With organisations like Women in Machine Learning and Black in AI acting as torchbearers for underrepresented groups in the AI industry, progress has certainly been made to foster diversity and inclusion over the past few years. But unfortunately, these wins are not the end of the story.
Recently, the alleged firing of prominent AI ethicist and co-lead of Google’s Ethical Artificial Intelligence team, Timnit Gebru, has sparked further debate. After a week of disputes over the company’s request to rescind a paper that she had co-written, Gebru left the team. And although Google maintains that the sacking was, in fact, a resignation, it should be noted that the paper in question was none other than one contending that companies must do more to guarantee that AI systems do not enact historical gender biases and offensive language.
Office politics aside, Gebru’s story is yet another pertinent reminder that bias thrives in our conferences and community, as well as in society at large. Put simply, when the leaders and decision-makers in the AI community do not accurately reflect the diversity of society at large, this presents significant issues. Whether Uber’s failure to recognise trans drivers, or sentencing algorithms that unfairly discriminate against black defendants, it is clear that more work must be done behind the scenes to remedy diversity standards within the industry before these errors become too difficult to correct.
If we are to build pioneering tech that works for all and serves all, one thing is for certain: this work must begin in classrooms and university seminar rooms, as much as it should be involved in the development of the tech itself. The question now is, what can companies do to help?
Taking steps towards inclusivity
It probably goes without saying that there is no silver bullet for improving diversity standards, and that meaningful change will be a steady and incremental process. But there are a number of things business leaders in the AI space can do to hurry things along.
From a purely technical standpoint, we must remember that AI is not implicitly biased; that’s why detection tools and white papers must be continually published to ensure that toolsets do not exacerbate pre-existing diversity problems, or put unrepresentative data to use.
Organizations however, can take the first vital steps towards fostering an inclusive company culture, and a workplace where all voices can be reliably heard. If an employee does not believe that a particular product or practice within the workplace reflects appropriate diversity standards, then they should be able to say so, so that teams can hold themselves to account. Above all else, this is the right thing to do – but it should not be overlooked that a diverse workforce can help companies perform better commercially, too.
More specifically, companies should also consider their employment practices. Asking important questions such as: are loyal employees from diverse backgrounds being recognized for their hard work? Organizations that have the means to promote these employees should do so. After all, if the students and professors of the future are able to see themselves in the leaders of today, then they might be more likely to consider a career in AI.
Beyond in-house changes to policy, efforts towards improving education and employment practices must also be bolstered if we are to benefit from AI on a level playing field and create a generation of strong role-models. To shift the status quo, Governments and state-led enterprises must take note and increase funding into STEM education and outreach.
Educational institutions, too, must conduct more research to address the nuances of how identity shapes students’ relationships to tech and the computer sciences. In this way, the benefits to society will be two-fold; not only will individuals from all walks of life be able to contribute to ground-breaking research at elite institutions, but society at large will be able to profit from the development of truly inclusive AI.
Ultimately, the future of the AI space is in our hands. Companies and Governments alike must take the necessary steps to ensure that the technologies of tomorrow are built to transcend geographical borders and cultural limits. Once we as an industry widen the scope of enquiry to examine how tech works in context with more rigour, and companies follow suit, I am confident that we will all be able to benefit democratically from the many benefits that AI has to offer.
Nikolas Kairinos is the chief executive officer and founder of Soffos, the world’s first AI-powered KnowledgeBot. He also founded Fountech.ai, a company which is driving innovation in the AI sector and helping consumers, businesses and governments understand how this technology is making the world a better place.