Presenting at the MRS annual conference in London today ( 11th March), he kicked off with a pointedly contentious topic.

“Can you imagine a world where your democratic vote is replaced by a synthetic vote?” he said. “In other words your visit to the ballot box, your postal vote, is replaced by an AI-algorithm-generated vote for you?

“Ridiculous question, I know, but would we have trusted Brexit if it was synthetic casting the vote? Would we have trusted Trump’s election if it was the synthetic data making that decision?

“Maybe we would have got different, better outcomes… But I’m not going to go there."

While market research professionals are clearly “not always electing presidents of the USA with research”, he added, “we are informing multibillion-pound business decisions”.

“It’s important we treat it with the same scrutiny and rigour that we would an election.”

Jeeawody went on to outline certain biases that can taint AI-generated insights.

“Inclusivity and bias is central to everything we do in our industry and there are certain types of bias that can manifest in synthetic data if we’re not careful how we handle it," he said, adding some brief explanations of the three types:

• Inherited bias is where synthetic data takes biases that originated in the original data – resulting in “garbage in garbage out”
• Inclusivity bias, where practitioners fail to accurately represent different audiences, “whether that’s ethnicity, demographics… not representing them as well as they should be”
• Systemic bias: instances when synthetic data amplifies those biases and can make it even worse

Jeeawody summed up his presentation with three key messages for those looking to dip their virtual toes in:

1. To make sure we embrace synthetic data ethically, we need to build trust
2. Don’t be afraid to ask questions to build confidence and trust
3. The balance of AI and human insight is where we can have the most powerful outcomes.