In a powerful critique of modern broadcasting, celebrated blind comedian and presenter Chris McCausland has condemned the rise of what he labels 'woke' artificial intelligence in television audio descriptions. He argues that new AI systems are failing blind and partially sighted viewers by deliberately omitting crucial details about a character's race, physical appearance, and even key plot points.
McCausland, a prominent voice for disability representation, revealed that broadcasters are increasingly using AI to generate the audio-described tracks that narrate visual elements for blind audiences. He claims this technology is being programmed with a restrictive 'woke' agenda, stripping away descriptive language in a misguided attempt to avoid causing offence.
The Heart of the Controversy: What's Being Left Out?
The core of McCausland's argument is that these sanitised descriptions are ultimately discriminatory. He provides a stark example: if a plot hinges on a character's racial identity or a visual gag relies on someone's distinctive hairstyle or clothing, the AI simply glosses over it. This, he asserts, leaves blind viewers confused and excluded from the full narrative experience.
'The whole point of audio description is to give you the information that you're missing,' McCausland stated. 'If you start editing that information... you are patronising blind people. You are saying, 'We're going to decide what's good for you.''
A Call for Context and Common Sense
McCausland is not advocating for insensitive or gratuitous descriptions. Instead, he emphasises the need for context and nuance. Describing a person's race or appearance is often vital for understanding context, character dynamics, and storytelling. He believes that well-trained human describers understand this balance, whereas AI, bound by rigid and overly cautious protocols, does not.
His comments have ignited a fresh debate about the role of technology in accessibility services. While AI promises efficiency and cost-saving for broadcasters, McCausland's intervention serves as a crucial warning: true inclusion means providing complete information, not a censored version deemed appropriate by an algorithm.
