Everyone’s talking about Chat GPT at the moment and the advancement of machine learning and natural language processing.
Chatbots like Chat GPT are designed to mimic human conversation and can reply to whatever people ask them to do. They’re already starting to be used by businesses, governments and individuals.
People are finding they can write essays or even stories at the click of a button (which might be bad news for the person writing this!), or get help with research.
In the past few weeks, there’s been stories of an AI-generated photo winning a photography competition, and of AI-generated music featuring the cloned voices of star singers.
And we’re only just at the start.
People are talking about many elements of modern life changing hugely as these chatbots become ever more intelligent, and people find ever more uses for them.
It’s being talked of as a new industrial revolution.
Perhaps the first question many people are asking is: will Chat GPT take my job? But there are deeper concerns too.
What’s the future for humanity if we invent Artificial Intelligence models that are far more intelligent than ourselves? What are the ethics behind the development and use of these tools? How much power will we give them and how will we keep them under control?
I saw it mentioned last week that AI in the future might be more intelligent than humans to the same extent that humans are more intelligent than frogs. Which is funny in a way, but also incredibly scary.
And how might advanced AI affect deaf lives specifically? Its reach might extend into areas we can’t even imagine right now, but I tried to imagine a few potential effects. Here they are.
If there are great leaps in technology, we might find automated speech-to-text systems becoming vastly more accurate. The subtitles on YouTube might become completely accurate, at last, as AI becomes far better at discerning speech. But that might be just the start. Having subtitles on life might be a reality, with the right pair of glasses.
As AI is ever more used to accelerate advances in healthcare, we may find that hearing aids and cochlear implants vastly improve to more closely match each user’s exact hearing loss, and become far more sophisticated at matching the sound of real life. They might even get so good that they develop programmes that are better than ‘normal’ hearing, making people who aren’t deaf want to use them.
As medical research advances ever more quickly with the assistance of AI, we may see huge advances in treatments for deafness and hearing loss. Advances that once took decades may only take months. The ‘cure’ that is part of the film The End, which depicted the last deaf person, could be a real thing.
Something that’s always seemed a long way off, or an impossibility, are AI interpreters. Sign language is so complex that there’s been a firm belief that nothing can replace a real-life interpreter. But with chatbots potentially being far more intelligent than humans, AI interpreting might become a possibility. This could see BSL interpretation added to all kinds of resources at the touch of a button, delivered by an avatar who looks just like a real person, but isn’t.
These developments might have an upside, leading to new opportunities for deaf people in employment, accessing culture, or new social possibilities.
It might also lead to a downside: the deaf world’s already changed a lot in our lifetime, with more and more deaf clubs and schools closing, and a feeling that the community is more disconnected since the rise of mobile phones and social media. In a world of huge advances in healthcare and technology powered by advanced AI, will deaf people, the deaf community and sign language continue to exist?
Charlie Swinbourne is the editor of Limping Chicken, and is also an award-winning screenwriter.
Adrian Pickering
May 15, 2023
Fascinating article and, I suspect, highly prophetic. Used well, AI could build bridges between hearing, deaf and Deaf communities. In my opinion, AI is the only way to make sign language ubiquitous, and to make everyone expect to deliver BSL versions of all their content. However, I do not believe AI should ever replace human translations and interpretations – who wants a computerised docotor to offer comfort and assurance? Regardless of AI advancement, they won’t offer the compassion of human beings. The reality is that there is such a shortage of deaf translators and hearing interpreters, that avatar signers are needed to fill the gaps. And so are more human signers.
Full disclosure: I am part of Robotica, the AI sign language avatar company. See https://limpingchicken.com/2022/09/28/a-bot-topic-debate-continues-over-the-role-ai-interpreting-has-in-bsl-access-industry/
Graham Turner
May 15, 2023
“Who wants a computerised doctor to offer comfort and assurance?” One short answer to this is likely to be ‘anyone who wants the service to be cheap’. And as long as the state is paying for such services, there will realistically be pressure to provide them as cheaply as possible, so long as they’re considered adequately effective. Are the bots adequately effective now? No. Are they getting more effective rapidly? Yes, very much so. Will people adapt to less human services? Well, what does the entire span of human history show? They will.
I don’t like it, I don’t want it, but I’m realistic enough to believe we’d better acknowledge what’s coming if we continue on the present path.
dionne7ad2457101
May 16, 2023
Really interesting thanks! I was wondering if AI chatbots would make better text crisis support than what is currently available? peopl needinging Samaritans, mental health crisis support and other services talking to an AI chatbot that may respond better than a human? Empathy, not scared of deaf people, able to adapt text to match the person, offer support without patronising. It could be a game changer!
Adrian Pickering
May 18, 2023
To Graham Turner:
>>”One short answer to this is likely to be ‘anyone who wants the service to be cheap’. And as long as the state is paying for such services, there will realistically be pressure to provide them as cheaply as possible, so long as they’re considered adequately effective.”
Of course there will be demand for the race to undercut humans, and this must broadly be resisted. The purpose of AI like this is to fill the gaps. Using AI to provide BSL on the cheap would be a disservice to deaf people, as well as to deaf translators and hearing interpreters.
>>”Are the bots adequately effective now? No. Are they getting more effective rapidly? Yes, very much so. Will people adapt to less human services? Well, what does the entire span of human history show? They will.
I don’t like it, I don’t want it, but I’m realistic enough to believe we’d better acknowledge what’s coming if we continue on the present path.”
I think there’s space for both – for AI to offer translations that otherwise simply would not exist. The rate at which new content is created means that humans alone could never catch up, let alone keep up. If companies focus on how AI can be used for good, we will one day see BSL available on-demand, everywhere. It’s absolutley critical that Deaf people and all BSL users shape the application of AI translations to ensure that it doesn’t unintionally cause harm.
Sally Chalk
May 19, 2023
This article from Calum Medlock talks about the importance of accuracy in sign language translation, photo-realism and the use of sign language rather than SSE or Makaton when developing AI products. It also discusses its use in transforming access in public spaces (where none existed previously) rather than using it in healthcare settings. https://www.signapse.ai/post/signapse-ai-sign-language-translation-photo-realism-accuracy
annoyed
May 24, 2023
sally from clarion and signapse stop using limpingchicken for spam and free adverts. putting an advert in the comments is wrong. you are exploiting goodwill. you do not have right.
remember this
https://limpingchicken.com/2021/07/02/deaf-news-little-mix-court-case-adjourned-after-interpreters-refuse-to-provide-access-following-complaints/