AI Use Case Conundrum
To start, I am 100% against large language model AI(LLM/AI). In art, in music, in writing, in coding, in chatbots, in everything. I am in the middle of transitioning from Google to DuckDuckGo because of the recent change to Google's search, because I do not want LLM/AI answers or to even see an "Ask AI" button anywhere. I have done some research, when I have the time and energy, and everything I learn about LLM/AI data centers is appalling. And it's not just the environmental impact that I am against (although that is a big part), it's also the impact on humanity, on human learning, and on the job market.
When LLM/AIs spit out false information with full confidence, even doubling down when pressed, there are people out there who believe it, because they've been led to believe the LLM/AI can't be wrong and that the LLM/AI is the next tech revolution. AI psychosis is a troubling phenomena that appears to be on the rise. College students don't know how to research or engage with writing and reading anymore because they rely so much on LLM/AI (and it makes me glad I got out of academia before this all blew up). There is actual loss of human life tied to LLM/AI use, and I think that is the most tragic of it all.
And the teachers and principal at my school use it all the time! They do a daily announcement video that is almost 100% AI generated, and it pains me to listen or watch (so I don't.) Last year I went to a workshop for principals and their secretaries, and I was flabbergasted by how many people said they fed the LLM/AI their entire schedule and let it dictate how their day went. Some even went so far as to use LLM/AI with student data, which is a huge FERPA violation. My principal uses AI for transcribing important interviews and meetings, and I have read the transcriptions; it is awful. I've done transcription work before, and have training for transcription listening, and to see AI take over that job and do it horribly is disappointing to say the very least.
All that to say, I'm against the use of LLM/AI. But. At the start of June I learned of a use case, and I've been mulling it over for the past few weeks, with no answer.
I have a friend who is deaf/hard of hearing. She has cochlear implants so she can hear some, but she still relies on lip reading to supplement that. We went up north for the weekend, to attend a small conference about making meaningful connections. Now, long before I met this friend, I had experience with and knowledge of deaf culture and deaf accommodation technology, at least up until the mid 2000s. But I haven't seen the newest technology, so I didn't know how the deaf/HOH handle phone calls these days. My friend made a few phone calls over the weekend and showed me how it works. Apparently, she uses AI transcription for her phone calls, a service she receives for free because of the ADA.
During the phone calls, I could hear and see how not great the transcription was, but she didn't seem bothered by it. And the whole experience had me thinking about AI and use case and whether I actually am 100% against AI, or if I'm 99.9% against it. Because it was legitimately helpful, and it does give my deaf/HOH friend access to phone calls as a means of communication. Good accessibility for those who have physical disabilities is important for a functioning, healthy, and equal society.
The question is, should that accessibility come at the cost of all the bad things from LLM/AI? But I don't have answers to any of the many questions that come out of this conundrum. Perhaps when the AI bubble bursts, the accessibility tools will be what remains. Or perhaps not.
What are your thoughts on AI and this particular use case? Have you thought about it before?
Post script: Writing this post and thinking about accessibility has got me pondering how accessible my blog is, with font and spacing and colors. I believe I need to make some changes 🤔
Thoughts to share? I have a contact page now!