Dyscalculia and ChatGPT

A transcript of Episode 305 of UX Podcast. James Royal-Lawson and Per Axbom discuss dyscalculia and low numeracy as well as the AI tool ChatGPT and how it compares to Google search.

This transcript has been machine generated and checked by Marc Prior.

Transcript

Computer voice
UX Podcast episode 305.

[Music]

Per Axbom
Hello, everybody, welcome to UX Podcast coming to you from Stockholm, Sweden. We are your hosts Per Axbom.

James Royal-Lawson
And James Royal-Lawson

Per Axbom
Balancing business, technology, people and society, with listeners all over the world from Georgia to Argentina.

James Royal-Lawson
And we have a link show today.

Per Axbom
Yes.

James Royal-Lawson
And a link show, that is when Per and I, we have checked the entire internet, we’ve read everything that’s been published. And we have pulled out two articles that we think are worth sharing of you and discussing.

Per Axbom
[Laughs] And the two articles are, the first one: designing for people with dyscalculia and low numeracy. And you’ll hear me struggle with the word dyscalculia. Today, but Dis – Calc – Kuliah, I’m not sure how to pronounce it.

James Royal-Lawson
And the article itself is saying Dis – Cal – Coo – Lia, Yeah, I know. But I know that’s not how we pronounce it.

Per Axbom
It doesn’t feel right…

James Royal-Lawson
We just have to trust them Per. And that same article is by Laura Parker, Jane McFayden and Rachel Malic. And the second one we’ve got for you is How ChatGPT is blowing Google out of the water: a UX breakdown. And that one is by Megan Ng.

Per Axbom
And it’s very topical. It’s fun to get something that is so acutely something that’s on everybody’s lips these days, and actually having a UX perspective on it.

[Music]

Per Axbom
So a couple of years ago, I actually don’t know how many years…

James Royal-Lawson
Four.

James Royal-Lawson
I’m guessing four, Per.

Per Axbom
You know, you probably, you’re aware. I attended an accessibility event here in Stockholm. And on stage was a person who worked with people with dyscalculia, Dis – Cal – Culia. And it was sort of mind blowing to me, because this was something I hadn’t heard about before. And it addressed something that I was not aware of, despite having worked a lot for a long time with online accessibility. And just a few days ago, you James came across this article, that — with a poster describing things you can do to help people with dyscalculia.

James Royal-Lawson
The reason why I think it’s it’s probably four, actually nearly five years ago you went to that talk is because I noticed in our planning board that we added dyscalculia as a potential topic show, back in March 2018. And we have actually mentioned it for those for those of you that are dedicated listeners and have listened to every single episode, we’ve talked about it briefly back in episode 196, which was Accessibility for designers. And then also again in Episode 253: The state of accessibility with Derek Featherstone, but we’ve only mentioned it briefly. And I was, I remember when you first told me about that talk. And I didn’t, I hadn’t heard of it back then. And didn’t realise it was a thing.

Per Axbom
Yeah, exactly. And so this is part of a poster series that people who work with accessibility probably are aware of that has come out of gov.uk over the years, and so this is like a compliment to that poster series. And it’s just lovely. And I think we should probably get on with describing what even is dyscalculia.

James Royal-Lawson
First though, we want to say, this blog post is on [the] UK Government blog, Designing government blog, which they have the motto for the blog: “And we believe working in the open makes things better. This blog is for people designing all aspects of public services from local to central government to share the projects, ideas and concepts or just to think out loud. And the three individuals who are behind this post are Jane McFayden, Rachel Malic and Laura Parker. All three are on Twitter. But one of them, Laura, is also on Mastodon.

Per Axbom
And that’s how you found the article.

James Royal-Lawson
It is, actually.

Per Axbom
So dyscalculia is a learning disorder that affects a person’s ability to do math. I guess that’s the simple explanation. But it also, that makes it hard for someone to read, understand, and work with numbers. So you’re working with numbers, probably more than you think all day long. You’re handling money, when you’re counting change, you’re actually telling and managing time estimating how long things will take when you’re doing stuff, and percentages, which I know a lot of people struggle with. And just remembering number facts about anything, and I guess years, even years are numbers. So just understanding when something happened in history is, of course, a struggle as well.

James Royal-Lawson
Public transport, moving around, there’s numbers absolutely everywhere.

Per Axbom
Just giving directions to someone, you’d say “so you walk 500 metres in that direction”, and people may not even understand what that means. So how does that affect life? Yes, people miss trains, they withdraw the wrong amount of money from cash machines, they get locked out of their accounts, because of course, passwords are a huge problem as well. And perhaps they sign contracts they don’t understand.

James Royal-Lawson
Yeah, that was actually one part of the article, which really does stand out for me was talking about financial services, falling into debt. And basically, how can people understand the consequences of what they’re getting into, if they don’t understand their options?

Per Axbom
Exactly. And you don’t even perhaps understand how much debt you are in.

James Royal-Lawson
Yeah,

Per Axbom
So this affects different or it’s actually influenced by different processing centres in the brain. And I was doing research and I found this article that also explained that it is affected by visual processing, it’s affected by your short term memory, it’s affected by your understanding of language, it’s affected by long term memory, your understanding of quantities, quantities, and amounts, and even your ability to calculate. So there are different processing centres in your brain. And each of these can be affected a little nothing, or a lot. So even if someone has dyscalculia, it may not present itself in the same way as another person who has dyscalculia. And that’s I think, is something important to remember with all different types of accessibility issues is that it’s very hard to generalise.

James Royal-Lawson
And I think on top of this though, one aspect I like about the article is that it’s, it has an inclusive design feel to it, it makes a point of pointing out that it’s not just people with dyscalculia, that struggle with numbers, that low numeracy can also be a factor in all those different aspects that you listed, then. All of them, many of them can also result from low numeracy.

Per Axbom
Exactly.

James Royal-Lawson
So that you don’t have to have a diagnosed or undiagnosed condition to struggle with some of these number related things.

Per Axbom
And there are I mean, there is that you could develop it later in life as well, because there would be like acquired dyscalculia, where someone perhaps develops dementia, or has some sort of brain damage. So it doesn’t mean that it necessarily is something that you’re also born with. So that means it affects a lot of people.

James Royal-Lawson
Yeah, the article itself quotes a figure of 1 in 20 in the UK, and makes a comparison with dyslexia. Dyslexia in the UK, according to the article says between 1 in 10 and 1 in 20. But that itself is interesting, because they say a lot of people aren’t diagnosed as having dyscalculia. So the 1 in 20 is just possibly the diagnosed. I mean, that’s an estimate or kind of officially diagnosed. So you probably can say it’s up there with dyslexia, as you know how frequently it appears.

Per Axbom
Definitely. In the other research I did, I found it affects between between 3% and 7%, according to experts worldwide. So that is actually a huge number of people. And that we probably aren’t taking into account the way that we should be. And that’s why this poster is so very useful. And I want to get into talking a bit about the way the poster is designed is that it has two columns, with the title Designing for users with dyscalculia or low numeracy. And the left column says “Do” and lists items that you should be doing. And the right column lists “Do not”, the items you should not be doing.

James Royal-Lawson
Per, I liked and I noticed that it says do and do not.

Per Axbom
Yeah.

James Royal-Lawson
Not, it doesn’t say “don’t”.

Per Axbom
Oh, yeah.

James Royal-Lawson
I think they deliberately written it out as “do not” there. It made me think of the like no step on aeroplane wings and stuff. They kind of made the language clearer.

Per Axbom
Yeah, exactly. So let’s get into some of examples in the do column. So what you should be doing: round numbers to the nearest whole number. And immediately I think of pricing pages. Because when you are shopping online, of course, you see numbers all the time.

James Royal-Lawson
The dot 99 has been a classic of many decades to make you think things are cheaper, 9.99

Per Axbom
And then this one actually in this case, it actually has the counterpart on the right hand side on there “do not”, it says do not use decimals unless it’s money. So there’s seems to actually be sort of, it’s accepted more when it comes to money that you actually can use decimals?

James Royal-Lawson
Well, I mean, there, I think the article doesn’t go into that. But one of the things there, of course, is what kind of price span you’re looking at. I mean, if you’re, if you’re looking at prices of things that cost, you know, 10s of pounds, we’ll use that as its a British article. So if it was it 10s of pounds or hundreds of pounds, you probably don’t need to worry about pence. So like the decimal point, you don’t need to say 523 pounds and 43p.

Per Axbom
Yeah.

James Royal-Lawson
You can probably round that there safely without causing monetary problems. But if you’re a one pound shop, and you’re selling everything that’s around about a pound, then you can’t round it up, maybe you need to be more differential, we’re saying point five.

Per Axbom
And that makes a second point under the do column, or in the do column. Very important that when you’re using numbers, leave space around the numbers, that makes it easier to process.

James Royal-Lawson
Since I mean, does this mean, like whitespace around the whole thing. So you got to, you know, leave it in little island of space, or…

Per Axbom
That’s my interpretation. Yes. But if you’re writing it as part of a sentence, that’s something else, because then you’re using the sentence to actually provide context and allow people to understand it more easily.

James Royal-Lawson
And that was one of the points on the do side.

Per Axbom
Oh, yeah, that’s a good point: use sentences to add context about numbers.

James Royal-Lawson
So the two of those things go, I guess go together or kind of are variations on how to make the numbers more approachable. If that’s the right word to use.

Per Axbom
Less difficult to deal with, I guess.

Per Axbom
Let people include spaces when entering numbers. This is an interesting one, because that obviously pertains to phone numbers to credit card numbers…

James Royal-Lawson
Postcodes…

Per Axbom
…I think that’s dated over the years. Sorry?

James Royal-Lawson
Postcodes as well.

Per Axbom
Yeah.

James Royal-Lawson
Zip codes, depending on where you are? Yeah, I mean, that this type of affordance is something that we’ve mentioned many, many times. And it’s, I mean, in usability testing is where it comes up as a frustration that when you do you enter things like a telephone number, with spaces in, and you get red text, and like error messages thrown at you, because you’ve entered it with spaces, it’s like, you know, it doesn’t make any difference, you can process it afterwards. So here is a wonderful encouragement to build a bit of affordance into some of this data entry, especially with numbers, and then ask yourself whether you really do need to force people to format things a particular way.

Per Axbom
Exactly. And when you say force, I’m actually looking at the last point in the do not column: do not force people to enter a number or do a sum to verify themselves. Because I think that some people are actually trying to make a point when you do like a CAPTCHA, yeah. That, Oh, you shouldn’t use all these complicated images, you should just do a simple math problem like two plus three. The thing to remember is, that is not a simple math problem for everyone. Because it requires a processing of things and language processing of the plus sign and the equal sign that isn’t easy for everyone.

James Royal-Lawson
Yeah. And even though it might it might be mentally, something that they can do. Yeah, being able to interpret it from the page. And in the stress of logging in, perform that task. It’s not as simple as it may appear, when you’re sat there in a sprint planning session and deciding to implement it. There was also another one of the do’s was: fill in the information you already have.

Per Axbom
Yes. Yeah.

James Royal-Lawson
Which is actually an interesting one, what kind of information are we talking about there?

Per Axbom
Oh, that’s interesting. I’m assuming it is because you are already logged in. And you perhaps want people to enter phone numbers, account numbers, stuff like that, that they perhaps have entered previously in another session and you want to use it again.

James Royal-Lawson
Or it could also mean that when you’re doing a form, that you actually mark the forms up in the correct way so that it utilises the browser’s saved information that you’ve entered previously.

Per Axbom
Yes, good point

James Royal-Lawson
…and that’s address line one address not doing your postcode and so on.

Per Axbom
That’s part of WCAG, you actually should do that. Yes. To actually comply with with the WCA guidelines.

James Royal-Lawson
And it’s very useful to make your forms more, or well, easier to complete generally. So conversion wise, if you’re working with that kind of thing, then you will be wanting to do this.

Per Axbom
Exactly. That’s a good point, always remember that people may have third party tools that contain the data because they don’t want to store it with you. But if you do mark up your fields correctly, then that tool will be able to fill in the content for people.

James Royal-Lawson
The final one of the do’s was the user research.

Per Axbom
Yes, that’s always the good one.

James Royal-Lawson
Yeah. With people who actually struggle with numbers. Yeah, always do the user research, which, again, we were talking about this this morning Per weren’t we? About how there’s still certain things that get crossed off the list a little bit too often in the rush to push things out there. And, sadly, I still find that recent user research is one of those things.

Per Axbom
It’s very true. But also for people who do user research. This is like another thing to think about as you’re recruiting participants, and that I don’t think a lot of people have been thinking about necessarily. The way that you and I were surprised about how common this is and how much of a struggle it can be, because so few people are thinking about it.

James Royal-Lawson
So there’s a couple more of the do nots, I think that we maybe should have a little look at too. Overwhelm people with too much content.

Per Axbom
Yeah, that one that one feels like Yeah, well, isn’t that always the case. But again, it’s such a good reminder that if you don’t overwhelm them, then the areas of the brain that they’re using for processing this type of data, that will go much easier. But it’s always applicable.

James Royal-Lawson
I can here yeah, here is, I guess, removing unnecessary numbers. And if this, if you’ve put too many, maybe this variant, or these kind of different different numbers appearing there, a bill that I think I can have a utility bill or something is one the examples they use there, that you can round things or remove certain presentation of numbers on the bill. So it doesn’t feel overwhelming as of when you look at it. I think we’ve all probably felt that when, especially with bills that you you look at something. I remember telephone bills back in the day when they used to used to list the numbers, you’d ring, you’d rung. In fact they still do, some of the ones now, depending on your contract, you’ll show or use an SMS or text messages to these numbers. And looking at that page is actually quite hard work because it’s just lots and lots of numbers.

Per Axbom
Exactly. And what I’m realising again, as we always say that, I mean, all of the things that you are expected to do or not do, those outlined here, that would help me as well, I don’t necessarily have dyscalculia, but it will help me because it will make the information processing easier. So it helps everyone.

James Royal-Lawson
Yet again, when we talked about accessibility and inclusive design, it’s rare you make things worse for an individual by applying some of these things. Granted, some things might not convert as well, because you’re not tricking people in the same way or are you kind of not pushing them in some way.

James Royal-Lawson
[Laughs] I love how you expressed that.

James Royal-Lawson
That’s perhaps a different episode.

Per Axbom
[Laughs] So I mean, obviously the link to this will be in the show notes. And definitely do check it out, print it out, put it up on your wall, and just learn things. I mean, I love being curious and learning things like this. It’s things like this, learning them, makes me realise there is so much still left for me to discover about how I can improve the web.

James Royal-Lawson
Yeah I know, this is I think, is a wonderful thing to share. And wonderful because this is one of those gaps in knowledge, I think a lot of designers probably aren’t on top of this, just like me and you weren’t really as on top of this until a few years ago when we started to realise it existed. And the article itself basically backs that up too. It says that this area is challenging because unlike plain language, there’s little guidance on presenting numbers and data. It is a hole out there. It is a gap in our education understanding around this thing. So read this, take it in and share it

Computer voice
[Music]

James Royal-Lawson
So our second article for today is, How ChatGPT is blowing Google out of the water: a UX breakdown. And this is an article by Megan Ng. It was published on…

Per Axbom
…Medium and the UX Collective yeah.

James Royal-Lawson
UX Collective, thank you Per. Vanished there for a second. And this last month or so. There’s been an awful lot of chatter and talk about ChatGPT. The new version came out in November.

Per Axbom
And not only in tech circles, I mean, everyone has been talking about it. I’ve seen it in like in general newspapers as well.

James Royal-Lawson
Yeah, a lot, a lot of coverage, hype and discussion. And you know, from “it’s going to take over everyone’s jobs” to you know, “it’s the end of mankind”. I mean, there’s also been a lot of various debates and discussions, that some are more accurate than others. But what I liked about this, or what attracted me to this article on this particular topic was that it had a little bit of a different angle to it. Or it picked up on something I myself noticed and felt when I was playing with ChatGPT, it did feel better than a search engine in many ways.

James Royal-Lawson
Yeah. So that’s a cue for me to just briefly describe what it even is.

James Royal-Lawson
Yes.

Per Axbom
Everyone will not know what ChatGPT is, because you may not have tried it actually, even though it reached 1 million users in only five days after its launch. So ChatGPT, it’s like, it’s an artificial intelligence tool. But it has a conversational interface. You just you type text and sentences or questions, or suggestions to this tool. And you ask these questions in natural language. And the system then just responds within seconds with something that is a paragraph, two paragraphs or like even an essay of content, depending on what query you had. And it’s, that’s what sort of is, and it’s the response is so well written and well formulated, and seems like it understood your question perfectly.

And that is, what is making waves, I guess, with so many people, is that they’re not understanding how does this even work. And the way it works, of course, is that this tool has harvested content from the web and from databases for many, many years. It’s a tool that’s been worked on since 2015. And it’s getting better and better every year, of course, and it’s just taking content that already exists, and learning how to present that to a human being as a response in a way that makes sense.

James Royal-Lawson
And blend it of course. I mean, that’s the kind of key aspect of this. It’s sucked in content across many languages, many domains, and spotted patterns in it, and patterns of whatever are in our questions and then pulls out, I guess, new answers or pulls out answers, based on that vast quantity of information it has at it’s disposable.

Per Axbom
And what many people may not be aware of is, if you ask the same question several times, you will actually get a differently worded response each time because it is actually made up on the fly.

James Royal-Lawson
Well, I think you’ve got to remember there Per, it’s updating itself all the time.

Per Axbom
Exactly.

James Royal-Lawson
So there’s a feedback loop there. So every time you ask the question, I mean, you know, there’ll be maybe other people asking similar questions. So things will constantly be changing as things people are constantly using the tool.

Per Axbom
You’re teaching the tool as you’re using it, that’s also important to remember.

James Royal-Lawson
Yep. So what Megan starts off by saying in the article is just that people are expressing out there in forums and social media and so on, that they do prefer ChatGPT to Google search. And she uses Google search, specifically in the article. And she goes on to list three ways. In three explanations, I guess, of why that might be the case, with a UX hat on, I guess. And the first one she lists there is that it’s a frictionless experience. That, I think you get, because you get a tailored, written formatted answer to your query that differs to the search result you usually get on Google.

So Google, by and large is presenting little clips, excerpts from webpages to you. And you do have to do a bit of information foraging on that result page to find it, to evaluate and to find what is the thing that matches your question, you know? They’ve given you a menu of things that might be an answer to your query. Sometimes Google nowadays do actually put fact boxes and so on, they do actually, you know, if you put in, I’ll go back to numbers, if you put in 10 times 10, in Google, it will tell you the answer at the top. So there are certain things where it’s providing direct answers now.

Per Axbom
But most of it is like pogo sticking. Like you open and go back, you open and go back to to make sure that you’ve read and compared different sources.

James Royal-Lawson
Yeah, I think another thing to remember. And Megan actually points this out, that it’s a little bit of an unfair comparison this in some ways because we’re comparing Google, which is a product that sells lots of advertising, to one, which is currently, I guess, a pure tool. I mean, this is just doing natural language processing and giving you answers, that it’s not, it’s not forcing loads of adverts onto you and other things. So Megan herself says Google’s business model is focused on generating revenue from ads, which clashes with the user’s goal of finding a relevant answer to their query. As a result, Google often shows paired links instead of providing a summary of the answer. Now, we probably would need a whole separate show to get deeply into in that paragraph. About does a particular organisation’s business model, remove/distance itself from the user experience, which is kind of what she implies there. That Google can’t offer the ultimate user experience because they’re needing to sell ads.

Per Axbom
Which is what I’m realising with this tool as well. I mean, this tool, I mean, we can expect it to go the same way as Google, because it will require a business model in the same way, when Google was non — didn’t have advertisements in the beginning either.

James Royal-Lawson
No, but will it though? Is it forced to go towards ads? Because what I mean, what does ads mean, in this context? It means that effectively, it pollutes, it’s existence gets polluted by that.

Per Axbom
Same as Google.

James Royal-Lawson
Yeah. Okay, maybe it would work. But…

Per Axbom
…you can even you can even make it. So you don’t notice the advertising necessarily? [Laughs]

James Royal-Lawson
I see. Now, that actually gets into one of the, you’re jumping straight down into number one, which is transparency. This is. So the first part, we’re trying to be a little bit structured there, the first bit, she talks about the ways in which why, explanations for why people might prefer Google and prefer the chat interface to Google. And then later on, she goes on to ways in which maybe you could improve ChatGPT to be even better. And one of those is transparency. And, you know, at the moment, ChatGPT doesn’t always, it doesn’t as far as I’ve seen, when I’m testing it, say what its sources are, it doesn’t always — sometimes it gives you explanations of why it’s, or it gives disclaimers, or explains things. But it doesn’t always give its full reasoning. It doesn’t, I’ve not noticed it give citations, generally.

Per Axbom
But it can give citations and it can even quote stuff that isn’t true. It can cite papers that don’t exist. So which is interesting as well. [Laughs]

James Royal-Lawson
Yes. Exactly. So then when you get into the kind of business model and advertising, if they were baking in, if they baked in impurities into the answers in the form of ads? That you know, I noticed one, I asked it for, to write some some JavaScript yesterday for me.

Per Axbom
Yeah.

James Royal-Lawson
And in the example code it replies with, because it does actually reply with fully functional programming code when you ask these questions. It used car brands, as an example, because it was something that involved like four different cases in this in this JavaScript structure. So it used four car brands.

Per Axbom
So it may even be testing advertising right now? [Laughs]

James Royal-Lawson
Maybe, I mean, I noticed there were four different manufacturers of cars. It wasn’t four different models of the same manufacturer. But you can see how there’ll be very, very subtle ways of getting advertising into this, which wouldn’t be legal.

Per Axbom
No, it wouldn’t unless…

James Royal-Lawson
…in many countries, you are actually supposed to make it known. I mean, like on Instagram you see it says “Paid partnership with” because in certain territories, you have to disclose when you’re getting paid to advertise something.

Per Axbom
That’s why influencers are getting in so much trouble, because they’re so bad at disclosing…

James Royal-Lawson
…disclosing. Yeah. And the AI, would the AI be subject to that? Because it’s an AI? It’s not a person.

Per Axbom
Ooh.

James Royal-Lawson
Anyhow, moving on. So, going back to the ways in which it might be better than Google, you’ve already talked about it, the natural dialogue.

Per Axbom
Yeah.

James Royal-Lawson
That it feels like you’re talking to a human being, it feels like a real conversation. So that is just a more human feeling of doing this. I’m asking a question. I’m getting an answer back. And you can even then come with follow up questions. And it maintains the conversation. It knows the context where you were, where you’ve got to, what you’ve said previously in this conversation, so there’s so much that feels quite natural.

Per Axbom
Exactly. And that’s one way of correcting things that it may interpret as — oh, I don’t have a good example — so you search for something related to “baker”, but it interprets as the profession but you were interested in a person named Baker, and then you actually correct it by just replying to it. No, I meant a person. And so when it comes to Google, it would actually have to know more and make assumptions about what you were searching for.

James Royal-Lawson
Yeah, it doesn’t normally directly connect the next search with the previous one. Whereas with this you would you would say, Oh, well, that’s a good answer, but no, but what about this? and you can add something else to it. And one flip side of this, the natural dialogue side is accuracy, this is one of Megan’s room for improvement things is the way in which ChatGPT replies is very confident. Like she says, an almost authoritarian tone. So it can give the impression of confidence and trustworthiness. But you might not question something, because it just sounds so goddamn right. Exactly. And without the citations, the sources, the reasoning, you might not for a moment question it. That in itself is risky, tying again into subliminal advertising or hidden advertising. You know, there’s a lot of ways this could could go out with, you know…

Per Axbom
Oh, yeah. Because we joke about I mean, anything you believe you can find a page on the internet that actually describes what you believe as a fact. In this case, you actually, you trust the answer, because it’s supposed to be intelligent, or supposedly intelligent. So it’s actually it’s sort of a confirmation from something that isn’t even human.

James Royal-Lawson
Yeah. And the third point, she lists as to why it might be better is its ability to handle various types of inputs.

Per Axbom
This is the bit I really like…

James Royal-Lawson
…yeah, there’s an inbuilt affordance here in the tool that it’s, you don’t have to spell things correctly. You don’t have to have great grammar. You can write stuff badly. You can miss words out, you can spell stuff badly, spell stuff incorrectly, and jumble words up. And it gets it. I haven’t tried mixing languages yet. But I know it’s multilingual. I tested that too. And you can write in various different languages, and it understands and replies back.

Per Axbom
And I really liked, based on what we actually talking about before with accessibility, that this is actually beneficial to accessibility, because it means that people don’t need to necessarily know the language, you don’t need to know grammar. And I actually tested as we were talking, to ask the AI or ChatGPT about, can you explain what one in five means? And it has a long paragraph where it actually explains what one in five means, which means that it can actually help people make sense of content.

James Royal-Lawson
Oh, right. Yeah. So if you if you haven’t understood something that you’ve read or heard, yeah, you could actually ask it and it will give you a — it’s like, in that sense, it is more like an encyclopaedia, a personal encyclopaedia that will, will talk back, that will describe to you and answer and try to explain it more than just a Wikipedia page or a search results page.

Per Axbom
Yeah.

James Royal-Lawson
It’s, I think that aspect is really I think there is a lot of accessibility benefits to some of this, the way it can find and present answers. But we’ve already highlighted some of the dangerous sides of this, too. But by and large, I think this is a wonderful thing. And I can really understand why people are liking it more than Google, because it does feel like I’m getting served more directly than Google. That’s appealing.

Per Axbom
It’s appealing and it’s very dangerous. But here are some things you need to understand. So it’s like, now you need to learn. What can I trust? What can I not trust? When is it useful? When is it not useful? So there’s a learning curve, of course, as always, with new tools.

James Royal-Lawson
Yeah. And you can’t use it for everything. I mean, Google still blows ChatGTP out of the water, when it comes to certain content, discoveries and new content. When it comes to like maybe searching products, especially when localities are involved, you know, like, opening hours — find me an Indian restaurant in Stockholm. I mean, maybe it can answer, I haven’t tried that. I haven’t tried restaurant recommendations. Maybe it is good at it. But of course, another thing that Google is going to be very good at is telling you where or giving you the location of a specific website, or destination. Like, you know, when you know where you want to find an actual website you want to go to when you’re trying to find the website, then Google is going to hands down be a good tool for that.

Per Axbom
Yeah, exactly.

James Royal-Lawson
Did you notice at the very end of the article that she revealed that it was ChatGTP that tweaked the title?

Per Axbom
Oh, no, I didn’t see that. Ah, interesting.

James Royal-Lawson
Yeah, no. So the original title of the post was Why people prefer ChatGTP over Google: a UX perspective. And she asked, she asked the AI “help edit this into a clickbaity title”. Clickbaity title. And ChatGTP replied with How ChatGTP is blowing Google out of the water: a user experience breakdown.

Per Axbom
Yeah… so we were, we were baited by…

James Royal-Lawson
…Baited by an AI. Driven by a human. I mean, it’s puppet mastery, isn’t it? But I mean, I do really like this, this is really, really interesting. And very, very useful. I’ve already in recent times now managed to get stuff out of ChatGTP that I can actually use in my work.

Per Axbom
I have as well, especially code.

James Royal-Lawson
Code is the thing. Yeah, I really found it’s incredibly, it’s mind blowingly good that you can actually just write a natural language sentence describing the code that you would like, and it gives you it. So, the whole thing about should designers code? Maybe not. Maybe you can actually just know how to ask an AI a question about code. That’s maybe the school you need. Recommended listening. We started with the recommended listening Per.

Per Axbom
We did, because you had the different times we had actually spoken about dyscalculia as well.

James Royal-Lawson
Yeah. Dyscalculia I’m gonna trust that pronunication. Anyway, so yeah, Episode 196: Accessibility for Designers. And then also Episode 253: The State of Accessibility with Derek Featherstone. They’re both really quite related to the first article we talked about.

Per Axbom
Remember to keep moving,

James Royal-Lawson
See you on the other side.

[Music]

Per Axbom
So James, I once knew a mathematician who hated negative numbers.

James Royal-Lawson
The mathematician hated negative numbers?

Per Axbom
Yeah, she would stop at nothing to avoid them.

 

This is a transcript of a conversation between James Royal-Lawson and Per Axbom recorded in December 2022 and published as episode 305 of UX Podcast.