Skip to main content

From ethics to apprenticeships: How AI will evolve in 2020

(Image credit: Image Credit: John Williams RUS / Shutterstock)

Technology advances slowly and incrementally. When a huge AI milestone is achieved – such as when Deep Blue became the first machine to beat a human chess champion in match play in 1997 – we praise something as a great leap forward that was the product of decades of incremental evolution and repeated failure.

However, the pace of progress is picking up. In only a few years, AI has gone mainstream, jumping from the laboratory and into our living rooms and offices. As the country enters a new decade, we spoke to several prominent academics, both one-to-one and in anonymised group sessions, for their thoughts on the impact the technology will have in the year ahead.

Technology advances slowly and incrementally. When a huge AI milestone is achieved – such as when Deep Blue became the first machine to beat a human chess champion in match play in 1997 – we praise something as a great leap forward that was the product of decades of incremental evolution and repeated failure.

AI won’t replace us, but we’ll trust it more

Automation and job losses are likely to crop up in any AI discussion. However, moving into 2020 it’s likely we’ll begin to appreciate AI for how it can enhance our work more than we fear it making us redundant. AI will change us definitely, but it won’t replace us.

Earlier this year, SAS hosted a Digital Challenge at its London offices for students from various universities, where they were tasked with applying data analytics to solve a business problem for an imaginary bicycle insurer. We also took the opportunity to get the views of some of the lecturers that also attended, to see what they were expecting from AI.

Mark Lycett, Professor of Information Management at Royal Holloway, University of London, said:

‘AI is certainly replacing repetitive and routine tasks, which is not necessarily a bad thing, but even in the field of AI, lots of the current skills in play remain human ones.

‘AI is pushing at some of these human skills but we’re a long way off where it will replace human intelligence. Machine learning, for example, is nearly always supervised and it’s about doing a very specific task very well.’

AI solutions still lack the emotional and contextual intelligence needed to be given free rein in decision-making. When faced with the simple reality of imperfect data sets, it’s not always guaranteed an AI will make the best, or even a fair decision. It depends on the quality of the analytical model and algorithms that support AI. Into next year and beyond, there will always be a human behind the machine to review its decisions.

What we will see, however, is people who work with AI beginning to trust its decisions more. Medical experts treating cancer, for example, are starting to trust the ability of computer vision solutions to spot the early signs of tumours over their own judgement. In this way, AI is able to liberate the doctor from the screening process while giving them the insight to more accurately prescribe treatment, ensuring people get the care they need.

As Anabel Gutiérrez, Senior Lecturer in Digital Marketing at University of Kent, remarked: ‘My view is optimistic in the sense that I believe AI will be designed to enhance what’s done by humans and improve their performance. We need to get that across to students and businesses – it’s not machines making decisions but us making decisions and deciding how machines and algorithms can support us in that process.’

Data education will take centre stage to address the AI skills gap

Data is the lifeblood of AI. It’s the raw material the technology uses to generate insights and make decisions. As Bhavini Desai, Head of Programme for MSc in Digital Marketing & Analytics, Regent’s University, commented: ‘Data is considered the gold dust of digital business. It has been proven that businesses that have used their inbound data efficiently have been able to sustain and succeed.’

However, to create value from these insights requires people who have a deep understanding of data along with the requisite analytics skills. Yet this is proving difficult to find. Only 18 per cent of digital leaders think today’s school leavers and graduates are entering the workforce with the necessary data skills and experience. 

Many industries are struggling to recruit. Anabel Gutiérrez spoke in particular about the skills challenge facing the marketing industry: ‘Marketing agencies told us they find it very difficult to recruit people to work in marketing who also understand data. Given how much marketing is now driven by data due to the digital footprint people leave behind, it’s vital that this gets addressed.’ 

A lack of data expertise could hold back the UK’s ability to emerge as an AI leader in the coming decade. That’s why we are likely to see an explosion in the number of data-focused courses across UK universities in the next few years. Many of these courses will combine data education with knowledge of a specific industry, such as Data & Marketing, preparing students to use the data skills they learn in a way that’s most valuable for their chosen sector.  

‘The benefit of combining analytics with other skills or disciplines is that individuals will have better context for the analytical findings,’ said Dr Gutierrez. ‘Without context it can be hard for people to understand how to apply analytics in a given scenario. This reflects what’s happening now in many organisations where they have data and technology and produce insights from that data, but they don’t always know what to do with those insights.’

Some academics also anticipated the roll-out of a nation-wide apprenticeship scheme focused on data and AI over the next few years. This would involve companies coming into schools to share their expertise, alongside inviting students into their organisations to get hands-on experience with the technology. Many saw this as the fastest and most beneficial solution given persistent problems with attracting data and AI experts into the teaching profession.

We’ll demand more ethical decisions from our AI solutions

As AI plays an ever-larger role in our lives and economy, ethical and legal scrutiny will only increase. Many academics are both surprised and impressed that, although AI ethics has been a topic of study for years, it has become so prominent in discussions about the technology today.

While regulators seem to be forever playing catch-up with technology companies, the ‘wild west’ era of AI legislation will start coming to an end in 2020. In particular, the gradual rollout of autonomous vehicles on public and private roads will oblige lawmakers to rule decisively on who’s to blame when AI goes wrong. 

Some academics believe that the challenges many governments have had with cryptocurrencies would also encourage them not to make the same mistakes with AI. By taking a soft-touch approach to crypto, it was felt that some countries have allowed the unregulated industry to balloon and be exploited for illicit purposes.

To combat the future misuse of AI technologies, many academics predict countries will begin collaborating on AI legislation and the setting of base legal standards. Alongside this would likely come a shared ethical framework to protect citizens from the impacts of algorithmic bias.   

Dr Gutiérrez stressed: ‘Humans must retain an understanding of the data and models being used, to ensure fairness in decision-making. We need to retain empathy and control to ensure we are taking care of individuals, as well as broader issues such as supporting society or helping the environment.’

Geoffrey Taylor, Academic Programme Manager, SAS UK & Ireland

Geoffrey Taylor is Academic Programme Manager at SAS UK & Ireland.