AI and the Blind

Announcer [over music]: Welcome to the Nation's Blind Podcast presented by the National Federation of the Blind, the transformative membership and advocacy organization of Blind Americans. Live the life you want.

Melissa Riccobono: Hello and welcome to the Nation's Blind Podcast. I am Melissa Riccobono, and I am here with my wonderful co-host.

Anil Lewis: Oh, I'm a wonderful. This is excellent. I'll take, "My wonderful co-host," and I'll flip it right back to you with, my marvelous co-host. This is Anil Lewis. Welcome to the Nation's Blind Podcast.

Melissa Riccobono: Well, thank you. How are you, Anil?

Anil Lewis: I'm doing really good. It's getting close to the... Well, it is not getting close to, it is the holiday season.

Melissa Riccobono: It is.

Anil Lewis: Loving interacting with everyone and looking forward to a tremendous 2024.

Melissa Riccobono: Absolutely. And we've gotten some nice feedback on our shopping episode, so if you haven't listened to that one, our previous episode, take a listen. It's probably a little bit late. It might not help you this holiday season with your shopping since you might not be hearing it till after the holiday season is done. But thank you to all those that have reached out and who have said that they enjoyed that episode and that they learned something. So thanks a lot for that. We love that.

Anil Lewis: Yeah, some of the content is definitely evergreen though, so it's not all about holiday shopping.

Melissa Riccobono: No.

Anil Lewis: It was interesting, I saw something on social media around the Instacart thing. There were some people that said that it just doesn't work for them, and then others were saying, no, it's wonderful for them. So it's interesting seeing the different perceptions of the usability of a lot of the different options.

Melissa Riccobono: No, absolutely, that is a really good point. And that's the thing, not one thing is ever going to work for everybody, and nobody's ever going to agree on the best way. So that's why we need to be communicating about all these things.

Anil Lewis: Yeah, it'll all be fixed, though.

Melissa Riccobono: Oh, well, it'd be great if they were all fixed, but we also need to be able to pick the right tool from the toolbox, so to speak. And sometimes you don't always have to use a hammer, you can use a wrench or pliers or something. You don't always have to use your hammer.

Anil Lewis: We won't have to worry about any of this in the future because artificial intelligence is going to take care of all of it for us.

Melissa Riccobono: It will. We're going to have our Rosie robot, the housekeeper on  The Jetsons,  and I think she might've done the shopping too. That'd be cool. I'd be okay with that. [Laughter]

Anil Lewis: There you go. There you go. So speaking of artificial intelligence...

Melissa Riccobono: Yeah, we're talking about artificial intelligence today. I love how you did that, Anil. You're like a professional or something.

Anil Lewis: Well, I tried. I'm working on it. Getting better at this.

Melissa Riccobono: Oh, absolutely. You're great at it. And we actually have two wonderful Federationists to talk through AI with us, and we're very excited to have them here. Last time you just heard from Anil and me. This time it's going to be wonderful to have other perspectives on the podcast. So we have Aaron Cannon. He lives in South Dakota and he is an accessibility specialist, I believe he said, for Ancestry.com. And Anil, do you want to introduce the other federation member that's on with us?

Anil Lewis: You've used the word wonderful several times. I thought you would've wanted to introduce our second guest, Mr. Gary Wunder.

Melissa Riccobono: Wunder, yes.

Anil Lewis: Long-time Federationist, mentor, editor of our Braille Monitor. Gary, glad you could join us.

Melissa Riccobono: And Gary's from Missouri. Hello, gentlemen. How are you?

Aaron Cannon: Doing well, thank you.

Gary Wunder: Very well.

Melissa Riccobono: And for those that didn't know, Aaron was the first person that spoke and Gary was the second person that spoke. [Laughs]

Aaron Cannon: I probably should have mentioned that.

Gary Wunder: And we are trying to decide which of us are artificial in our intelligence.

Melissa Riccobono: Wow. [Laughter]

Anil Lewis: I've been artificially intelligent for quite some time now.

Melissa Riccobono: Well, why don't we start there? Of course, if we're going to talk about AI, I guess I would be interested, and I think our listeners would be interested as well for both of you, I'm guessing that you have used AI, either Be My AI or something else similar, to help you complete tasks or just to see what it's all about. And we would just love to hear from you what your experiences have been so far, what things have gone really well, what surprised you and amazed you. And then on the flip side, what maybe are you worried about or didn't go so well?

Gary Wunder: So, Be My Eyes, the group that has put together a bunch of volunteers to help us has added a new wrinkle, which is called Be My AI. So you can either talk with a volunteer and get them to look at things and ask questions of them, or you can use this artificial intelligence called Be My AI. It's available 24 hours a day just like the volunteer service is for Be My Eyes. And then there is for more word-oriented things, ChatGPT and other equivalents that use large language model vocabulary.

I've had really good luck with both of them. I sometimes love the interaction that I get from a volunteer, but sometimes I just want information. And when I do that, the artificial intelligence part of Be My AI is great. The only downside that I have seen to Be My AI is that it tells me just how wide my camera view is, and reminds me how often I have underestimated the picture that I am sending to volunteers. I think that I'm sending them a picture of a piece of cake that I'm about to eat, but in reality they see the fact that I've got three coffee cups on the desk, and   [Laughter]

Anil Lewis: So Gary, can you walk our listeners through an actual experience? So what do you do? You take your smartphone and snap a photo of something in your environment?

Gary Wunder: Yeah. So I go to a lot of hotels, and a lot of hotels since COVID have been providing three bottles. They are mystery bottles. And I take my phone into the shower and I snap those, and Be My Eyes through Be My AI will tell me, "There are three bottles. The one on the left is shampoo, the one in the middle is soap, and the one on the right is conditioner." I list it that way because mostly that isn't the way it works. Mostly you have your shampoo and your conditioner together, and your soap is off to one side or the other. But it just goes to show you can't count on those things. So I am more comfortable with Be My AI in a shower than I am asking-

Anil Lewis: A live volunteer? [Laughter]

Gary Wunder: Exactly. And if I am more comfortable, I am way sure that the live volunteer or the paid person is way more comfortable.

Anil Lewis: And with Be My AI, you can actually ask questions about the photo once it gives you the initial description, right?

Gary Wunder: mExactly. Or your focus is not good enough, it's too high or it's too low to give you an accurate description, you need to do this. I wish sometimes that it were a little more interactive and would let me ask my question at the beginning, because it gives me a lot of information that I may not want or need. Like, "You are standing in a shower stall with a towel rack... And then there are these three bottles." I kind of know all of that stuff already. So it'd be great if I could do the question first, but it's a great thing.

Anil Lewis: Nice. That's some good feedback we get to share.

Aaron Cannon: I had my daughter take a couple photos the other day. I could do a quick demo.

Melissa Riccobono: Sure. Yeah, that'd be great.

Aaron Cannon: I actually did this as part of a chapter meeting. I was invited by the Minnesota at-large chapter to talk about how I use AI.

Melissa Riccobono: Well, while we're waiting, we do have some comments from members. We did post on Facebook a couple things, and there have been people that say they have used it for helping them find outfits, for getting more descriptions of colors and things on online shopping. I'm kind of fascinated by that. I wonder if they take a picture of their computer screen, or if they actually get a screenshot and put it into the app. And I'm kind of curious about how that works. Someone said they changed the ink in their printer without needing help. That's really interesting. But then, well, there are people that say there's things on the flip side. So we'll talk. Gary did mention the wanting to ask the question first, and there's also some other people that have some other examples, but let's get the demonstration.

Aaron Cannon: So I have taken, or rather my daughter has taken a picture of... I just asked her to take some random photos of around the yard. And so I'm going to open this photo in the photo viewer and then I'm going to click share, and then I'll go up here and I'll get my phone close to the mic so you can hear it.

Melissa Riccobono: And then you just share with the Be My AI app?

Aaron Cannon: Yeah, that's correct. So I'm going to click Describe with Be My Eyes.

iPhone Voice: Describe with Be My Eyes. Cancel button. Cancel button. Please wait. Be My AI is... Please wait. Be My AI is writing a message.

Aaron Cannon: It's thinking.

iPhone Voice: Be My AI is still writing a message.

Melissa Riccobono: It's thinking very hard.

Aaron Cannon: Yes.

iPhone Voice: Be My AI: the picture shows a group of chickens outdoors. There are patches of snow on the ground and the chickens are gathered around what appears to be a few pieces of broccoli. The chickens are of different colors including brown, white, and speckled patterns. The sunlight casts shadows on the ground, indicating it might be late afternoon or early morning. The ground is a mix of bare earth and snow.

Aaron Cannon: So quite a bit of detail there. And like we said earlier, you can ask questions. How many chickens are there, and can you tell what their breeds are?

iPhone Voice: Inserted, "How many chickens are there, and can you tell what their breeds are?" Please wait, Be My AI is writing a message. Be My AI is still writing a message.

Aaron Cannon: Tough question.

iPhone Voice: This is taking longer than usual. Please wait.

Aaron Cannon: Sometimes the technology does go down. I've noticed it slow down in the middle of the day.

iPhone Voice:Please wait. Be My AI: There are ten chickens visible in the photo. Identifying specific breeds from the image can be challenging without clearer details, but there seem to be a mix of brown and white chickens, some with red combs and others with Taylor combs. The brown ones could possibly be Rhode Island Reds or a similar breed, and the lighter colored ones might be leghorns or another light feathered breed. There's also a speckled chicken that could be a Sussex or a Wyandotte.

Aaron Cannon: All right, not too bad. It got one of them. So it's not always accurate, but we do in fact have Rhode Island Reds. The rest of them are Brahma. But still pretty good.

Melissa Riccobono: Very cool.

Gary Wunder: Which is an interesting thing, because artificial intelligence implies some speculation. It isn't, what do I know based on just being a search engine? But, what might I conclude from the limited information I have? I found it interesting that one of the people who commented said that she asked for AI to tell her which outfit looked better. That's a stretch for me. I've never tried that.

Melissa Riccobono: Yeah, and I don't know either. And she was saying it would say things like, "This looks smart," or "This looks cute," was another word that it used. I don't know, to me that bothers me too. You can actually go in and change the settings, which I think is interesting. Because sometimes I feel like Be My AI can get very judgy about, "This scene looks cozy," or "It looks bare." That bothers me a lot, actually. I don't really need it or want it to tell me if it looks cozy most of the time. I don't know. I don't know, Aaron, if you have thoughts about that, or Anil, but that has been one thing that has really bothered me about it.

Aaron Cannon: Yeah, I personally like it a little bit because you have that opportunity to ask follow-up questions. And so if it tells me the room is cozy or the outfit is a cute outfit or whatever, if I'm taking a picture of my daughter, I can ask it, "Why do you say that? What makes you say that? What makes you say that it's cute," or "How did you reach that conclusion?" And sometimes it has a good answer, other times it doesn't. I guess the bigger concern for me is, if I have on a hideous outfit, is it going to tell me that or does it only say nice things? [Melissa laughs]

Anil Lewis: I think this highlights the positives and the negatives of AI, because all of those really are somewhat subjective because style is kind of individual. And so it's wonderful that we, as blind people, have the option to get information in this dynamic way. I love that chicken demonstration. That was pretty cool. But then as Gary was mentioning earlier, all of this is built on top of large language models, so it's really about aggregating data throughout that digital space and coming to, I guess basically what we'll call it is artificial assumptions on things. And we know what happens when you make assumptions, even with artificial intelligence.

But I think that with a lot of it, if you're leaning on the AI to tell you what's cute, what looks good, then it's going to be subjective based on the large language model of data that's available. But to your point, Melissa, I think that then the user has to practice their own individual decision-making by getting the information that they can really make the decision based on their own personal preference. But again, if individuals would love for their particular personal shopper to be their artificial intelligence engine, then of course that's their choice as well.

Gary Wunder: Sometimes we're confused by too much data, but I think sometimes we get too little data also. And although artificial intelligence has its own problems, whether a thing is cute or sexy or not very attractive on you, those are three opinions you might get from three inquiries that you make of different people. So my wife, for example, does not really want to go shopping with one of the people we pay to do shopping assistance. She would rather go with her daughters when it comes to clothing because there's an issue of trust, and she believes that her daughters all share her perspective about what would look good and what wouldn't look good. When I go shopping for clothing, I'd like AI perhaps because I think that the clothes salesman is going to tell me whatever it will take to close the sale. So it's all, how much data do you want and how much decision-making do you retain in your head versus what you hear through your ears?

Melissa Riccobono: And also, we haven't even talked about the fact that sometimes AI does hallucinate, which means that it says there's something there when it really isn't there. Somebody had a post the other day about, "I had a holiday card and it said that it was congratulating me from graduating from law school." And then I said, "No-

Anil Lewis: Congratulations, Melissa, you got your J.D.! That's awesome!

Melissa Riccobono: That's funny. No, it wasn't me, it was somebody else. And then they were like, "No, it doesn't say that." And they're like, "Oh, I'm sorry, it's a birthday card." They're like, "No, it's actually not my birthday either." Eventually it was able to read. The cool part was, it was able to read the handwriting, which has often been a problem. And to me, that's been something that's been amazing, and I can think about... Particularly because, number one, my two daughters are blind, number two, my oldest has never really been taught to write and read cursive. And I think about letters and different things that my mom has that she wrote to my dad and my dad wrote to her, and I've told my mom already I want those letters, or I want somebody to have these when you no longer want them.

Because I don't want them to be lost. I'm very much into family history and making sure that that is preserved. And I've often thought, how am I going to read those when the time comes? And so knowing that there's some technology out there that might allow me to do that, at least with a fairly reliable accuracy rate is pretty exciting to me, in that I won't have to have somebody that doesn't know my family, potentially reading something that could be a little bit more personal. So that's pretty exciting to me.

Anil Lewis: Gary, Aaron, have either of you used the ChatGPT?

Aaron Cannon: I'm a big sucker for ChatGPT. I've been a paying subscriber for I think almost a year now.

Anil Lewis: Can you describe, what is ChatGPT?

Aaron Cannon: Yeah, ChatGPT is a service that allows you to essentially chat with an AI, and GPT is the model that's behind the Be My AI service. So when you're talking to Be My AI, you're also talking to ChatGPT. It's basically the same thing. ChatGPT is more of a general service; it's not specifically for the blind. They have a website and a mobile app, both of which are so-so when it comes to accessibility, but you can muddle your way through most of the time. Originally, they only allowed you to chat with the AI, and that was about it. But that was huge. You could ask it to write a poem for you. In fact, one of the demos that I did in the chapter demo that I did a few days ago for the Minnesota at-large chapter, was that I asked it to write a poem about Kenneth Jernigan, and it actually did a really solid job. Not perfect, but it was a lot of fun and it was reasonably well done.

Anil Lewis: Can you tell it what type of style? Can you say write a sonnet about Kenneth Jernigan?

Aaron Cannon: Yeah, absolutely.

Anil Lewis: Oh, wow.

Aaron Cannon: For a laugh, a friend of mine was upset about a Facebook post that he read, a local group. And so I said, "Well, would it sound better if you read it like this?" And I had ChatGPT rewrite it in the style of Dr. Seuss.

Anil Lewis: Nice. Would you, could you? Yeah, I love it.

Aaron Cannon: Would you, could you fix the roads? Yeah, it was fun. So really, sky's the limit. That's the hard part about AI, I think, at least for me, it feels like there's so many great ways in which I could be using it right now that I haven't even thought of yet. And that's really where we're at right now with the technology. We don't know how to use AI as a society. We simply don't know how to use it. And so learning is part of the excitement, and finding new and creative ways that we can use it is what keeps me subscribing to ChatGPT. But anyway, so you have the ability to chat with the model and ask it to do things. Now you can also upload photos, so you can also have similar functionality to what you have with Be My AI, and ask it questions about the photos. You can upload documents. And one way in which I've used this is, with work we were selecting our healthcare and I was a little bit confused, and so I uploaded the whole document, the PDF of the terms, and asked it some questions.

And it was actually able to go in, read the document, figure out some tables that weren't very readable with the screen reader, and then I was able to use it to perform some calculations on that data to figure out which was going to be the best health plan for me. I was pretty astonished that I was able to do all that. The way it was able to do the calculations is that it's able to also write and run small computer programs. It uses the Python programming language. But it does that behind the scenes unless you ask it to give you the source code that it's writing. Then the other fun thing it can do, well, there's two more, it can listen to your voice and transcribe your voice. So you can actually chat with ChatGPT using your voice. And it's far superior to anything I've ever used before. Much, much better than Siri, much better than Dragon NaturallySpeaking. It's quite astonishing.

Anil Lewis: So it only transcribes or does it talk back to you?

Aaron Cannon: It transcribes and it also talks back, yes.

Anil Lewis: Oh, wow.

Aaron Cannon: Yeah, so it's basically like having a conversation. It says something and then it waits for you to talk, and then it talks. And then the other way you can use it is for generating photos. You can describe a photo that you want it to create or a drawing or something, and it will create that for you.

Anil Lewis: Okay. Now, this one I have to comment on. I had [inaudible 00:21:01] staff when Kennedy and Carl told me that it could do this. I'm like, "This is amazing." So I said, "Let's tell ChatGPT to draw us a picture of God." So Kennedy put together all of these different whatevers, omnipresent, all powerful, just all that into the AI engines. I'm not going to call them out because well, I really don't remember which ones were, but the one that Kennedy used gave him four photos of old white men with white beards, et cetera. And then the one that Karl used came back and says, "Nope, we can't do that. There are too many diverse opinions on what this entity..." That kind of thing. So I think that again, shows the difference between the good and the bad of this whole ChatGPT AI piece. If we just yield to whatever that one perception is, it could really be problematic.

Aaron Cannon: Yeah.

Melissa Riccobono: I think I want to talk about that, especially when it has to do with blindness. But why don't we take a really quick break and then I have a question for Gary. And certainly, Aaron, you can chime back in too because I think you might've had something to say to Anil. But I do want to make sure that we can hear a word from our sponsor, so to speak. So let's take a quick break for a promo and then we'll be right back with you.

Promo announcer [over music]: The PAC plan is our Pre-Authorized Contribution plan. This is a way for anyone to make contributions to our organization on an ongoing basis, and that ongoing basis enables us to be able to budget for the money that's coming up. Anyone who wants to contribute to the plan for the first time or who wants to up their contribution, just share with us some financial information. It can either be a debit card or a bank account and routing number, and we will make sure that your generous donation is processed each month. There are two ways that you can do this. One, you can go to NFB.org/pac, that's P-A-C. You can also give us a call at 1-877-NFB-2PAC. That's 1-877-632-2722. Leave a message that does not contain financial information, but that contains your contact info and one of our folks will call you back.

Melissa Riccobono: All right, so Gary, I have something to ask you. You're an editor of a magazine, the Braille Monitor, the Federation's flagship publication. A lot of people are now using ChatGPT or other things like it, to write various things, or at least to get them started in writing different things. What are your thoughts about this and do you think it's not a great idea? Have you thought through this at all?

Gary Wunder: I have definitely given it a lot of thought, and I pretty much unconditionally think it's a good idea. I have so many people who have stories inside them. Some of them are heart stories that they haven't yet gotten to the head, and some of them are head stories, but their biggest problem is they can't communicate it from them to me because they don't have the confidence to believe that they can do it. So part of my job as an editor is to convince them that whatever they write down, we will work together to make look good. And the me part of we will become invisible when it comes to the publication of their article. So if they can just help pull it from the heart to the head, to my head in some way, shape, or form, we'll do it. So if AI can do that, if AI can let them jot down and outline some thoughts that may even be random and AI can put in a logical sequence and give me a draft to work with, I'm all for it.

Now, no one has ever admitted to me that this is the way they gave me their first draft. So I'm going to do what my wife says is stupid and put my job on the line and say that, there are times when I feed AI an outline and I make it write all of the diverse facts that I have given it. And then I go through and I make it sound like Gary Wunder. I'm not afraid of that technology any more than I'm afraid of using a six-dot Braille keyboard to write, even though a mistranslation in Grade Two can be really hard for a print reader to figure out. I'm not any more afraid of AI than I am Dragon NaturallySpeaking or Siri, because I still get the last word on whether or not it goes or not. So I think the things are wonderful.

Anil Lewis: I love the way you described that, because that's the responsible use of AI. I mean, I've been hearing stories like on NPR where they've been using the AI to write legal briefs, and the artificial intelligence actually makes up citations. So it's like creating legal precedence on fictitious precedents. It's just really interesting. But even more so, it was interesting, a friend of mine who's a college professor says that one of her students actually used artificial intelligence to write a term paper. And the reason that she knew that it was not the student's own is because the ChatGPT or whatever engine they used, actually aggregated information from a paper that she had actually written. So basically, the artificial intelligence was plagiarizing, but of course it's the student's responsibility to own up to that. That's them using it inappropriately and probably getting dismissed from college because of plagiarism. It's all about responsible use of the technology, I think is important.

Gary Wunder: Well, I think that's right. And I get that AI does offer that possibility, but the Braille Monitor has had to apologize for running an article which in fact was plagiarized, because the author wouldn't own up to the fact that he had taken it from another magazine. And it was very embarrassing. How college professors are going to deal with this, I really don't know. And will one day the Braille Monitor have to write an apology because something else happened that way? Maybe. But I don't have all the information out there. Maybe one of the functions of AI is to figure out, has this been written someplace else and is it plagiarized?

Anil Lewis: We'll let the artificial intelligence supervise itself.

Melissa Riccobono: Itself. It's also interesting because, of course you can ask, ChatGPT especially, about anything. So there was actually a resolution at our last national convention regarding information about blindness and what the ChatGPT has and what it comes back with. And sometimes it is pretty disturbing what it comes back with. Sometimes it's more what we would, I guess want it to come back with, other times it's very, "Oh, the blind person's going to need a lot of help," and this and that. And of course, you can teach the ChatGPT to some extent. Aaron, I don't know if you've thought about this and what our responsibility is regarding, I don't know, policing or figuring out that information about blindness... And information about a whole bunch of things, but of course we know about blindness and the National Federation of the Blind, so that's our frame of reference. What are your thoughts about making sure that information about blindness is accurate, and how do we go about doing that, do you think?

Aaron Cannon: Yeah, it's a tough question. I thought a lot about, what are the wrong ways to do this? [Chuckles] And I think the wrong way is to try to legislate, at least this early in the process. There are some pushes, especially in the EU to regulate Ai. And as a technologist, that makes me really nervous because we've already seen situations in which that legislation or threat of legislation has had what some would consider negative impacts on us, the blind. For a while, Be My AI was very hesitant to describe people. I believe they've since walked that back because I haven't run into that issue. But this is one area in which actually ChatGPT is a little more reticent than Be My AI to describe people and their features. And so my fear is that we don't understand this technology yet, so it feels really premature and makes me super nervous for us to start talking about legislation.

I agree with Gary that a lot of it really has to be on the shoulders of the user. They are ultimately the responsible party. When a major news organization was caught generating local stories and local recommendations for, I think local food and things like that with AI, and then they were caught because they were just some ridiculous articles, who's responsible for that? Well, they are because they used the tool. This is ultimately just a tool, and I think we need to work on understanding it before we work too hard to limit it. Every technology has had the capacity, or has increased our capacity to do evil things. Every single technology, especially the computer. You can now be across the world and scam people. You couldn't do that as well before computers. So every technology has increased our ability to do bad, but it's also largely increased the ability to do good and to increase opportunities for us, the blind. So I am very optimistic about AI. Yes, there are some very serious concerns, but I think we have to be careful to keep those in the proper perspective.

Anil Lewis:
Yeah, and I'm equally excited about the technology itself too. I know I've been kind of the naysayer, but only to offer balance, because I don't want us to move into this space with a degree of naivety that creates more problems. So going into it with full understanding of the real potential and the pitfalls allows us to really exploit the technology in a real way. So I'm excited about AI, the whole aspect of getting access to information in a very dynamic fashion. But I do think that what we're going to have to do is be extremely responsible in the use of it.

And if we can find ways to make sure that we can populate more positive information about the true capacity of blind people in that big digital space where the large language models are created, then that's great. I love the fact that we have such a huge repository of our information in that digital form through our Braille Monitor publications, on our website, and all the other information that we have there. So hopefully we can continue to inform the technology in a way that represents blind people and the true perspective that we have come to realize.

Gary Wunder: I think that this still falls back on us, and we can't blame artificial intelligence if there is public misunderstanding or public information that is shared that we are not comfortable with. We have a view about what it means to be blind, and we have different views about what it means to live the life you want, and those views have to sink or swim in the pool of the public arena, whether you're talking about artificial intelligence or real intelligence. So I think it just ups the stakes for us as a federation to echo what we know, to deal with things that we know to be outright different, and to exercise a certain amount of tolerance about how people are going to feel about things, and to let the artificial intelligence echo that without putting up so many guardrails that it's our script all the time. If you'd like, I'd like to read you the question that I ask about blind people and the NFB and tell you what ChatGPT said to me.

Anil Lewis: Oh, absolutely.

Melissa Riccobono: Yeah, definitely.

Gary Wunder: Okay. I wrote, "What can you tell me about the National Federation of the Blind, its philosophy, and how what it believes compares with what others believe in the field?" And it wrote me back and it said, "The National Federation of the Blind, open paren NFB, has a distinct philosophy and set of beliefs that guide its actions and initiatives formed in response to conflicts and challenges faced by blind individuals. The NFB has always been driven by a desire for equality and the opportunity for blind people to participate fully in society. One core belief of the NFB is that blind people possess the same basic talents and abilities as others and should have equal opportunities in education, employment, and other aspects of life.

This belief has fueled their efforts to challenge and change perceptions and policies that limit the potential of blind individuals. The NFB has been a strong advocate for the rights of blind people, seeking to address inequities in legislation and the national policy that affects them. The organization's philosophy is rooted in the idea that blindness is not a defining characteristic or a mental psychological handicap. They emphasize the normality of blind people and aim to raise expectations for them, believing that low expectations are significant barriers to achieving dreams. This stance is reflected in their public education efforts, including the publication of Kernel Book series, which contains firsthand accounts." And it goes on. I mean, I find nothing objectionable about that.

Anil Lewis: But it didn't get to the root of your question though, because it was supposed to hopefully give us a comparison to the beliefs from others in the field. I was thinking that there's going to be a counterpoint to that.

Gary Wunder: There is. I skipped a few paragraphs. It says, "Comparatively, other groups and individuals within the blindness community might have different views or priorities. While some might share the NFB's emphasis on independence and equal opportunity, others might focus more on specific aspects, like technology, accessibility, or different approaches to education and rehabilitation for blind individuals." So you could drill down over that.

Anil Lewis: Very nice. Yeah, I like that. I like that.

Aaron Cannon: Very diplomatic.

Melissa Riccobono: Yes, and I do agree with that. The problem to me is when you ask it more in general about... And again, to a certain extent, you're kind of leading it. But if you ask it, "What might be difficult for a blind person?" I don't know, I'm just using that as an example. It might come back with a whole host of things. "Oh, and watching children might be very difficult because you can't visually observe them," those kinds of things. But again, if you're asking AI and taking that as the gospel truth, there's so much on Google already that has poor information, and so we're never going to completely get away from that. I do agree that it is our responsibility as a movement to continue to keep educating and to figure out ways that we can educate in our space. Not to stifle everybody else's views, but to put different ones in there so that our narrative is at least one that is presented, because not everybody does agree with us.

And people might have very different thoughts about blindness based on if they just went blind yesterday or they've been blind their whole lives. So I do agree with that. I'm with you, Anil, in the sense that I think it's pretty cool technology. I'm somewhat of a naysayer to a certain extent. There's just something about it that just... I love what it can do. I've seen what it can do. I've used it. I've enjoyed that. I think for me as a writer and as a musician, I really do worry about the formulaic possibility of everybody using technology like this to start things, or that it is going to get so good that you really can't tell. I don't know, that just doesn't sit right with me. I hope always there will be somebody that will put something different and continue to push for that creativity in the world. And I love, Aaron, how you say that, because you're right. We can do good and we can do evil with everything.

And so I think it's just really important to keep exploring. And I'm not certainly putting it out of my life or saying that I'm never going to use it, or that my children and my husband can never use it. I mean, I'm a huge Beatles fan. I think it's amazing that they were able to create a new Beatles song with this technology. I mean, taking something of John Lennon's that really wasn't usable and using AI to take it apart and make it into usable things, that's amazing to me. And that's responsible use. I guess I just worry that somehow it's going to be very irresponsibly used and we're going to be in some kind of a, "There Will Come Soft Rains  type situation where the house runs itself.

Anil Lewis: No need to worry, because it is going to be misused in certain ways, but that's the way it is with so much. I tell people, it's so inspiring, but it's equally scary. But we can't run from it because it's here now. And I think that what we have to do, and I agree with Gary, it's the same fight we fought, but we are going to have to develop new strategies to continue to remain relevant in this space. Because AI doesn't sleep. It can happen in an instant. So we've got to find creative ways to make sure that we're able to be true to who we are as an organization in this new realm of information sharing.

Gary Wunder: Melissa, when I listen to what you say about AI, I go one step deeper. And maybe you don't go this deep, and so it's not as scary for you. AI makes me question what it is about being human that makes me unique. Because every time we have an advancement that chips away at some of that, it makes me realize that I have to look internally into me rather than externally.

Melissa Riccobono: Yes. No, and that's so true, and that's what I really hope we don't forget along the way, that we don't just become these automatons. That we continue to realize that we are all unique and that it is what makes us unique. And again, the AI can help us, I do believe that, as long as we can continue to put our own humanness. Like you said, Gary, to help it still sound like Gary Wunder is writing this and that it wasn't just spit out from something. And you're going to fact check and you're going to put your little humor in, or whatever it is that helps people know, "Oh, yep. Yep, that's Gary." Or, "Oh, nope, that's Anil that definitely wrote that speech. That's an Anil thing."

Anil Lewis: But it learns. It learns. And each evolution it's going to get closer and closer to being Gary.

Melissa Riccobono: Yeah, that's true.

Anil Lewis: Yeah.

Melissa Riccobono: Well, and then we could have Gary AI. [Anil laughs in the background] I mean, we can actually go a step further. We had Mark AI at convention, and that still creeps me out when Mark AI talks to me. I'm like, "No, stop!" [Laughter]

Anil Lewis: And the other piece about AI that really drives me nuts, it doesn't sleep, again. Most of the stuff you get back, it's so lengthy.

Melissa Riccobono: It's like a lawyer. It's like-

Anil Lewis: We need to teach AI to do more technical writing.

Aaron Cannon: What blows my mind is that we don't actually, at least not completely, we don't understand exactly how this is working. I am certainly no AI expert, but I know a little bit about computer science and I've read a little bit about how large language models work. And from what I understand, originally what we were trying to do is build a neural network, which is a type of computer algorithm essentially, that allows the computer to quote, unquote, "learn."

Well, what we were teaching it to do was sentence completion. Word prediction and sentence prediction. So basically, if I say this, what am I most likely to say next? And get the computer basically building a statistical model to try and predict what we were going to say next, or what a human is likely to say next. And so out of that came this really strange technology that we have now in ChatGPT, which is something you can actually hold a conversation with and seems to actually be thinking, at least as far as our layman's understanding of what it's doing. So it's really interesting that this emergent property came out of what was supposed to be just something that was good at predicting what you were going to say next.

Anil Lewis: Yeah, I love the evolution of technology in that way. And it'll probably surprise our listeners that Aaron is really not a real person. He is actually artificial intelligence. So throughout this whole podcast... [Laughter] I'm kidding. I'm kidding.

Melissa Riccobono: No. No.

Anil Lewis: But it could be that smooth. As it evolves, it really could be that engaging.

Melissa Riccobono: There are things that it does that I'm like, "That actually kind of sounds real." I guess that's been around for a while. You have the, "Hello, I'm Cynthia from the Breast Cancer Society." It sounds a little robotic, but sometimes it's done pretty gosh darn well and you have to take that pause and be like, "Are you really... No, I think you're a robot." But there is that already, I guess.

Gary Wunder: My grandson wrote a little prompt, and the prompt was, "Write a song, make it a folk song, talk about Gary being the grandfather, his granddaughter being Finley and riding on his shoulders as they go throughout the neighborhood." And it wrote this song that was about 90 seconds long, about Finley riding along and being amazed, and she had a great view from where she sat. And while she enjoyed the view, her grandfather wore out his shoes. And it didn't sound corny at all. It was really good. Would it have been the first pass that I'd have wanted to release if I wanted to follow in the footsteps of Jimmy Webb or some other great songwriter? No, but it was quite good. And again, the chords made sense, the lyric made sense, and it did it in 30 seconds. It's amazing.

Anil Lewis: And now it's in the top 10 of the folk music charts. [laughter]

Gary Wunder: Oh, absolutely.

Melissa Riccobono: With Gary Wunder on vocals and guitar.

Gary Wunder: Yep, and trying to collect enough revenue to buy myself a new pair of shoes.

Melissa Riccobono: Because you wore those out.

Gary Wunder: Yep.

Aaron Cannon: This might seem a little bit abstract for some of the listeners, because I suspect there's folks out there who don't use Be My AI, there are folks out there who don't use ChatGPT, and so maybe this feels a little bit remote. But as far as I understand, Microsoft is planning to release in the next few months a new version of, what is it? I think they're calling it Cortana still. I don't know, they have a specific name for it. But essentially, they're going to be releasing an AI built into Windows 11, and possibly Windows 10, and it's going to be built into Word and other Office applications. So you can use the AI to help you draft documents in real time. Just because you haven't encountered AI yet, you probably will.

Anil Lewis: Is that the Copilot?

Aaron Cannon: Yes, thank you. That's what it's called. Thank you. Yes. I should know that because I use the coding version.

Melissa Riccobono: Please say that it won't be a paper clip with eyes, the little Clippy office assistant from way back that everybody hated. Let's hope they grew from that. [Laughter] But I mean, they've had that for a while. That used to pop up. "It looks like you're trying to write a letter. Would you like me to... These are the templates I can give you." So I guess we've kind of had this for a little while. It's much different now, but they've always been trying to-

Anil Lewis: But now it's actually going to write the letter.

Melissa Riccobono: Yes, yes, yes. Correct.

Anil Lewis: And it'll be a 10-page letter, and it'll make up some citations if you need it. [Laughter]

Melissa Riccobono: Yeah. Right. We need the introverted version of AI. My youngest daughter is, I think her philosophy sometimes, and she takes after my dad, why say something in five words when you can say it in three? Why say something in three words when you can say it in one? So that's kind of what we need to teach the AIs so they're not so verbose. We need to have the introverted Calvin Coolidge type AI. [Laughter]

Gary Wunder: Well, was it Will Rogers who said, "He don't say much, and when he does, he don't say much"? [laughter]

Melissa Riccobono: No, Elizabeth says a lot. She can put a lot of punch into that one word. [Laughter]

Gary Wunder: Oh, I was thinking of Calvin Coolidge, but yes.

Melissa Riccobono: Oh, Calvin Coolidge. [Laughter] Well, I think this has been fascinating. I can't wait to see where this goes.

Anil Lewis: What happens when we have this conversation even like a month from now, the way the technology is evolving so quickly? I could imagine definitely by this time next year, it's going to be a whole new world related to AI. And I love the fact that it's everything. It's creating text, it's creating graphics, it's creating folk songs. This is amazing. This is just amazing.

Gary Wunder: What I'm a little afraid of is that people want, for privacy's sake, for AI not to tell us things that people who can see are going to get to see on their own. And I think that just like we have to take the offensive when it comes to ballot marking devices, there are going to be times when we're going to have to be an advocate for AI's ability to tell us things that we have the right to know. And I get that it's a little difficult when I say, "What does she look like?" And the AI tells me, "She's five foot nine and a little overweight," or "She's six foot one and she's a bombshell." Or what it already does that I'm not real happy with is when I take a picture of me and it says, "An elderly man who is losing his hair, who is sitting in a wrinkled shirt." [Melissa laughs] If I hadn't paid $1,000 for the iPhone, I'd have smashed it on the ground. But this may be the reality.

Anil Lewis: Yeah, I agree, and we have the proponents of getting access to that information. It's not about whether we like it or not, you can't make life decisions with limited information. And I think the analogy of, is it information that someone else who is able to see is able to obtain? And if that answer is yes, then we should have the same access to that information.

Gary Wunder: Yeah. Sometimes I think candid information is really hard to hear, but it's really helpful. I have artificial eyes, and when I got my original eyes removed and my artificial eyes replaced, people gushed with how good the new ones looked. But they didn't stop there, they had to tell me how bad the old ones looked. And if I had known that, I might have been less reluctant to give up the old ones, which were painful, and embrace the new ones. There was a time when I was in sunglasses and got some pictures taken, and they appeared in the Braille Monitor for a long time because people said that was the best shot that they had of me. I may not like that, but what if I look better in glasses than I look with my regular eyes or even with my prosthetic eyes? Do I need to hear that? And do I need to get it from enough sources that I can make a decision about how to enhance my visual appearance? Or am I going to be so defensive about it that I'm going to say, "Shut that down."

Anil Lewis: But the key is, you cannot make that decision if it doesn't exist.

Gary Wunder: That's right.

Anil Lewis: If you don't have the ability to access the information, then that's moot, at least in this particular instance. Just like you can go to your friend, you can either ask them directly to give you that feedback or not, but if you don't have that friend to ask, you can't get that information. So I think that those analogies, in my opinion, make it very clear as to our position, the position that we should take around artificial intelligence.

Melissa Riccobono: Yeah, absolutely.

Anil Lewis: Well, I got to believe that this is going to stoke some feedback from some of our listeners, so we're looking forward to the social media feedback and the phone calls, et cetera, because this is only the real beginning of this. I can't wait to see how it evolves and what's that next issue that's inspiring and scary all at the same time.

Melissa Riccobono: Yeah, definitely. What's that next issue, but also what are you using it for? We had some really cool examples. We had one person say they were using it to interact with a book with their child, and that they dropped something on the floor. And they were able to use the AI to help them find it so they didn't have to step on it. There's lots of really cool ways to use it, and so we would love to hear those comments, and then we'd love to also hear the not so great or the things that scare you. Let us hear it all. Find us on Facebook and you can comment. You can even comment on the posts that we've already put up, or you can make a post of your own. You can also definitely email us at [email protected].

Anil Lewis: Or you could call us at 410-659-9314, extension 2444.

Melissa Riccobono: I really appreciate you, Aaron, and you Gary, taking the time and letting us know about your uses of AI and just having such a candid conversation with us. I really enjoyed it and I'm sure that our listeners will enjoy it as well.

Anil Lewis: Absolutely. Can't wait to have you back again in a few months to see how all this has changed. But until then, remember, you can live the life you want.

Melissa Riccobono: Blindness is not what holds you back.

Announcer [over music]: We'd love your feedback. Email [email protected] or call 410-659-9314, extension 2444.