Artificial Intelligence (AI)
The use of artificial intelligence (AI) and automated technologies is changing workplaces. Although data analytics and automation are not new, AI technology has advanced rapidly in recent years alongside innovations in algorithms, data volume, and computing power. AI-powered platforms are now used to screen job applicants, streamline the application process, and provide on-the-job training. AI is also powering exciting innovations in assistive technology for people with disabilities. While AI holds tremendous potential for both employers and employees to make workplaces more inclusive, it also carries risks for people with disabilities related to privacy, ethics, and bias.
What is Artificial Intelligence?
Artificial intelligence refers to the use of computer systems to perform tasks that traditionally require human intelligence and senses. AI “learns” through the use of statistical techniques that allow it to incrementally improve performance on a task. This process of “machine learning” allows the machine to generate rules and predictions on its own by analyzing large quantities of raw data, rather than being explicitly programed.
AI can process lots of information by enabling advanced data analysis and pattern-finding in large, complex data sets. It is also used to automate low-level, repetitive tasks and make complex tasks more efficient.
AI and the Future of Work
AI advances have opened up new opportunities for people with a wide range of disabilities to become more engaged in work due to software that can learn how to recognize and respond to images, sounds, and linguistic expressions. Gartner predicts that by 2023, the number of people with disabilities employed will triple, due to AI and emerging technologies reducing barriers to access.
These emerging technology innovations include:
- Auto captioning tools
- Autonomous cars, which when built with Universal design (UD) means “the design of products, environments, programs and services to be usable by all people, to the greatest extent possible, without the need for adaptation or specialized design.” (UN Convention on the Rights of Persons with Disabilities (CRPD), Article 2). UD, a design process rather than a description of products, encompasses accessibility, going beyond it to address the widest possible user base. UD does not exclude assistive technologies where needed, as they are part of the usage environment that must be taken into account.” data-original-title=”universal design”>universal designprinciples may provide expanded transportation options for people who are currently unable to drive
- Facial recognition and image recognition to support navigation and interaction with the environment for people who are blind or have low vision
- Text summarization to enhance comprehension for people with cognitive disabilities
AI, big data, and data analytics are also becoming commonplace in recruitment. Over 1/3 of respondents to LinkedIn’s 2018 Report on Global Recruiting Trends shared that AI was the top trend affecting how they hire. These recruiters are using data analytics to quickly sort applicants and passive contacts, and to scrape and analyze predictive behavior based on social media profiles. Employers are also relying on pattern recognition methods to analyze current job descriptions and the success rates of past hires when determining their strategies for seeking new hires.
These methods are praised for their efficiency and cost-effectiveness. Many companies assert that they augment their diversity goals by reducing human error, unconscious bias, and nepotism in the hiring process. AI also holds potential to help companies directly match applicants with positions based on competency assessments, leading to better hires. Companies seeking to meet Section 501 diversity quotas, for example, could benefit from using such tools to target and reach a diverse candidate pool.
Unfortunately, AI is also limited by reflecting the implicit biases of its makers. Models learning from biased training data may reproduce historical bias against minority groups, including people with disabilities. Further, training data typically underrepresents outlier populations, including people with disabilities. Because people with disabilities are a widely heterogeneous group, mitigating bias is a more complex problem than controlling for factors such as gender or race. Jutta Treviranus, Director of the Inclusive Design Research Centre at OCAD University in Toronto, Canada notes that success in reducing disability-related biases in AI requires approaches rooted in the jagged starburst of human data—rather than simple bell curves.
Gathering inclusive datasets will prove essential for building effective solutions, though finding those solutions is still in progress, and also holds significant privacy challenges. Employers must be sure to work carefully with vendors on these concerns to avoid excluding people with disabilities from their talent pool.