Here’s how tech companies are mitigating AI bias to promote inclusivity

As AI transforms industries and automates workflows, companies are prioritising responsible AI by building frameworks to ensure safety, fairness, and transparency. AI leaders are implementing guardrails to prevent bias, toxicity, and misuse while promoting inclusivity and ethical practices.

This was noted by representatives of top tech institutions during a session on “AI Everything’ at the Dubai World Trade Centre (DWTC), GITEX GLOBAl, event in Dubai, UAE which presents the biggest, most international edition in its 44th year, welcoming over 6,500 exhibitors, 1,800 start ups, and 1,200 investors alongside governments from more than 180 countries.

The speakers at the session include Boris Krumrey, global VP of  Automation innovations, UiPath, UK, Kaushal  Kurapati, SVP of product, Gen AI platform, Salesforce, USA and Tamika Curry Smith, SVP, chief diversity, equity & inclusion officer, ARM, USA.

One of the key challenges in deploying AI systems is managing bias, which can stem from the data used to train machine learning models, Boris Krumrey, global VP of  Automation innovations, UiPath, UK.

“Historically, systems trained on internet data have inherited bias from that content. Companies are working to counteract these effects by actively developing tools to detect and mitigate bias in AI-generated outputs,” he said.

Krumrey added that this includes detecting harmful stereotypes in language models and addressing biases in image generation, such as earlier instances where searches for engineers yielded predominantly white male results. Thanks to efforts to raise awareness and promote inclusivity, companies are now developing AI that recognises the importance of diverse representation in society.

Kaushal  Kurapati, SVP of product, Gen AI platform, Salesforce, USA  said tech firms are also combating new vulnerabilities such as prompt injection attacks, where malicious actors might exploit AI systems to behave in unintended ways.

“In response, AI developers are implementing trust layers that monitor and filter input and output to ensure compliance with ethical and privacy guidelines, especially when personal information is involved.

“Inclusivity remains a vital focus. Speech recognition systems have faced criticism for struggling with accents, speech impairments, and the voice patterns of older adults. Addressing these shortcomings requires diverse teams of people from various backgrounds to improve AI’s inclusivity. For instance, Google’s Pixel 6 camera incorporated inclusive technology to accurately capture darker skin tones, reflecting the benefits of a diverse team driving product innovation,” he added.

Kurapati emphasised that the business impact of responsible and inclusive AI is significant. By ensuring that AI systems reflect the diversity of their users, companies are not only doing the right thing socially but also improving user satisfaction and driving growth. Google’s success with the Pixel 6’s inclusive camera technology highlights how ethical practices can result in increased sales and customer loyalty.

As governments and regulatory bodies like the European Union introduce AI regulations, companies are building solutions to comply with these frameworks. AI trust layers and evaluation mechanisms are becoming standard practice, ensuring that all inputs and outputs align with the company’s policies on ethics and transparency. Red team testing and cost mining offer additional layers of security, providing businesses with full oversight over their AI systems.

In addition to ethical AI, Tamika Curry Smith, SVP, chief diversity, equity & inclusion officer, ARM, USA said that companies are investing heavily in creating a more inclusive workforce. A dedicated team has revamped the global hiring process to ensure consistency and mitigate bias. This initiative is part of a broader effort to increase diversity within the tech industry, particularly in AI development.

“We have redefined how we source talent globally,” said Smith. “From working with external partners that focus on women and underrepresented communities, to building partnerships with organizations like NASA and UNICEF, we’re trying to attract diverse talent early on.”

These partnerships are essential to fostering interest in STEM fields, especially for younger students. By sparking curiosity in high school, companies aim to address the knowledge gap and make the tech sector more accessible, she added.

“Representation is key, to make AI as inclusive as possible, we need voices from diverse backgrounds at the table, shaping the future of AI,” Smith added.