Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:00):
Hi, this is Stephen. Today we have a special bonus
episode which is a crossover episode with the Business English podcast.
Rob and I talk about artificial intelligence, its effects on
the world, its possible future effects, how people are using
it to learn English, and other topics. This is the
(00:24):
first half of the episode and if you want to
listen to the second half then you can find it
on the podcast feed of the Business English podcast, where
it has also been released today. So you can listen
to both parts of the episode, but you have to
find the other half in the podcast feed of the
(00:45):
Business English Podcast. There's also a link in the show notes.
So seven is free, but if you would like to
have access to all of our transcripts, including this one,
vocabulary lists, worksheets, and world news quizzes, you can become
a supporter at senseven dot org. Okay, here's my conversation
with Rob. This is Stephen from SEN seven, the Simple
(01:11):
English News Daily.
Speaker 2 (01:12):
Podcast, and this is Rob from the Business English Podcast.
Speaker 1 (01:17):
Wow, it's great to be here with you.
Speaker 2 (01:20):
Rob.
Speaker 1 (01:20):
Today, we're doing something a little bit different. We're doing
a crossover episode. We are having a part of the
episode dropped on the SEN seven feed and part of
the episode dropped on your feed, and today we are
talking about AI, which is something which really I think
you know a lot more about than I do. But well,
(01:43):
here we go.
Speaker 2 (01:44):
Yeah, I mean, full disclaimer, full disclosure up front here.
I'm not in the AI space professionally, although I am
a consumer of its services, so to speak, and I
also use a little bit of the commercial services as well.
I sort of integrate it into my business so it
sort of assists me there. So yeah, happy to riff
(02:06):
on a few points with you today, Stephen Brilliant.
Speaker 1 (02:09):
Yeah, I also use a little bit of AI, not
in the process of making the podcast, but sometimes I've
started dabbling in a few things which we can talk
about as we go along. But you mentioned maybe talking
a little bit about the history of AI, starting from
Alan Turing.
Speaker 2 (02:30):
Yeah, so I wanted to give a little bit of
a backstory, and obviously the roots run very deep here
and I wanted to highlight one specific element of the
history of artificial intelligence, a man named Alan Turing because
he was working here and was born here in the
(02:53):
UK an Englishman. Actually, he was quite a famous mathematician
of the day, probably one of the most famous ones.
Have you ever heard of him, Stephen, I.
Speaker 1 (03:05):
Have, but embarrassingly I don't know much about him other
than he was a code breaker and quite recently he
was added to our fifty pound notes, which are quite
rare to see because they're so valuable. But that's an
impressive thing if he's been able to get on there.
But yeah, I'm sure you can tell me a lot
(03:25):
more about him.
Speaker 2 (03:27):
Well, it doesn't get much better than that, does it.
Your face is now on the most valuable piece of
currency in your country. Yes, that's pretty good. Yeah. So
a brilliant mathematician, Alan Turing, and he is widely considered
one of the pioneers of what we would now call
artificial intelligence, building one of the first thinking machines. During
(03:52):
the Second World War, as you alluded to, he was
part of some secret code breakers working at a place
called Bletchley Bletchley Park. Fantastic film actually called The Imitation
Game with the Benedict cumber Patch, fantastically representing the man
(04:13):
and the legend Alan Turing, but potentially most notably he
is he came up with this concept like the Turing test.
Have you ever heard of it? Yeah?
Speaker 1 (04:28):
I have heard of the Turing test. Maybe I can
tell you what I think it is. Yeah, yeah, please,
you can please do it.
Speaker 2 (04:34):
Let's do it.
Speaker 1 (04:35):
Yeah, you can correct me. I'm sure the Turing test test?
Is it the Churing testing? Yeah, exactly. Let's see if
Stephen can pass the Cheering test test. So I think
the Churing test is something like whether a computer is
able to pretend to be a human or pass tests
as if it's a human, so that it can essentially, yeah,
(04:57):
pretend to be a human. Is it something like that?
Speaker 2 (05:00):
It is that exactly? The machine or the computer passes
the Churing test if the human that it's interacting with
cannot tell that it's interacting with a machine. And this
should be sort of throwing up red flags now in
(05:21):
people's minds, as I believe that we're edging ever so
close to this moment, certainly through the advent of chat bots,
for example, interacting with website chatbots or Instagram chatbots. Any
experience with.
Speaker 1 (05:42):
This, Yeah, yeah, definitely. I mean I mentioned before that
I've started to use a little bit of AI, and
actually the main thing I've done is something I never
would have imagined that I would have done, which is
actually having a conversation with a chatbot chat GBT basically
about just ideas, just trying to get general ideas for things,
(06:07):
and it's really impressive. It's just so much faster than
looking up looking things up on the internet, you know, like,
for example, if I need some new software and I
want to work out which new software I need, rather
than going through loads of Google searches, which I might do,
chat GBT is much faster at giving me five suggestions
(06:31):
that might be useful or something like that. So I
have been speaking to AI, which is if you told
me that I would have been speaking to AI in
a conversational way five years ago, I wouldn't have believed you.
Speaker 2 (06:41):
Yeah, rather interesting pivots actually that I think that the
space is experiencing in general. There are a lot of
thought leaders now that are looking at this but basically saying,
you know, Google is in serious trouble because people are
not really Googling much anymore, and they're going straight onto
the AI version to just get the direct answers with
(07:02):
the references afterwards. So yeah, at Google producing some great stuff,
but will come onto that in just a minute.
Speaker 1 (07:09):
Amazing if you don't mind me asking. You know, we
were just talking about the cheering test there do you
think that chat GBT passes the cheering test.
Speaker 2 (07:20):
It's a great question. I feel that this is that
I think there are tiers, There are layers to this.
I think through a basic interaction, I think many people
it would seem to me that many people would struggle
to see the difference between an AI response and a
(07:42):
human response, for example, a chatbot on a website. If
you're interacting with an AI chatbot, it may ask you
how your day's going. It will pick up your IP address,
ask about the weather where you are currently. It will
be able to go into the calendar and at availability
dates and get back to you make adjustments, which isn't
(08:04):
beyond you know, the norms for a normal customer service representative. However,
on the flip side, what is the meaning of life?
You know, potentially some cultural nuances, or talk about some
Shakespeare and what it means to you as a person.
I feel that it may give widely generic answers or
(08:26):
some social, socially pleasing responses. Any experience with that.
Speaker 1 (08:32):
Well, in fact, what I was just thinking of when
you were talking, there was an article I read in
the newspaper the other day. They go, this is me
being very twentieth century rather than the very twenty first century.
I actually still read newspapers. And it was about how
universities are having a big problem now of knowing whether
(08:54):
or not students have submitted AI generated essays, and even
there have started to be some lawyers that are working
for students to representing students who have been accused of
using AI to generate their essays and them saying no, no, no, no,
(09:16):
I haven't and then they go to this lawyer to
help them out to fight back against the university. So crazy,
crazy case. But in that would suggest that if university
professors who are marking these essays are not able to
tell whether or not this has been made by AI,
or actually think that it has been made by AI
(09:37):
or something, then yeah, maybe it does pass the cheering
test or well, especially if it was actually written by
a person and then they think it's written by AO. Yeah,
then that's almost like a person not passing the cheering test.
Speaker 2 (09:51):
It gets messy very quickly. Yeah, it does get messy. Yeah,
we can touch. I want to blow this wide open
a bit later on with using AI to learn English,
because I've got quite a bit to say on that
as I'm sure you have as well about this idea
of using AI to help you communicate across cultures, across languages,
(10:14):
and to improve your English. But what we'll touch on
that in a moment. Well, we've sort of covered the
allenturing elements, so allent turing tests. Yes, we're potentially there
in some areas but not in other areas, but you
can be sure that it's coming. What about currently, So
moving into the resent from the historical context of AI,
(10:39):
it'd be nice to just riff about the current usage
any platforms and things like that.
Speaker 1 (10:47):
Okay, Well, I've just told you a little bit about
me now talking to AI occasionally to get some ideas.
You mentioned before that you probably use it a lot
more extensively than I do. So what do you do
with AI?
Speaker 2 (11:02):
So as a consumer just using sort of the chat
bots interface, which would be you know, chat GPT or
Claude or grok or Perplexity, Gemini. You know, there's a
whole host of sort of front runners out there. I
think Facebook or Meta sorry naughty Meta are now doing
(11:22):
something with Lama I think.
Speaker 1 (11:25):
Sory to cut you off, but they're all basically the
same thing, right, all of those names you've just mentioned.
They're all just chatbots, right.
Speaker 2 (11:34):
They on the front the interface for the individual. Yes,
you can just chat to it, but they do have
other sort of other systems that you can tap into
APIs which you can tap into to leverage the large
large language models. Basically, So that's the list of sort
(11:58):
of the most popular ula consumer facing AI chatbots. Obviously,
there are many more applications that the corporate levels that
businesses can use to integrate into their systems, but purely
focusing on the chatbots, these seem to be the front
runners at the moments. Again, disclaimer, I'm not in the space,
(12:21):
so please feel free to tear me apart in the
comments for any slip ups. But the idea is is that, yes,
the average individual can use a chatbot to surface any
information or use it as an echo chamber. For me,
I'm using GROC. I quite enjoy using GROC. However, I
(12:43):
know that claud by Anthropic is also a very good
large language model LM to help you generate ideas. It's
very good for sort of written content, copywriting. It seems
to be well used by copyright writers and people in
the creative space.
Speaker 1 (13:02):
Yeah, and then there was a few months ago, deep
Seek came out, and I think the surprising thing about
that was that people I don't know, people who know
more about AI than I do, thought that China was
a long way behind the United States on the creation
of AI. But then they brought out deep Seek, which
was very very similar, like a similar standard to what
(13:26):
chat Gibt is and Claude and whatever.
Speaker 2 (13:29):
Right, Yeah, that's right. Yeah, I don't know much about
deep Seek. I've never used it, but yeah, very very
much aware of the fact that the countries that were
developing large language models sort of took a sharp intake
of breath. It was like, oh, what's this then, And
(13:49):
obviously deep Seek doing some great work, doing some great work.
Speaker 1 (13:53):
But I mean, I can talk about controversies on my
news based podcast. Controversy is all right, okay, the deep
Seek model, if you ask it certain questions that go
against the Chinese Communist Party's vision of the world, it
refuses to answer them. Like I've seen articles, because again
(14:16):
I haven't used it myself. I've seen the articles where
they say, can you say an insult of you know,
Mexican President Claudia Shinbaum and it will do that. And
then if it says, can you also insult President Chooesing Ping?
And it just refuses to. And if you say, you know,
can you tell me some nuance about the situation of
(14:39):
Taiwan as to whether it's a country or not or
something like that, it will just say Taiwan is an
inalienable part of China or something like that, like really
basic without going into more more nuance of the situation,
or can you say what happened in Tiana Mill Square,
it will just refuse. No, I don't want to talk
(15:01):
about that. Let's talk about something like that, which which
kind of gives rise to this is the most obvious
way of AI being manipulated. I'm sure it's manipulated in
other ways by different people, but as far as having
the power to choose what people can see and what
they can't see, I think the deep Seek model showed
(15:22):
the most obvious case of their biases being plugged into
the system.
Speaker 2 (15:28):
Absolutely. Absolutely, Like you say, it's sort of echoes, well,
if it's made by human it's going to have errors
in it. So I think that's one of the challenges.
I don't know what model it was, who it was,
if it was open AI, if it was Google, but
not too long ago, in sort of the last eighteen
(15:51):
months or so, I remember an article being written about
the generative sort of image generation qualities of one of
the of one of those big players, when it was
asked to produce pictures of the British monarchy, and this
(16:13):
was done in such a manner that it could not
have been possible. You know, that the British Monarchy were
portrayed in a way that they just never would have
been and so there were questions raised about how factually accurate.
I think I just want to take this one step further.
I think Elon Musk, and I'm quoting from Elon or paraphrasing,
(16:37):
should I say where they were said about miss You
know what is worse? Is it to misgender someone? Or
thermo nuclear global war? And the answer was misgendering someone.
And it's like, I don't know about this. It's a
bit challenging.
Speaker 1 (16:51):
Actually, as seeing as you've mentioned Elon Musk, there seems
to be able to slip his way into every single
conversation in the world, doesn't he. But actually, a few
weeks ago, I think I think it was only a
few weeks ago. There was a short amount of time
where Grok again, I didn't, I didn't use this myself.
I saw saw this happen. I read about the Yeah,
you know what I'm saying, really yeah, where Grock was
(17:14):
randomly telling people about white's genocide in South Africa when
they weren't asking about it at all. They were saying like,
can you give me a recipe for a cupcake? And
then you'd be like they'd be like, yeah, this is
a good recipe it and regarding the white genocide going
on in South Africa, and that it was just completely misplaced,
and from what I could read, it seems to have
(17:35):
been that Groc's company what is the company behind Greek?
Is it x or Xai Xai.
Speaker 2 (17:43):
Yeah.
Speaker 1 (17:43):
They said that this was a like something like a
rogue employee doing something small for a few minutes, which
in theory, I mean that rogue employee must have been
either Musk, right, would actually have the power to change
something like that, and would also have the kind of
(18:04):
desire to have this focus on white genocide in South Africa.
I mean, it might not have been, but it seems
quite likely that that would have been something that Elon
Musk would have done.
Speaker 2 (18:14):
I mean, these hallucinations in these models are really like
you said about the you know, the deep Seek or
Grok or any of the others. I'm sure it's going
to carry on dropping clangers. I think that's quite a
good phrase to me. If you don't know that one,
go and look it up to drop a clanger. Great,
you making mistakes. But yes, the idea of you know
(18:36):
the current language large language models in their current form
very exciting space to be I want to address this idea,
and it almost seems to be the battle cry of
the anti AI people, is it's going to take my job.
Any initial thoughts on this, any opening arguments?
Speaker 1 (19:00):
Okay, I hope you've enjoyed the first half of this episode.
If you would like to listen to the second half,
it has been released today on Rob's podcast feed, so
you can search in your podcast app for the Business
English podcast. There's also a link in the show notes.
(19:20):
The next episode of CENT seven will be Monday's News
in seven minutes. Have a great weekend.