By Roberta Haseleu, Practice Lead Green Technology at Reply, Fiorenza Oppici, Live Reply, and Lars Trebing, Vanilla Reply
Most people are unaware or underestimate the impact of the IT sector on the environment. According to the BBC: “If we were to rather crudely divide the 1.7 billion tonnes of greenhouse gas emissions estimated to be produced in the manufacture and running of digital technologies between all internet users around the world, it would mean each of us is responsible for 414kg of carbon dioxide a year.” That’s equivalent to 4.7bn people charging their smartphone 50,000 times.
Every web page produces a carbon footprint that varies depending on its design and development. This must be more closely considered as building an energy efficient website also increases loading speeds which leads to better performance and user experience.
Following are six practical steps developers can take to reduce the environmental impact of their websites.
- Implement modularisation
With traditional websites that don’t rely on single page apps, each page and view of the site is saved in individual html files. The code only runs, and the data is only downloaded, for the page that the user is visiting, avoiding unnecessary requests. This reduces transmitted data volume and saves energy.
However, this principle is no longer the standard in modern web design which is dominated by single page apps which dynamically display all content to the user at runtime. This approach is easier and faster to code and more user-friendly but, without any precautions, it creates unnecessary overheads. In the worst case, accessing the homepage of a website may trigger the transmission of the entire code of the application, including parts that may not be needed.
Modularisation can help. By dividing the code of a website into different modules, i.e. coherent code sections, only the relevant code is referenced. Using modules offers distinct benefits: they keep the scope of the app clean and prevent ‘scope creeps’; they are loaded automatically after the page has been parsed but before the Document Object Model (DOM) is rendered; and, most importantly for green design, they facilitate ‘lazy loading’.
- Adopt lazy loading
The term lazy loading describes a strategy of only loading resources at the moment they are needed. This way, a large image at the bottom of the page will not be loaded unless the user scrolls down to that section.
If a website only consists of a routing module and an app module which contain all views, the site will become very heavy and slow at first load. Smart modularisation, breaking down the site into smaller parts, in combination with lazy loading can help to load only the relevant content when the user is viewing that part of the page.
However, this should not be exaggerated either as, in some instances, loading each resource only in the last moment while scrolling can annihilate performance gains and result in higher server and network loads. It’s important to find the right balance based on a good understanding of how the app will be used in real life (e.g. whether users will generally rather continue to the next page after a quick first glance, or scroll all the way down before moving on).
- Monitor build size
Pre-processors come with the possibility to prevent a build to complete if its files are bigger than a variable threshold. Limits can be set both for the main boot script as well as the single chunks of CSS to be no bigger than a specific byte size after compilation. Any build surpassing those thresholds fails with a warning.
If a build is suspiciously big, a web designer can inspect it and identify which module contributes the most, as well as all its interdependencies. This information allows the programmer to optimise the parts of the websites in question.
- Eliminate unused code
One potential reason for excessive build sizes can be dozens of configuration files and code meant for scenarios that are never needed. Despite never being executed, this code still takes up bandwidth, thereby consuming extra energy.
Unused parts can be found in own source code but also (and often to a greater extent) in external libraries used as dependencies. Luckily, a technique called ‘tree shaking’ can be used to analyse the code and mark which parts are not referenced by other portions of the code.
Modern pre-processors perform ‘tree shaking’ to identify unused code but also to exclude it automatically from the build. This allows them to package only those parts of the code that are needed at runtime – but only if the code is modularised.
- Choose external libraries wisely
One common approach to speed up the development process is by using external libraries. They provide ready-to-use utilities written and tested by other people. However, some of these libraries can be unexpectedly heavy and weigh your code down.
One popular example is Moment.js, a very versatile legacy library for handling international date formats and time zones. Unfortunately, it is also quite big in size. Most of all, it is neither very compatible with the typical TypeScript world nor is it modular. This way, also the best pre-processors cannot reduce the weight that it adds to the code by means of ‘tree shaking’.
- Optimise content
Designs can also be optimised by avoiding excessive use of images and video material. Massive use of animation gimmicks such as parallax scrolling also has a negative effect. Depending on the implementation, such animations can massively increase the CPU and GPU load on the client. To test this, consider running the website on a 5 to 10-year-old computer. If scrolling is not smooth and/or the fans jump to maximum speed, this is a very good indication of optimisation potential.
The amount of energy that a website consumes — and thus its carbon footprint — depends, among other factors, on the amount of data that needs to be transmitted to display the requested content to users. By leveraging the six outlined techniques above, web designers can ‘slim’ their websites and contribute to the creation of a more sustainable web whilst boosting performance and user experience in the process.
How schools can improve employee experiences amid the cost of living crisis
By Dr Angelina Sun, Workforce Management Solutions Director at WorkForce Software
It is a grim fact that the cost of living crisis is hitting some of our most crucial workers the hardest. With reports this month finding that a third of primary school teachers are struggling to afford essentials due to soaring prices. For example, the rise in fresh produce costs broke records at 12.1% in September, putting the impact of inflation into stark context. Similarly, researchers at the University of Portsmouth recently warned that the cost of living crisis is seeing schools lose teaching assistants to better-paid roles in supermarkets, which not only pay more but also offer other benefits such as flexible working hours.
It is no surprise that the National Education Union (NEU) has issued a fresh warning of strikes as teachers demand an above-inflation pay increase. As the Union warns, if poor pay and conditions continue, the trend of 1 in 3 teachers in England leaving within the first five years will only get worse. Similarly, it has been found that 1 in 8 college staff leave within a year.
There is no doubt that the question of teachers’ pay must be addressed, but it is also clear that there are broader barriers to recruitment and retention in the education sector. Across other industries, there is an increasing awareness and appreciation of the importance of delivering a rounded, positive employee experience. Beyond remuneration, this encompasses other aspects of work-life from flexible working to support for wellbeing. Education has struggled to adopt a similar approach, both for practical reasons (the nature of this work often prevents certain flexibilities from being offered) and because culturally, the profession has often been slow to adapt.
However, this must change. There are ways to allow educators access to similar elevated employee experiences enjoyed in other sectors, and these must be considered if vital teaching staff are not to be lost to other competitively placed professions.
Flexibility in practice
The nature of the school day may lead educators to assume that flexible approaches to working are off the table. As a recent report from the NASUWT-The Teachers Union found, more than half of teachers (52%) say that their school/college does not offer flexible working and is now calling for flexible working to be a day one right for all teachers. As its General Secretary, Dr Patrick Roach highlighted, “There is a great deal of evidence showing that some schools continue to believe that flexible working is not compatible with delivering high educational standards for pupils. This is patently false.”
There are ways to explore more flexible working arrangements and schedules that give teachers greater autonomy over their time. In fact, the government has shown support for flexible approaches, recently announcing investment of £750K in a ‘culture change programme’ to embed flexible working in schools and multi-academy trusts, such as offering compressed hours, adjusting start and end times based on personal needs, and removing the requirement to spend non-teaching hours at school. When teachers can complete planning, preparation and assessment tasks at home, and even run parent conferences online, there is no reason they cannot be offered the same kind of remote flexibility as others.
Schools also enjoy the rewards of flexibility – retaining experienced staff, recruiting from a broader pool, promoting wellbeing and improving work-life balance, which leads to more engaged, productive teams.
There are also options to apply flexibility to financial rewards. While schools’ hands may be tied when it comes to salary uplifts, there are ways they can support staff to manage finances more flexibly. For example, pay-on-demand gives educators access to wages when needed. Staff can access money as it is earned, meaning there is no waiting for payday. This can make managing personal finances a lot easier, alleviating some of the financial stress that teachers increasingly find themselves under.
Delivering such elevated employee experiences across a school or college workforce can be a challenge. However, there are successful examples of similarly dispersed, deskless workforces that have adopted intelligent workplace management technologies to help manage flexible working and pay-on-demand. Using cloud technologies educators can access streamlined, personalised digital processes that mirror the technology-enabled employee experiences now enjoyed across private sector roles.
For example, modern workforce management technologies that utilise sophisticated AI and machine learning, alongside user-friendly design, can reinforce better working practices for all staff, whether frontline in the classroom, or supporting staff in the back office.
As highlighted by the Shared Headship Network, “schools need to be smart about the use of technology to enable certain aspects of school life to be more flexible.”
Modelling effective communications
If flexible working is to become a workable, embedded culture in education, it must also be supported by a foundation of effective communications and dialogues between teachers, staff, managers and leaders. For deskless workforces, such as teachers and lecturers, maintaining communications can be a challenge, – even simple administrative tasks such as leave requests can become burdensome and protracted.
Automating routine management tasks can save time, streamline workloads and improve efficiency while also allowing managers and leaders to focus on supporting teachers with the real work. Workforce management solutions are an easy win in this regard. Not only that, but they can also help leaders take the temperature of their organisations in the moment. For example, ‘pulse surveys’ can pick up on work dissatisfaction very quickly. Achieving consistent two-way communications through technologies can give the individual employee a voice while empowering employers with the insights they need to address trends affecting their larger workforce groups.
Seeing the bigger picture
There is no single solution to the issues facing recruitment and retention of teachers. Technology is not the panacea, especially when teachers are rightly fighting for better pay. Teachers deserve salaries that reflect their vital role in society. However, pay is not the only issue facing the profession. Too often educators are looking at their workplace experience in comparison to private sector roles and seeing a whole host of tempting benefits they do not have access to.
Digital provision, innovation and modern workplace cultures that are increasingly embraced in the private sector are not often available to those in the public sector. Again, this serves to highlight inequity of experience beyond the figure on the pay packet.
It is time to use all methods we can to create educational workplaces that reward staff. From how we engage technology, to pay, to employee experiences, it is a holistic approach that will save our schools and ensure teaching talent is rewarded and secured.
Closing the digital expectation gap in higher education
Ben Murphy, Client Partner, Great State
Digital education has been a long talked about topic, trialled across universities and schools for many years, alongside traditional classroom learning. It was when the Covid pandemic hit that higher education in particular faced scrutiny and backlash on how digital services were run – both in terms of academia and student wellbeing support – and how effective this method of teaching really was. This raised concerns from both students and parents about value for money in terms of the quality of the university experience delivered, causing a crisis confidence in these institutions.
With digital experiences weaved throughout our daily lives – from food shopping to checking the live bus schedule all by the touch of a button – this generation of students have high expectations of what universities can deliver digitally. Every university has ‘gone digital’ to varying degrees, but some universities are simply not set up to deliver the high-quality digital services and experiences at the pace required. Yet expectations of service delivery and the wider campus experience are becoming increasingly important factors in students’ eyes.
The consequence of this is an emerging digital experience gap between what students expect and what universities offer. Some HE institutions are already making the necessary changes to ensure digital experiences for students are slick, effective and useful – some still have a long way to go. A recent report by Jisc found more than a third of students rated support for online learning to be average or worse. Universities must act now to develop high-quality, inclusive digital experiences which will not just satisfy, but serve the TikTok generation of today.
Using digital to support academic learning and mental wellbeing
Students are looking for more than just learning support when it comes to using digital platforms at university. Previous experiences of digital may have been capped at submitting an assignment through an online portal or joining a society page on a university forum. Now, this has moved on to being able to reserve library computers easily, access lecture notes online, book university accommodation, and schedule video calls with university support staff. These all require well-designed, user experience informed interfaces to function well and actually be of value to the students using them.
Students expect the digital experience to support them not just academically, but also socially and mentally. With our research showing that 90% of students believe a good digital experience can help them perform better academically, create a greater feeling of belonging, and help maintain mental wellbeing – digital services need to incorporate every aspect of the student experience.
For those studying within a hybrid model or even fully remotely, digital can offer a way to feel connected with peers through a digital community. It’s these social connections which play an important part in the university onboarding process in place of, or enhancing, the physical experience.
And this doesn’t have to stop at graduation. Digital services offer the opportunity for graduates to feel connected to the university and their alumni far beyond learning and into their working lives.
Leveraging data for digital experiences
There is a big opportunity to leverage the data provided by students in more intelligent ways. Collecting data from digital services provides universities with an opportunity to learn more about students’ mindsets and behaviour, allowing them to tailor and drive the student experience towards positive outcomes. Data patterns can establish what makes each student tick, what learning style works best for them, and what their motivations are, to allow the university to provide better individual support in the best way possible.
Data can also provide an indication about the utilisation of services based upon segments of the student population within the university. Data implementation doesn’t need to be large, intimidating or have an organisation-wide impact – it can begin with delivering pockets of personalised content or communications, centralising some data and making small steps towards a single customer/student view.
Closing the digital gap
Given it takes two to three years to implement a digital strategy across a large organisation, complacent HE organisations may not act fast enough to close the gap as student patience runs out and expectations increase.
Universities that fail to act now may be left behind, creating an opportunity for challenger universities – such as former polytechnics and newer institutions – to close the gap or overtake ‘premier league’ institutions within the Russell Group, for example. Challenger universities have had to work hard to attract and retain students. Getting the digital experience right could provide the USP they’ve been searching for and put them ahead of the game.
Universities must understand the specific needs of their student cohorts by identifying where student needs are not currently being met, and the areas of the experience they most value, as these might not be the ones they expect. Whilst most of the student population are digitally native, it should not be assumed that they all are. There are important differences in the way that students from differing socio-economic and ethnic backgrounds have access to and use digital, and universities must consider all of these demographics when designing digital solutions to serve all.
Embracing “customer-centred” design thinking and service design principles to make sure that services are designed around students – rather than what is perceived as best practice and to suit the traditions of the organisation – is crucial to closing the digital experience gap.