The Impact - Safety, the Unexpected, and Why One Avatar Is Not Enough

ALT + TEXT: Illustration showing a busy public transport environment during an emergency. Deaf and hearing passengers, staff, and a driver communicate using multiple methods: sign language, written text, mobile alerts, smartwatches with haptic warnings, and face-to-face interaction. A sign language avatar appears on a screen, but the focus is on real people communicating directly. The image emphasises that accessibility is not tokenistic or limited to one avatar, but requires human awareness, sign language skills, understanding of Deaf culture, and two-way communication supported by AI. It represents Deaf Awareness Training 2.0, where technology supports — not replaces — inclusive human interaction.

If a sign language system only works for scheduled announcements,

it isn’t translation — it’s repetition.

And if accessibility is reduced to a single avatar on a public screen,

it isn’t access — it’s a constraint.

Deaf people don’t need repetition. They need access to the unexpected.

Accessibility matters most when nothing is scripted

Real risk happens during moments like:

• emergencies

• evacuations

• security alerts

• live staff instructions

  
  

These situations cannot be prewritten, predicted, or simplified.

Any system that depends only on:

• fixed wording

• predefined messages

• predictable data

will fail precisely when safety matters most.

Calling repetition “real-time translation” creates a false sense of safety. False safety is not neutral — it is dangerous.

Accessibility is not one channel

An avatar on a screen can be one option.

It must never be the only option.

Real accessibility means choice, because Deaf people are not a single user type.

Access should work across:

• watches — haptic + visual alerts during emergencies (space area notification)

• mobile phones — personalised language, speed, and clarity

• AI assistants — interactive, context-aware communication

• kiosks/totems — two-way text, symbols, and sign language

• public displays — real-time, visual-first information

  
  

Different situations demand different tools.

Different Deaf people need different interfaces.

One solution will never fit all.

Two-way communication is the top margin of accessibility

Hearing passengers can:

• Speak to the staff

• Ask questions

• Get clarification

• Change plans instantly

This is two-way communication — and it represents the highest margin of accessibility.

Any accessible system should aim for the same outcome:

• Deaf passengers can ask, clarify, respond, and be understood

• in real time

• especially when situations change without warning

One-way avatar announcements sit at a lower margin of accessibility.

They may inform, but they do not empower.

The unexpected is the real test

Many Deaf people already navigate unpredictability every day.

For example, Deaf drivers rely on visual signposts, not audio directions.

Access works because information is live, visible, and adaptable.

Public infrastructure must meet the same standard.

This is why generative, live systems matter:

• one spoken voice → multiple sign language contexts

• live tannoy announcements → live signing, text, and personal devices

• language that adapts, not scripts that freeze

  
  

Technology alone is not the solution

AI can enable access —

but only if it is designed with honesty and accountability.

Honesty matters.

Language matters.

Choice matters.

Deaf involvement matters — from design to deployment.

Accessibility that only works on schedule

or offers only one method

is not inclusive.

It is restrictive.

Accessibility is not about avatars.

It is about agency, safety, and real communication —

especially when the unexpected happens.