Attendees at the Employers Network for Equality & Inclusion’s annual conference heard how decisions made by artificial intelligence must be reviewed to ensure that AI doesn’t adopt our biased behaviour. Contributor Peter Clark, co-founder – Qlearsite.
Hosted by NatWest, the conference, Diversity & Inclusion: The Changing Landscape heard from experts in ethics, psychology and computing who warned that our biased decisions would have long term consequences as artificial intelligence systems become more prevalent in recruitment and selection.
They explained that AIs learnt from existing data, and highlighted how information such as performance review scores and employee grading was being fed in to machines after being subjected to human unconscious bias.
Dr David Snelling, the programme director for artificial intelligence at technology giant Fujitsu, illustrated how artificial intelligence is taught through human feedback. Describing how huge data sets were fed into the program, David explained that humans corrected the AI when it used that data to come to an incorrect conclusion, using this feedback to teach the AI to work correctly. However, as this feedback is subject to human error and bias, this can become embedded in the machine.
Peter Clark, co-founder of conference sponsors Qlearsite, used a live survey to demonstrate how AI was able to understand language, enabling it to summarise huge amounts of employee feedback into key themes and gauge sentiment.
Tracey Groves, founder and director of Intelligent Ethics, described how the conclusions derived by artificial intelligence can impact on the lives of real people, and encouraged adopters of AI solutions to remember that people are not just data. She insisted that for AI to have a positive effect on humanity it must be held accountable and its decisions and conclusions reviewed.
Amid warnings of the risks inherent in trusting unaccountable AI, business leaders also warned that the concept of trust was changing. Social media expert Sam Flynn explained that both customers and potential employees were more likely to challenge brands on social media, and Green Park chief executive Raj Tulsiani was clear that peer review services such as TripAdvisor and Glassdoor were trusted far more than organisation’s own communications.
Psychologist Dr Pete Jones expressed how the prevalence of social media was contributing to a form of unconscious bias known as source bias. Using people’s different social media profiles, Pete highlighted that we place different value on information from different sources. Drawing different pictures from different social media platforms meant that our risk of making biased decisions was increasing due to the pervasiveness of poor quality or out of context information.
Closing the conference with a keynote speech, Baroness Ruby McGregor-Smith CBE was unambiguous in her view that organisations had to be more proactive to win the war for talent. Declaring that the best performing organisations will be those most respected by their customers, Ruby called on employers to be aspirational on diversity, using their leverage to influence their supply chains and to look at the benefits a diverse workforce brings at a team level.
Building on her theme, Ruby insisted that to be successful future business models could no longer be driven by shareholders and promoted employee owned and socially owned businesses as the alternative structures to watch out for.