Current use of AI
Artificial Intelligence has to be defined when examining how best to maximise the industry as a whole. Rather than specifying the differences between artificial intelligence, machine learning, deep learning and so on, discussing them as a group provides a better framework for developing their establishment and continued growth. Examples of AI in use include Ocado’s use of AI to optimise their storage and distribution methods, communicating with computers in natural language, generating photo realistic images from scratch, and beating world champions in Chess and Go. It pervades everyday life, affecting everyone, not just relegated to specialised computer labs and the realms of science fiction.
As a tool, AI has the ability to improve the functioning of most digital operations, products, and services. It is able to increase the effectiveness of using data. In essence, it is often used to do the same job that a human could do, but the sheer amount of data that it can manipulate and analyse and the speed at which it can do it is well beyond the capacity for a human. But AI can also be used in different ways from human tasks, providing a different route to reach a conclusion. In this way it can complement human intelligence in tasks, rather than substitute for it, which offers more opportunities for innovation, and helps to dilute fears of automation.
Because of its presence in everyone’s daily lives and its ability to augment human activities, its global potential is massive, and it can help deliver economic and social benefits. Many existing industries, should they capitalise on it properly, will be able to increase their productivity thanks to AI, which can help foster innovation to create new products and services. Estimates for the global AI market range widely – it could be more than £30bn by 2024, or $15.7 trillion by 2030. While the numbers may differ, the message is clear: AI is huge for the entire world.
As one of the leading countries for AI, the UK needs to decide how best to take advantage of the technology, in order to ensure that it is not left behind. Another advantage the country has is its position related to the history AI’s development, with the British computer scientist Alan Turing being widely regarded as an AI pioneer, and his legacy means that public support for AI in the UK is high, while it is also currently regarded as a centre of expertise. Building on that legacy of history that garners public trust and confidence is essential for the UK to remain competitive, as industry outside of the country is investing in AI.
The UK Industry
By international standards, the UK technology sector is in a strong position, estimated to have a turnover of £170 billion in 2015. There are 1.64 million digital tech jobs, and the growth rate was more than double that of non-digital jobs between 2011 and 2015. The report notes cybersecurity as a prime example of an established digital sector that will improve through using AI, where it is able to identify, categorise, and analyse the cyber threats that companies face every day more efficiently than humans. But it will also be able to help defend against large attacks through its ability to work simultaneously across many devices and on different tasks, highlighting anomalous behaviours quicker, which can then be passed onto the human engineers. This is how AI can complement human intelligence, rather than supplant it.
The UK has a healthy mixture of startups and large tech companies developing and using AI, with many of the smaller startups being acquired by major global tech companies, as a system of these majors building their pool of expertise, a practice known as “acqui-hiring”. Prominent examples of this include DeepMind now being part of Google and Swiftkey being incorporated into Microsoft. It has been estimated that more than 200 startups and small businesses exist in the UK that are developing AI products. Some of these define themselves as AI specialists – such as Swiftkey and Ravn – some address specific areas of industry sectors such as Legaltech, an AI that assists lawyers in legal searches and document review.
Meanwhile, larger companies outside the tech sector have been experimenting with the use of AI to increase the efficiency of their operations, such as with Ocado’s distribution systems. Uptake of AI is easier in businesses that have already considerably digitised their services, especially if their capacity for data is well-organised. AI is particularly useful for companies to improve their management of resources, including both people and tie by creating better scheduling that maximises efficiency, and automating tasks such as management of pay.
Public sector organisations, such as HMRC, have also been using AI in a similar manner. AI is able to help these established industries and public sector agencies dramatically increase the speed of their operations, resulting in faster services for the customers. In the area of issuing government paperwork such as passports and visas, this increased speed is an obvious benefit. The UK government has already launched the Government Digital Service (GDS) to automate and process user comments on gov.uk while there are other plans for further experiments with AI for the government.
Access to Data
The ability of AI to operate on multiple tasks at the same time takes on a more significant role with the explosion of data that is now produced globally. The rise of the internet and mobile personal devices has driven much of this, and shows no signs of slowing down. Data economies have become influential factors in spurring changes to national and global economies. AI is nurtured by increased access to data, as machine learning is trained by being fed vast sums of data. Without relevant data in a sector, AI technologies are unable to develop fully; it also makes AI’s presence compulsory when the flow of data becomes so large that only AI can deal with the amount and the complexity. Companies may find it difficult to externally share data, due to privacy and security concerns, or even commercial advantage concerns. It doesn’t entirely come down to trust either, sometimes companies simply don’t have the expertise to form agreements over data sharing, which is why there have been a lot of development of AI that can assist in this area recently.
The Digital Economy Act 2017 has provisions for using data in the public interest, but there is still more to be done to maximise the use of data. The report puts its recommendation into three broad areas: making data more open, improving machine readability, and managing trust of sensitive data.
The access to sensitive data is in interesting issue, because if data cannot be released or accessed in a sector – which may be the case for some organisations for many valid reasons centring on security and privacy – then AI development is stifled in that area because it will not be able to be trained to assist that sector, and therefore is unable to be effectively applied. Healthcare is cited as an example of this, where AI could add enormous value but is hampered by a lack of access to data. Part of the problem with healthcare data is the personal nature of it, and public trust and confidence has to be retained in healthcare, while the cost of it also has to be kept low and sustainable.
In order to help open up access to data, the formation of Data Trusts by the government and industry is put forward as recommendation. In order to use data for AI, data holders and users would come together – on a case-by-case basis only – to agree terms that meet both parties. Such trusts would not be a legal entity, but a set of relationships in a repeatable framework. This ways, the parties know and understand their obligations, while being able to comfortably use and share data in a fair and legal way.
Supply of Skills
Artificial intelligence requires an advanced set of skills, meaning that to remain competitive in this industry, the UK must ensure that it is providing training of these skills in the country, as well as making itself an attractive destination for talent across the world to assist in meeting the industry requirements. Fostering both domestic and international talent is key to developing AI and becoming a global leader, because AI, as the coming together of two fields (computer science and maths/statistics/data science) that are themselves constrained by a lack of supply, the problem of a shortage of talent is particularly felt and needs to be tackled head on.
Foundational skills will be key for training people to fill support roles with AI, which will be required as well as the specialists in AI in order to full incorporate AI into many sectors. Such skills that need to be considered include data preparation and protection, explaining the functions of AI to customers or staff, or managing reporting and accountability of AI. Increasing the availability of foundational skills, such as further compulsory mathematics education, will allow this supply to flourish. In the shorter term plan for increasing the skills supply, master’s level and post-graduate level places will need to be increased to help supply the growing demand for AI specialists. University places are also well-placed to ensure that the UK’s AI specialists are the best from the country and internationally, as well as representing a diverse set of voices, creating opportunities for women and other underrepresented groups.
The Massive Online Open Course (MOOC) Introduction to Cybersecurity, developed by the Open University was completed by over 80,000 people, demonstrating that MOOCs are a good way of teaching basic digital skills and increasing awareness of the relevance and need for these skills. With the idea that we may need “skills retraining” with the rising wave of automation, MOOCs offer an excellent method to determining the best suite of skills for the individual due to their accessibility. It also illustrates how useful and relevant AI is, wherein might not have been considered before, so that it is easier to make the transition to incorporating AI into sectors and businesses.
Maximising AI Research
The report calls for the Alan Turing Institute to become the national institute for AI and data science, giving AI a central role in its focus, as well as collaborating with other national institutes around the world. It is also able to already play a significant role in helping to make the UK a serious player in AI research, working with UK universities to help both their supply of skills as well as improving their research capabilities, and enshrining it with national status and having it be the dedicated national centre will solidify that.
Creativity and innovation are vital for the field of AI, and they often come out of the university network. In order to properly commercialise these ideas, universities have their role to support and nurture the entrepreneurial taken among its students and researchers so that the UK is an excellent place of global renown for starting a business. To this end, the transfer and exploitation of intellectual property, and not just its protection, should be an aim for universities, so that they are fulfilling their aim of having a positive impact on society. Making policies and practices for the licensing of IP and forming spin-out companies as clear and accessible is the recommended action for universities.
Increasing Uptake of AI
Improving transparency over the use of data in particular is always of prime concern to every stakeholder, and many AI innovators are pushing for clearly stated ethical principles in data-driven activities. Google, DeepMind, Amazon, IBM, and Microsoft all founded the Partnership on Artificial Intelligence to Benefit People and Society to share and promote the application of ethical practices, while also publicly demonstrating how dedicated major companies are to that cause.
The difficulty in highlighting transparency in AI is that any single decision is hard to explain due to the process of machine learning. Neural nets work on a different architecture and learn iteratively so that it is not easy to provide an explanation of the factors for any one single decision. The more open with data that businesses and research centres are able to be then the better the explanations we will be able to generate – if the structures and mechanisms for data sharing aren’t in place, then it will be impossible to have accountability in AI. The report recommends that a framework for explaining processes, services, and decisions delivered by AI is developed by The Information Commissioner’s Office and the Alan Turing Institute.
Finally, demystifying AI is what will ultimately increase its uptake. It is currently perceived by many businesses in all sectors as being new, overly complex, out of reach, even not well understood. Because of this image, it is then a challenge for businesses to comprehend that AI could improve their productivity, and in what ways that could manifest. Understanding that AI already plays a role in everyone’s lives would continue to address this image problem, while also promoting the government as both a user and a customer can help instigate a bottom-up approach.
Governments could well their cues from accountancy and law firms that have successfully engaged with AI. Areas that it could be deployed in are to increase the flow of work, information, and analysis in services where AI has the ability to deal with the mountains of data, improved decision making in complex areas, and improving the efficiency, effectiveness and usability of services for users. This would have the result that AI is seen as both a powerful tool, and a common one, where businesses, governments, and individuals alike are coming into contact and using AI, and this is definitely an area the UK government should be placing its focus on.
.jpg)
.jpg)
.jpg)
.jpg)
.jpg)