SignGPT, starting points, and the risk of a “Digital Milan”

Many people think SignGPT is the same as ChatGPT, just with sign language added. This is where confusion begins.

ChatGPT starts with:

• text

• or speech

This makes sense because ChatGPT is built for people who write or speak first.

Because of this, many people assume SignGPT should work the same way:

• start from text or audio

• Then generate sign language

  
  

But this assumption creates a serious misunderstanding.

For Deaf people:

• Sign language is the first language

• Meaning starts in sign, not speech or writing

• Sign language is not a translation layer

  
  

When a system called “SignGPT” does not start with sign language, people become confused:

• Is sign language the core, or just the output?

• Who is the system really built for?

  
  

This confusion matters — because the starting point equals power.

History reminds us why this is important.

In 1880, sign language was removed from Deaf education in Milan. That decision caused long-term harm by pushing sign language aside.

Today, with Generative AI and LLMs, there is a modern risk I think of as a “Digital Milan” not by banning sign language, but by ignoring it at the foundation.

When AI always starts from speech or text:

• Education barriers can return

• Communication barriers are repeated digitally

• Sign-first people are excluded again, quietly

This is not about blame or intent. It is about design choices and their consequences.

There are two valid human starting points:

• Speak/write first

• Sign first

ChatGPT already supports the first.

SignGPT should support the second.

Clear language and clear design matter. Otherwise, we repeat old patterns in new technology — and call it innovation.