AI full of possibilities with the right tools and understanding

Consumer identity is just one of many areas where artificial intelligence (AI) can be helpful in analysing vast amounts of data from millions of sources to gain business insights, according to Hans-Peter Gϋllich, CEO and founder of AI-assisted risk management firm Dydon.

“AI presents many possibilities and opportunities in areas such as cyber security and compliance, but businesses need to understand the limitations of the technology so they know what it can and can’t do for them,” he told Computer Weekly.

Another thing that is important for businesses to understand, said Gϋllich, is that AI technologies typically complement human activities rather than replace them.

“This perception that AI technologies will lead to unemployment is incorrect because in most cases is it more likely to be an effective support tool that will make work life easier.”

That said, Gϋllich emphasises that AI-enabled systems are typically very challenging to build, especially as there are very few people in the market with AI knowledge and skills.  

To capitalise on this opportunity and put AI technology within the reach of more organisations, Gϋllich’s company Dydon has made a platform to enable businesses to develop their own systems to deliver reports to users that support their daily work.

“Ordinarily, to create any AI-enabled system you need an AI expert to help construct the algorithm and then build a software application around it, but our platform is designed to remove the complexity and to help a non-AI expert to build the required AI-enabled functionality by guiding them through the whole process,” he said.

This means that businesses can develop systems to gain insights from large volumes of public or open source data to solve business problems, such as identifying cyber risk issues or identifying consumers and analysing their behaviours and preferences, without the need to involve an AI expert.

“Businesses just need to set the parameters for the information they need and the systems goes out, looks for the relevant information, determines the essence of that information and presents the five most useful documents out of the millions analysed, saving time and effort,” said Gϋllich.

In the context of consumer identity, Gϋllich said AI-enabled systems are particularly useful in reading and analysing text automatically using natural language processing capabilities.

“And when a company is addressed by a consumer, that organisation typically wants to know as much information about that person as possible, which would normally take a lot of time to find the relevant information and understand the essence of it,” he said.

“But this is much quicker and easier with AI-enabled text analytics that can find the right information quickly, extract the core content and, if required, create a predictive risk rating driven by the data found.”

When applied to cyber risk, AI-enabled systems are particularly good as collecting relevant information from a wide variety of public sources to identify potential cyber threats and generate security warnings.

“For example, we looked at a cyber attack in the UK retrospectively as a test exercise and found that there had been enough indicators in public tweets about previous attempts that could have been identified and analysed by an AI-enabled cyber risk system to flag up the possibility of the big attack before it happened,” said Gϋllich.

Looking to the future, he said AI-enabled cyber risk systems have the potential to scan relevant sites in the dark web to analyse marketing reports on cyber criminals’ activities and their products to identify new and emerging cyber threats.

However, although the potential of AI-enabled systems is great, Gϋllich said it is important to ensure that the information and insights extracted from the data is presented to users in such a way that it is easy to understand and use.

“This is an area that we have focussed on in developing our platform because unless information is presented in a simple, user-friendly way, they tend to not to use it, which is the reason many AI-enabled systems fail to reach their full potential,” he added.

Gϋllich is to discuss these topics in more detail in a session entitled: AI for new age of cyber risk & consumer identities at Consumer Identity World Europe 2018 in Amsterdam from 29 to 31 October.