Why inclusive AI matters | Chenayi Mutambasere | TEDxSt Helier

AI Bias and Inclusion Gaps - AI systems rely on the data they are trained on, and unseen or unheard voices are not reflected in the AI mirror [2] - 90% of data used to train AI systems is in English, while only 18% of the world speaks English [6] - Only 1% of the data used in large language models for global AI models is African language data [7] - Globally, only 20% of AI professionals are women, and in the UK, only 1.7% of AI professionals are black [7] - CV filtering algorithms show bias, with white-sounding names having an 85% chance of moving to the next stage compared to 9% for black-sounding names [9] - Facial recognition systems have higher error rates for darker-skinned women (34.7%) compared to lighter-skinned men (0.08%) [10] Impact of AI Bias - Biased AI systems can affect hiring decisions, medical treatment, credit scoring, and content streaming [8] - Exclusion in AI development leads to scaled exclusion, while equity promotes inclusion and opportunity [12][13] Addressing AI Bias and Promoting Inclusion - Digital exclusion leads to data exclusion; in Africa, only 40% have internet access, and in the Caribbean, as low as 11% [15][16] - Global standards and policies are needed to protect, value, and respect data, treating data breaches with the same seriousness as money laundering [18] - Companies should disclose the inclusive characteristics of their data, similar to nutritional information on food packaging [19] - Diversity of thought in AI is a skills gap that governments must address by encouraging the participation of women and black people in AI development [20][21]