Artificial intelligence is becoming an integral part of digital strategies, but many CIOs are struggling to turn the technology into effective use cases.
That's according to CIO UK and Computerworld UK's 'Making Artificial Intelligence a Business Reality' research report, released at CIO UK's AI Summit at the May Fair Hotel in London on Thursday 28 February. The study revealed CIOs believe overwhelmingly that AI will augment human roles rather than replace jobs, but reported a lack of understanding about the business and ethical implications of AI at board level.
Although 65% of the respondents to the new survey said they are considering or piloting their first AI deployments, less than a fifth have already implemented AI projects or will do so in the next six months.
The peer-led study was directed by a steering group of CIOs and CTOs; committee member Nadine Thomson, a digital consultant currently working as Technology Director at News UK, provided a useful definition for the technolgy.
"True artificial intelligence is a simulation of human intelligence where machines are capable of intelligent behaviour," she says, adding that the applications of the concept are typically more limited applications.
"When we talk about AI today, we're really speaking about machine learning, where computers learn, make connections, self-correct and carry out tasks without being specifically programmed to do that task."
The steering group agreed that AI should not be treated as a separate strategy, as this would create a technology-led approach. Instead, they should start by analysing the problem and then consider whether AI could help solve it.
"At a practical level, start with the business strategy and business processes, and marry these back to what AI can offer today," says Rajat Dhawan, Chief Technology Officer at The Travel Corporation. "Start small, in an area that you have good control over and where you are struggling. Find out where AI will – and can – have a real impact."
The business benefits are primarily finding new revenues and increasing efficiencies by getting richer insights into customer behaviour and internal processes.
In the survey of 200 CIOs and IT executives from organisations of more than 250 employees, 54% of respondents said freeing up staff for higher value work was driving their investment. Another 44% said it was staying ahead of the competition, while 41% cited cutting costs, 31% said gaining a better understanding of their customers and 25% mentioned unearthing new business opportunities.
The key to achieving these objectives is to focus on the users.
"Don't make technology decisions; make decisions based on your users," says Chris Bradbury, Director of Product and Technology at Reed Online. "AI is not going to make your users or customers happy because you have it in your technology stack."
Building a team of data scientists offers businesses the best chance of success.
"One thing we didn't expect was the benefit of having data scientists on site to help find applications for AI," says John Gillespie, CIO at Amnesty International. "Having an approachable data scientist has allowed colleagues to raise ideas. Some of the best ideas have been cooked up in the canteen."
This helps address the skills shortage that 90% of survey respondents believed was a problem. Many of them felt that technical experts should be in charge of AI projects. Data scientists were named the best-placed specialists to lead the AI team by 60% of respondents, ahead of data engineers (cited by 52%), BI and data analysts (50%) and developers (4%).
The team that they employ to work on AI should have a varied set of skills and background.
"Team diversity should be a prerequisite success factor," says Inna Stelmukh, Senior IT Business Partner at Skanska. "You need a mix of technology and business experts, people with a different psychological profile, different backgrounds and fresh eyes."
This will also help avoid the algorithmic bias that is one of the biggest issues in AI, a technology that needs strong ethical standards to avoid dangerous consequences.
"AI ethics are very important; you need to consider an ethics committee," says Que Tran, Head of IT for Europe and Russia at DP World.
Internal staff can be supported by partnerships with external organisations that specialise in AI, which will help them quickly adapt to developments.
"The landscape is changing quickly, and it is very difficult to maintain internal talent to the level of specialist expertise available outside the organisation," says Thomson.
Google (68%) and Microsoft (59%) were named by survey respondents as the vendors that provide the top AI solutions, followed by Amazon Web Services (44%), IBM (35%) and Salesforce (11%). Startups can also help, as well as academic institutes.
"Go talk to other organisations, in other sectors, and find out what they're really doing," recommends Aline Hayes, director of technology EMEA at Experian.
"Go talk to your local university and find out what they're teaching. And talk to sensible suppliers; Microsoft, for example, put me in touch with people in the construction industry I'd never have known about."
The board will also need to be supportive partners, which means CIOs need to improve their understanding of AI - something that 37% of respondents to the survey said they currently lack.
"Any good CTO worth their salt will be spending time continually coaching, informing and educating their fellow board members in technology to ensure that when you bring heavy technical decisions into the boardroom you don't end up with a cringe-worthy Zuckerberg vs Congress situation," says Bradbury.
Delivering business value
The steering committee members are already delivering business value from a variety of AI applications, from Thomson using the technology to analyse behaviour of customers and personalise their experience to Bradbury deploying machine and deep learning to help Reed users and customers find the best jobs and candidates.
Nearly 80% of our survey respondents agreed that AI will transform their organisation over the next three to five years.
Expectations for the technology are going through the roof, but CIOs need to remember that it won't be the solution for every problem.
"Think carefully about what problems you're trying to solve, then choose the technology," advises Bradbury. "If that technology is something under the AI umbrella, then good luck; you're in for a heck of a ride."