Truhs and Myths of Artificial Intelligence

ETAS panel dismisses common perceptions about artificial intelligence and machine learning, asking for business cases to take off.

Publication date: 11/2016

To tackle the challenges in new technology advancements, such as machine learning, artificial intelligence and automation, the market has to first overcome its own bias, said three panelists during the European Technology Architecture Summit in London.

 

It was evident, during the discussion, that the audience was concerned about the typical clash between humans and machines, with some questions revolving around automation threating human capital.

 

This existential crisis is not justified by the facts, said the panelists, attributing it to the fear of the unknown.

 

Chris Donnan, CTO of Tyler Capital, argued that there are two reasons why people shouldn't be afraid of these new technologies.

 

The first is that machine learning has been present in other aspects of our lives, without us realizing it, such as the social media. "When you're using Facebook, do you feel that someone stole your job?" he said.

 

Furthermore, he said that in the long run, as automation advances, there will be a change in the way people work and therefore it will replace some jobs.

 

"But that's not new and there is no reason to be terrified," said Donnan, giving the example of robotics in medicine, where a proportion of people were replaced by machines for diagnostic purposes.

 

The Risks of AI

 

The risks in implementing these technologies have been exaggerated, according to the panelists, who said that there are no special or additional risks within artificial intelligence.

 

They said a powerful tool that is not fully explainable and doesn't have clear rules needs to be controlled to operate properly. It requires more scrutiny, but in the end it is not substantially different from any other software.

 

Igor Lobanov, head of systems architecture at Legal and General, said that any insufficient complex system can misbehave. "You can use the same traditional techniques to avoid the risks," he said. "You test it, you implement independent controls and you just hold the same engineering practice."

 

The Security Threat

 

The issue of cybersecurity and the threat artificial intelligence and machine learning could pose to the financial firms was also dismissed as a myth.

 

All three panelists agreed that while security is general is something that everyone should take seriously, there are no additional risks involved in automated systems.

 

In fact, said Andy Steele, global head of reference data operations and technology at Barclays, in this field artificial intelligence can play a critical role in confronting cyber threats.

 

"The speed of algorithms gives additional security," he said. "It can detect the differences in the hacked system based on stable characteristics and that way it advances our protection."